Abdul-Razzak (1998), JGR	Cloud droplet formation in the atmosphere occurs via heterogeneous nucleation onto aerosol particles. relationship between aerosol and cloud droplet concentration is complex due to a number of complex processes involved, e.g., aerosol activation process, mixing processes. Only aerosol activation process is considered in this paper. The parameterization relates the number nucleated directly tofundamental aerosol variables.
Abdul-Razzak (2000), JGR	nucleation scavenging of aerosol, one of the primary means for cleansing the atmosphere of particulate matter.
Abdul-Razzak (2004), JGR	Surface-active organic compounds, or surfactants, affect aerosol activation by two mechanisms: lowering surface tension and altering the bulk hygroscopicity of the particles. Large uncertainties about the effect of insoluble organic films on aerosol activation.
Acevedo (2012, BLM)	vaguely defined coupled and decoupled status of stable boundary layer with 1 layer temperature: the term decoupled refers to the state with very weak turbulence and consequent low temperatures near the surface, while the coupled state has stronger turbulence intensities and higher temperatures. why the long-tail stability functions works better on the average than the short-tail functions: within a given region there may be localized patches of turbulence even when the mean regional stability is beyond critical for the existence of turbulence
Ackermann (1998)	MADE; ammonium react with sulfaric acid resulting in irreversible aerosol formation of ammonium sulfate; while excess ammonia react with nitric acid resulting in the reversible formation of ammonium nitrate.
Adams (2002) JGR	GISS-GCM II-prime; aerosol microphysics model = aerosol model We assume that all sulfate exists uniformly as ammonium bisulfate, a global average chemical composition. Accommodation coefficient of sulfuric acid: 0.65 Nine vertical layers between the surface and the model top at 10 mb. most of these layers are in the troposphere with only one or two in the stratosphere stratiform clouds appear in a GCM grid cell once the average relative humidity exceeds 60% The reaction in cloud droplets of sulfur dioxide with hydrogen peroxide is the largest source of atmospheric sulfate A large nucleation source in the upper troposphere, especially in the tropics
Aghedo et al. (2007), ACP	Lighting has a small impcat on surface O3 < 1ppbv
Alexander (2005), JGR	For sulfur chemistry, H2O2 is the dominant oxidant under typical bulk cloud pH condition(pH<5) Because of high CO32-/HCO3- content of seawater, sea-salt aerosols have a high acid buffering capacity or alkalinity. Once alkalinity is consumed the pH of the aerosol drops rapidly, the sulfate formation through O3 oxidation effectively shuts off
Alkezweeny (1984),sci. Total envir.	If we consider the trace metals as tracers for pollutant sources, it appears that the Northwest region has not been significantly impacted by anthropogenic sources.
Allen(1996), JGR	Meteorology fields are interpolated to the appropriate transport time before using. Turbulent mixing: a fraction of material in each model layer within the PBL is mixed uniformly throughout the PBL. GEIA global emissions inventory activity.
Allen (2006), JAWMA	Primary emissions are the dominant source of carbonaceous fine PM during most seasons around Houston. Mean diurnal PM2.5 pattern is shown
Allen (2008), JAWMA	Spatial homogeneity of PM concentration is an indicator of the extent of local source strength because localized sources will tend to introduce spatial gradient in PM concentration. When FRM mass tend to be high and spatially homogeneous in sourtheast TX. synoptic scale wind tend to come from east or northeast.
Al-Saadi, Jassim (2005)	MODIS AOD cannot be retrieved in certain situations: where it is cloudy, where there is strong sun glint from bodies of water, and over snow/ice and bright desert areas.
Ames (2001), AE	comparison between IMPROVE and CDN (CASTNet Deposition Network)
Amundson (2006), ACP	SCAPE2 and ISORROPIA rely on a priori specification of the presence of phases at a certain RH and overall composition, which may be undesirable.
Amundson (2007), JGR	PSC activity coefficient for inorganic multicomponent solutions and the UNIFAC activity coefficient model for water/organic mixtures are combined in thermodynamic model: UHAERO, which included dicarboxylic acids. So interaction between inorganic and organic is considered. This version of UHAERO is an extention of the inorganic only thermodynamic model UHAERO (Amundson et al., 2006)
Andersen (1999), AE	Emission of NH3 is observed when NH3 Conc. is below the compensation point over a spruce forest.
Anderson, 2001, MWR	The emsemble adjustment Kalman filter performs significantly better than the traditional ensemble Kalman filter.    The problem facing the EnKF: the sample sizes of practical ensembles are far too small to give meaningful error covariance statistics, at least  > 100 according to van leeuwen, 1999?    Models have many parameters for which appropriate values are not known precisely. EnKF could produce a smaple of the probability distribution of such parameters given available observations.
Anderson and Neff (2008), ACP	Some concepts regarding mid-latitude PBL may not be valid over Arctic area. Fossil convection?
Andreae (1999), JGR	The oxidation of marine biogenic DMS is the source of both NSS sulfate and MS-(methanesulfonate)
Andreas (1990), Tellus	Sea spray droplets of radius 10 um and less will almost certainly reach thermal and moisture equilibrium before they fall back into the sea.
Andres-Hernandez (2009), ACP	NOx observed by aircraft must be to a large fraction related to lightning episodes rather than being exclusively convectively pumped to upper levels Enhanced free radical production in the level of 450-700mb
Aneja (2001), AE	detrimental effects of excess nitrogen (oxidized and reduced) deposition to nutrient sensitive ecosystems.
Angevine (1996), JGR	No mixed layer is present in the lower atmosphere over the ocean. Vertical mixing occur rarely, due to intermittent process, if at all.
Angevine (2004), JAM	Pollution episodes in New England are associated with moderate, definetely non-stagnant winds. Thus transport contribute to O3 buildup, which contracts with episodes in other part of U.S., which occur in stagnant or recirculating conditions.
Angevine (2006a), JGR	A shallow stable boundary layer over cool waters help efficient pollutant transport by isolating overlaying flow from the surface. If the boundary layer is more stable, the transport is more efficient.
Angevine (2006), JAMC	Lack of penetration of sea breeze is seen from their model simulation.
Ansari (1999), AST	Equation-oriented approach is used in SEQUILIB, SCAPE2, ISORROPIA, in which the equation system is solved by Newton Raphson and bisection methods (Two root-finding algorithm). Assuming that all PM2.5 particles have the same chemical composition(Bulk equilibrium) results in the mixing of acidic with alkaline particles introducing errors in the predictions of nitrate. These errors can be minimized with models employing multiple aprticle groups. where crustal elements are of significance, SCAPE2 is good choice
Ansmann (2003), JGR	The main dust layer is located above the boundary layer up to 3-5km, which is observed by lidar network in European.
Anttila (2008), GRL	It remains uncertain which compounds are involved in nucleation or through which mechanisms freshly-nucleated particles grow to larger sizes.
Appel et al. (2007), AE	CB05 Vs. CB04, systemtatically higher O3 predicted by CB05. Better performance when O3>60ppb but worse when O3 < 60 ppb. CMAQ systemtatically overpredicte lower-range O3 concentrations.
Apsimon (1994), AE	Preliminary results from the ASAM (abatement strategies assessment model) model imply that strategic allocation of resources to cutting back sulphur emission in Europe has substantial benefits relative to a uniform cut-back everywhere.
Arnold and Dey (1986), BAMS	For OSSEs, identical-twin approach should not be used. Fraternal-twin experiments are preferable. Simulated observations must be given proper observational errors. The Ref. Atmos. (or nature run) and the assimilation model should stand in the same relationship to each other as the asmosphere and the model do. A model is always less sophisticated than the real atmosphere, so the assimilation model should therefore be less sophisticated than the RA. This can be accomplished by using a coarser grid and less physics.
Arteta (2006), AE	Online RAMS-RACM is used to investigate pollution plumes in highly variable meteorological condition.
Asman (1998) AE	The actual direction of flux depends on the difference between the surface air concentation (Xsurface) and concentration in the first layer. NH3 Xsurface can have a larger than 0, but for most other component Xsurface will be 0.
Asman (2001), AE	NHx is removed by dry deposition of NH3 at distances less than 1km from the source or by wet deposition of NH4+ at distance larger than 100Km from the source. All NH3 at cloud level dissolves instantaneously into cloud droplets
Athanasopoulou (2008), ACPD	A hybrid scheme for gas-to-particle mass transfer is necessary for accurately simulating semi-volatile aerosol components when coarse sea salt aerosol is included.
Atlas, 1997, JMSJ	Direct (in situ) measurements. Nature run should agree with observations to withiin prespecified limits.
Atlaskin (2012), qj	to avoid numerical instability potentially resulting from thermal decoupling of the surface and atmosphere, operational models tend to overestimate the level of background turbulence in very stable conditions
Avery,2010,JGR submitted	Convective lofting over ITCZ is either a two-stage or a rapid mixing process since undiluted boundary layer air is not sampled in the convective outflow
Ayers (1999), JAC	HCl is more volatile than HBr. If the dehalogenation is simply due to the acide displacement of volatile hydrogen halides by less volatile acid, Cl- loss would be greater than Br-. But acctually Br- loss from the sea-salt is much larger than Cl- due to autocatalytic halogen activation process. Acid catalysis is impotant to dehalogenation process
Ayotte (1996), BLM	Representing entrainment is the most important role of any model.  Some models (e.g., MYJ) show significan errors comparing with LES results.  At high resolution the choice of physical parameterization becomes less important.
Baek (2004), Environ. Pollution	Excess input of NH3, HNO3, and NH4NO3 may lead to soil acidification and eutrophication of ecosystem. ammonium sulfate aerosol has a low vapor pressure, which allows it to condense even at low RH
Baek, and Aneja (2006), Int. J GEI	Particles in the nuclei mode may be transferred into the accumulation mode by coagulation. However, acumulation mode particles do not ordinarily grow into the coarse mode, because concentrations are too low for coagulation to be effective. Relatively small amounts of particulate sulphate, nitrate and chloride species in coarse particles originate from the SO2, HNO3 and HCl reactions with coarse PM instead of equilibrium chemical reactions. often half or more of the PM2.5 mass is composed of secondarily formed species through gas to particle conversion. ammonia has a short lifetime approximately 0.5 to five days or less due to the rapid GTPC of NH3 to NH4+ and relatively high dry deposition velocity to the surfaces near its source
Baik (2001), JAM	Two flow regimes are identified in dry simulations of urban heat island: stationary gravity waves near heating region; the other regime is characterized both by stationary gravity waves near the heating region and by a downwind updraft cell
Bailey (2007), JGR	Model-3 only care about tropospheric chemistry. Two-way nesting of REMSAD: seems only exchange on boundary from both way. REMSAD: designed as a screening model
Banta (2005), BAMS	Emissions often do not vary much from day to day. So meteorological process will control if high-pollution event will occur. Sea-breeze front, PBL height, noctournal low-level jets etc. are important meteorology ingredients that need to be predicted accurately in order to predict pollution event correctly.
Bao, J.-W. (2005), JGR	Cumulus scheme (Grell) is turned off for the 1.67km resolution simulation. The 1st guess for ETA analysis and AVN analysis are the same: from NCEP's operational global prediction model. Sensitivity of model forcasted meteorology to different initial condition(ETA, RUC, AVN). Cold bias is found for the simulations with alll three ICs for the TexAQS2000 episode. Two major advantages of online model over offline model. effect of small scale meteorology impact is lost in offline model. no chemistry-meteorology feedback in offline model.
Bardouki (2003), AE	The statistically significant correlation (r2=0.73; P <0.05) between sodium in excess and NO3- concentrations with a slope of 0.8, indicates that this reaction can account for the majority of the nitrate observed during the experiment
Barker, 2004, MWR	overfitting to observation is not warrented when observation errors are larger than back ground errors.   This version of 3D var use climatological back groud errors.  VAR are categorized as those data assimilation systems that provide an analysis via minimization of a prescribed cost function.
Barna (2000), AE	A diagnostic model extrapolates existing surface and upper air observations to develop a mass consistent flow field within the model domain. observational nudging, a type of FDDA. Observational nudging relaxes the model state towards to observed state by adding, to one or more of the prognostic equations, artificial tendency terms based upon the difference between the two states.
Barrie (2001), Tellus	3 general circulation models (GCMs) that generate their own meteorological fields and 6 chemical transport models (CTMs) that are driven by gridded meterological fields produced from observations. Vertical mixing of surface emissions from the planetary boundary layer into the free troposphere in source regions is a major source of uncertainty in predicting the global distribution of sulfate
Bassett (1983), AE	The 2nd thermodynamic model following Saxena? include mixed salts. Bulk equilbrium, size distribution is not considered Russell (1983) is the first one to couple gas-phase air quality model to an aerosol equilibrium calculation,but only ammonium nitrate is considered in his model When aerosol size-related effects such as visibility reduction are of interest, it will be necessary to predict size distributions.
Bassett and Seinfeld (1984)	The kinetic limitations are most important for large particles or for species with very low vapor pressures. Nitrate will tend to evaporate from the smaller particles and deposit on large particles where surface curvature effects on vapor pressure are minimal.
Bates (2004), JGR	In MBL, sea-salt is highly depleted in chloride from reaction with sulfuric and nitric acid vapors. Dust has higher buffering capacity comparing to sea-salt, thus sulfuric and nitric acids may preferentially associated with them. Supermicro dust and sea salt provide a major sink for nitric and sulfuric acids since their higher settling velocity and shorter atmospheric residence times.
Bates (2005), JGR	Organic aerosols were a dominant component of the total submicrometer aerosol mass in marine boundary layerr off the New England coast during summer of 2002. Reduction of sulfate may eventually lead to dominance of organic aerosol.
Bates (2008), JGR	The supermicrometer aerosol measured over the Gulf of Mexico was highly enriched in nitrate and chloride is depleted.
Bauer (2002), JGR	O3 overprediction at rural area and in free troposphere. Online model but just gas phase. no aerosol phase.
Bauer (2007), ACP	For computational consideration, EQSAM is used in GISS to solve gas/aerosol partitioning. GISS model runs for 6 years, the 1st year simulation serve as spin up. Present and future climate conditions are driven by sea surface temperature and sea ice conditions from another model.
Baumgardner (2002), JGR	Use measured co to estimate BC co is produced almost entirely from combustion precipitation remove BC if hygroscopic compounds such as sulfates or organic acid have formed on the BC surfaces.
Baumagardner (2004), AE	The number size distributions are integrated from 0.03 to 18 um to obtain a total number concentration that simulation the measurement by CN counter, CCN is integrated from 0.06-18, mass integrated from 0.05-2.5 to simulation measurements of PM2.5. 0.06um is the activation diameters at 0.5% supersaturation.
Baumgardner (2006), STE	In many coastal cities, the mass concentrations of particles changed very little between sea and land breazes. recirculation of pollutants that had originated from the land but returned with the sea breaze.
Beekmann (1994), JGR	Potential Vorticity(PV) a tracer of stratospheric air intruded into the troposphere. PV has significant positive correlation with O3 in lower stratosphere(225hpa) and middle troposphere(500hpa)
Belair (1999), JAM	Classification of different PBL schemes,  orders, local vs. non-local.         inclusion of nonlocal term has little impact on the simulated PBL due to small surface heat flux in this specific case.
Bell (2004), AE	Climate change induced increases in biogenic VOC emissions could significantly impact ozone concentrations.
Bell (2008), JGR	From TRMM data, it is observed that during summertime, rainfall over southeast U.S. is significantly higher during the middle of week than on weekends due to intensification of afternoon storms.
Berge (2001), JGR	Above the PBL, the impact of the initial conditions could last for at least more than 3 days. The large impactor factors in the free troposphere strongly complicate any model evaluation by use of measurements in the upper air.
Berkowitz (1996), JGR	O3 correlations with CO and aerosols will be the main feature used to identify excess ozone. CO is used to identify air masses impacted by human activity because CO is long-lived species associated with imcomplete combustion of fossil fuels.
Berkowitz (2000), JGR	24 hr of "spin up" allows time for chemistry to replace the IC with more realistic simulated values. The associated mixing with these large-scale cloud field was reflected in modifications to the vertical velocity and turbulent kinetic energy (and thus eddy diffusivity) fields. Thus grid-scale cloud mixing is included. However subgrid, fair weather cumulus is not considered in this study Ozone is not produced at surface, but is produced somewhere above the surface and subsequently brought to the surface by mixing
Berkowitz (2004), JGR	O3 problem in Houston is unique because of very high value, and rapid onset, while in othere area, the high O3 is associated with more regional scale and slowly developping event. High correlation between HCHO and high, rapid formed O3
Berkowitz (2005), AE	VOC reactivity as given by Atkinson(1994)
Bessagnet (2001), AE	absorption process (transport of semi-volatile species from the gas phase to particles) = mass transfer process
Bessagnet (2004), AE	Sea-salt may be significant at coastal sites when the strongest winds occur, which may cause large negative biases if the model don't consider the wind blown sea-salt emission. At mountainous sites, the model is sometimes deficient due to transport is difficult to model.
Betts (1992), JGR	entrainment may be underestimated in many parameterization models
Betts (1997), MWR	MYJ PBL scheme is used.  The Eta model didn't predict sufficient BL-top entrainment and didn't predict the overshoot cooling at BL top produced by entrainment.  Eta model predicts too stable a nocturnal BL. So greater coupling between the surface and atmosphere in stable conditions may be desirable.   Also transfer coefficient in the unstable BL need to be higher.
Bey (2001), JGR	The tropopause is diagnosed in the model using the standard criterion of a 2 K km-1 lapse rate. Above the tropopause we use a simplified chemical representation.
Binkowski (1995), JGR	ACM handle vertical turbulent trnsport under convective conditions while K theory handle other conditions. Dry deposition is represented in the model as a boundary condition on the finite-difference form of the vertical turbulent transport term. To calculate the aerosol LWC, A fit to the experimental water activities reported by Apann for H2O-to-SO42- molar ratios as a function of NH4--to-SO42- molar ratios is used RPM only contains two modes, so soil dust and sea salt are not included. RPM also assume the aerosol consists entirely of SO42-,NH4-, and H2O, consequently, nitrates are not considered
Binkowski (2003), JGR	The water content of sulfate aerosols depends strongly upon the ionic ratio of ammonium to sulfate. The ammonia deficient regime: the ionic ratio of ammonium to total sulfate ion is less than 2.0
Birmili (2000), GRL	80% of the new particle formation events showed a remarkable increase in SO2 by an average factor >7, which is explained by entrainment of polluted air from aloft
Bottenheim et al. (2002), AE	Why ODEs just occur in Spring? because CHBr3 peak in winter?
Bottenheim and Chan (2006), JGR	O3 depleted air over barrow appears typically 1 day's travel from a putative region where O3 depletion chemistry is active. Air parcels depleted in O3 may remain the way for several days. This is possible in the strongly stratified stable air over the Arctic in the Spring
Bottenheim et al. (2009), ACP	Turbulence may bring down O3-richer air to surface and terminate ODEs. ODEs are often interuptted by such turbulence Increase in atmospheric stability created the conditions for the prolonged ODEs in Spring Frequent short-term variability in the winter patterns started to decrease significantly after polar Sunrise
Boy (2006), ACP	kinetic nucleation of H2SO4 is the primary mechanism for the formation of new particles in this case. Mechanism responsible for formation of new particles in PBL is still not completedly understood.
Boylan; Odman(2002), AE	Ramp-up Days; Adaptive grid model introduced by Dr. Odman! The impact of gas-phase ICs diminishes more quickly when compared to aerosol species because fine aeroosol have very low deposition velocities. All the modeling results were sensitive to the boundary conditions and long-range transport of both gas and aerosol species from the boundary can affect the concentrations of pollutants thousands of kilometers away. The performance for EC is better because EC is inert. The inert species include magnesium, potassium, calcium, EC.
Boylan; Odman (2005),JAWMA	normalized quantities can become large when the observations are small. So whenever the observation is smaller than the threshold value(40ppb), that estimate-observation pair is excluded from the calculations. Some important meteorological parameters can bias the predicted pollutant concentrations and skew the source/receptor relationships. Not modeling large point source plumes at subgrid scale can result in plumes being diluted too fast and the chemistry pushed into a different chemical regime.As a consequence, the spatial relationship between pollutant sources and downwind pollutant conentrations can be affected
Brankov (1998), AE	High o3 level in Northeast tend to occur when air arriving at the monitoring site has prviously traveled over the southeast and midwest.
Brasseur (1998), JGR	In-situ emission; surface emission; MOZART is driven by the dynamics of CCM, which are provided every 3-hr. Within 3-hr period the variables are intepolated linearly as a function of time. Semi-Lagrangian transport is inherently non-conservative, a mass conserving correction is applied to the solution after each time step. gas phase only interacts with sulfate aerosols in this version of MOZART. Microbial production of NO in soils
Braun (2000), MWR	in this case only ECMWF IC BC and Betts-Miller can produce hurrican.  If PBLH is predicted higher than the convective cloud base it is not good? since cumulus transort is supposed to be responsible for the transport above this level.
Bright (2002), WF	Local schemes have been extended to 3rd oder while nonlocal schemes have been limited to first order.   MYJ Eta PBL scheme is a local 1.5 order scheme.         highr-order, local PBL scheme (e.g., MYJ) tend to produce too shallow, too cool and too moist PBLs.     nonlocal PBL schemes are better to simulate deep PBLs over the southwest of US.
Brock (2002), JGR	In summer time, when relatively cloud-free conditions prevail, gas-phase oxidation of SO2 to H2SO4 and subsequent condensation onto particles is significant source for total aerosol mass. So we may need to pay attention of this for TxAQS2000 case.
Brock (2003), JGR	Photochemical oxidation of SO2 is the Key process regulating particle mass growth in all the studied plumes in TexAQS2000 episode. In-cloud oxidation of SO2 is believed to have been negligible because of negligible cumulus clouds on 27, 28 August. The rapid decrease in particle number concentration in the ship channel plume with increasing plume age is qualitatively consistent with expected high rate of coagualation combined with dilution. Compounds other than (NH4)2SO4 such as nitrates or organics, contributed to most of the observed particle growth in plume from petrochemical industries along the Houston Ship channel HNO3/NOy is a better proxy for integrated exposure of the plume to OH
Bronnimann (2000), AE	Background conditions are often addressed by analysing ozone at remote or high altitude sites, which are far from emission sources.
Brown (1996), BLM	Local closure produce insufficiently well-mixed profiles.        Although the failings of local closure scheme in convective conditions are well known, they are commonly used in NWP models.
Brown (2007), JGR	NO3 and N2O5 aloft have larger concentrations and lifetime. NO3 could oxidize VOC (ISOPRENE, Monoterpenes)
Burgers (1998), MWR	Observations must be treated as random variables at the analysis steps, i.e., one should add random perturbations with correct statistics to the observations and generate an ensemble of observations.
Butler (2003), JGR	Most PM2.5 species peaked during the summer months (Sulfate and OC I think); however, nitrate, metals and EC showed some enhancement in the winter due to lower inversion heights. Trace metals contributed less than 1% to the total fine mass on average
Buzcu (2006), JGR	Junquera et al. [2005] didn't include SO2 emissions from fires. The increase in sulfate during the wood smoke episode could be explained by heterogeneous/surface reactioon on wood smoke particles.
Byun (2007), AE	One of the most important findings of TexAQS2000: reported emission of light olefins were not consistant with measured concentration. Semi-conservative species (co and NOy) CMAQ modeled atmosphere lacks reactivity to produce the observed high ozone event.
Caldwell (2005), JAS	Higher entrainment at night and very little entrainment around local noon
Campbell (2002), AE	The ratios of sulfate and calcium to sodium are much higher than sea water, which means some terrestrial origin of gypsum. It is better to model them as externally mixed particles. EQUISOLV only fails when formation of solid NH4NO3 or NH4Cl. Other non-volatile solids don't matter
Capaldo (2000), AE	hybrid mass transfer approach Bulk equilibrium approach ignore the differences in the chemical driving forces of the specific aerosol "bins" In size-resolved equilibrium approach, the chemical composition of discrete and internnally homogenous aeroosl size sections are considered to generate specific drving forces for each section
Carmichael (1991), AE	Once the cloud forms, the number of sulfate aerosols activated and dissolved to form aqueous sulfate is assumed to be the same as the number concentration of cloud droplets.
Carmichael (1998), JGR	For cloud cover index values greater than 0.5, convective transport was assumed by "pumping" the near surface gas concentrations to cloud top elevations by "enhancing" the vertical eddy diffusivity. IC was chosen at the lower end of the concentration range high ozone levels were transported downward by high pressure systems which trail the cold front
Carmichael (2003), JGR	Forecast vs. hindcast Lightning NOx emissions are from cumulus area Upper air is dominated by the boundary conditions, so the prediction for upper air is less structure biomass emission transport in the warm conveyor belt at altitudes above ~2km
Carmichael (2007, 2008?), JCP	Boundary layer structure and wind direction are poorly predicted meteorological variables. Direct sensitivity analysis is source oriented approach. Adjoint sensitivity analysis is receptor-oriented approach. Advection scheme may change from a high order, large stencil scheme inside the domain to a lower order, short stencil scheme near inflow boundaries 4D-var application typically increase the simulation time by factors of 50-80 over that of forward model. Uncertainties associated with aerosol: emissions > Wet removel > chemical formation > vertical transport. BC may not have formation uncertainties but it's emission uncertainty is higher than that of sulfate. So it may not be a good accuracy indicator for models. sulfate should be the mostly accurately predicted aerosol species.
Carmichael (2008), AE	source-receptor relationships were sensitive to uncertainties in sulfur emission, and secondarily to the driving meteorology. The variability of prdicted nitrate due to different aerosol module is as large as variability due to different CTMs. Boundary conditions are large uncertainty in regional model.
Castro (2005), JGR	By RCM we mean a limited area model run for an integration time greater than 2 weeks, so that the sensitivity to IC is lost. Internal nudging did improve the representation of large scale freatures but weakened the variability for small scale features. Surface boundary forcing appeared to be the dominant factor in generating variability for small scale freatures and exerts greater control on the RCM solution.
Chai (2006), JGR	Two assimilation methods are extensively used: Kalman filter method using a sequential approach, and 4D-var with a variational approach. 4D-var is used to adjust the IC of some species. assimilate some observation into the model, but use other independent measurement to validate the model.
Chai (2007), JGR	The advanced data assimilation such as four-dimensional variational data assimilation (4D-var) are more effective than the traditional techniques such as nudging and optimal interpolation(OI). MM5 use AVN during forecasting and NCEP FNL analyzed data during postanalysis. Chemical BC use MOZART GFDL during forecasting and MOZART NCAR during postanalysis. MOZART NCAR assimilated MOPITT CO data In this data assimilation, only initial ozone concentrations were adjusted.
Chan, Chak Chan (2007), AE	The ZSR equation assume all solutes (inorganic and organic) uptake water independently
Chang and Allen (2006), EST	CB04 is modified by adding chlorine chemistry. Including chlorine chemistry may increase peak O3 prediction by around 5 ppb for some locations.
Chang and Allen (2006), AE	HCl is created as soon as molecular chlorine is emitted. Formation of ammonium chloride in Houston occurs only when the atmosphere is amonia rich, typically in early morning hours.
Chang, Ming Chih (2000), AE	Time integrated sampling of labile species often leads to errors (artifacts). Particlulate ammonium nitrate and organic carbon are the two most predominant labile PM species
Chandrasekar (2003),AE	Running a coarse prognostic meteorological model and ingesting its output to a fine resolution diagnostic model simulation to construct meteorological field are good approach in terms of both computational cost and accuracy.
Chapman (2008), ACPD	The non-positive definite advection scheme may cause overprediction of PM2.5, especially for point sources. The positive definite advection scheme in WRF/Chem 3.0 could solved this problem. SO2 emission may be still overestimated.
Charlson (1987), Nature	DMS from the oceans is a significant gaseous sulphur precursor of NSS-sulfate of biological origin.
Chen, D.S. (2004), Environ. Info. Arch.	First guess meteorology field of MM5 were obtained from global tropospheric analysis datasets provided by NCEP and were available every six hours When the stable continental high was dominant, local atmospheri ciuculation would be main factor affecting pollutants dispersion
Chen, F (1997), BLM	Three surface layer schemes perform similarly and the resulted surface exchange coefficients(ch) do not lead to significant model predictions given the same roughness length.    The model is more sensitive to the choice of roughness length for heat.
Chen, F(2009), GRL	NOAH land model overestimated Ch for grass and crops, which may have provided too much water vapor for the U.S. Great Plains
Chen, J (2008)	Forecasting system over pacific Northwest region (95*95*21 grid cells) CMAQ performs better for episodic ozone events (>50ppb), but systematically overpredicts levels < 40ppb. IC of CMAQ is from last hour of previous da's forecast, which reduce the need for model spin-up associated with static IC.
Chen, J.J (2006), JGR	CACM and MPMPO in CMAQ. MPMPO accounts for both absorption and dissolution into aqueous-phase water.
Chen, L.-W. (2002), AE	Summertime fine PM is dominated by sulfate source. Ammonium sulfate accounts for 50% of PM2.5. Fraction of local and regional source varies for different season.
Chen, L.-W. (2003), JAWMA	In mid-Atlantic region, SO42- and carbonaceous material each accounts for 50% of clean air PM2.5 (<10ug/m3). but change to 60% and 20% for more polluted air PM (>30ug/m3)
Chen, S (2009,2010?), AE	In polluted conditions the mechanistic details are less important than in cleaner conditions probably because of the dominance of reactive nitrogen chemistry under polluted conditions
Chessa (2004), WF	Limited-area Ensemble prediction systems nested into a global ensemble.           Users are often not interested in 'the most probable weather scenario' (described by a single deterministic forecast or by the ensemble-mean forecast) but need to access and quantify the risk of occurance of cimatologically rare but destructive events.            The effect on the forecast error of model uncertainties (different parameterization) is as important as the effect of initial uncertainties.    Better parameterization may have more positive effect on prediction than higher resolution.
Chin (2000), JGR	GOCART driven by the assimilated meteorological fields. The results from other GCM in general represent multiyear values averaged over a large area. OH, NO3, H2O2 monthly averaged fields. Anthorpogenic sulfur is emitted as SO2 except a small fraction as sulfate(~5%) scavenging of SO2 is limited by availability of H2O2
Chow (2006), JGR	High PM concentrations occur in winter. High concentration of NH4NO3 is vally-wide homogeneous, which accounts for 30-60% of PM25. The conceptual model could explain this: At night HNO3 form in the upper layer over cities, which will be transported to rural area wich has high NH3 emission. So the mix of NH3 and HNO3 will create NH4NO3. Next morning, the vertical mixing will bring NH4NO3 down to the surface.
Chow (2006), AST	Poorer comparability among different measurements is found when the aerosol is volatile (e.g., dominated by NH4NO3)
Chow (2006), Chemosphere	The diurnal pattern of OC and ED at Angiola reflected the transport of secondary pollutants to the site. In summer of Fresno, there were distinct morning and late afternoon rush hour peaks. In winter of Fresno, OC and EC peaked at night due to residential wood conbustion and the development of a nighttime inversion.
Chow (2008), JAWMA	Standard TEOM underestimate PM2.5 FRM mass by 20-30% at Atlanta, Fresno, and Houston. FRM underestimate PM2.5 mass by 10-40%. Continuous ion measurement by thermal-gas methods underestimate filter-ion concentration by 10-50%
Chu, D. A. (2003), JGR	The derivation of PM concentration from satellite measurements may be possible once we know the detailed aerosol vertical distribution.
Chu, S.-H. (2004), AE	In Atlanta ammonium sulfate were high in summer, OC is flat year-round. In wintertime OC in Fresno were predominately primary in origin, only 30% of OC could be attributed to SOAs on somedays.
Chu, S.-H. (2006), AE	Natural emissions such as wildfires make CA subject to high risk of violation of 24-hr NAAQs. In east US sulfate and OC are the key PM components, which may make east US violatethe annual PM25 NAAQS.
Chuang (1997), Nature	Cloud droplet activation is described by Kohler equation, which assume cloud droplet activation occurs as an equilibrium process. marine stratiform clouds are climatically the most important cloud type
Civerolo (2003), PAG	A common synoptic-scale feature associated with O3 episode over the eastern US is the presense of a high pressure system aloft (500mb), which usually accompanied by subsidence.
Clappier (2000), JAM	dynamic module and chemical module run simutaneously but with different time steps.
Clarke (1997), AE	Dry deposition is not measured directly in CASTNet, but is determined by an Multilayer inferential model. (i.e., fluxes are calculated as the product of measured ambient concentration and a modeled deposition velocity). Routine monitering using direct measurements of dry deposition is not practical for regional networks.
Clegg and Seinfeld (2006), JPC	Thermodynamic model, which has the interaction between the inorganic and organic species?
Clement (1996), AE	Accomadation coef. is important for CCN. Accomadation coef. should not be much less than unity.
Cocker (2001), AST	Wildfire lead to an increase of hygroscopicity of aerosols
Cohen (2004), JGR	wind blown soil was 7% to 9% of PM2.5 total mass in ACE-Asia experiment Black carbon represents of 8%-9% of total fine mass
Colby (2004), WF	two-way nesting: the inner grid provide the data points for the outer grid where they overlap. Sea breezes usually did not penetrate inland more than 20 km in this case. ensemble runs might be a better used of current increasing computing power.
Colette (2005), AE	The frequency of export of ozone from the BL is uniform throughout the year since distinct processes are responsible for upward vertical transport.(warm conveyor belts are dominant in winter and local circulations and convection play a role in summer) The BL is vented toward the free troposphere 90 times a year
Collins (2001), JGR	The assimilation in this paper is equivalent to a 3D source or sink for aerosol although it is not a dominant term in aerosol mass budget. Satellite retrieved AOD is assimilated.
Constantinescu (2007), JGR	Errors in the initial conditions (the background errors) are evolved according to the tangent linear model associated with the forward model.
Cooper (1994), JGR	Lidar observation showed the transport of pollutants under convective PBL is not a continuous and gradient-driven process, but low frequency and spatially discontinuous.     Directly accociated with eddy and plume dynamics.
Cooper (2001), JGR	Midelatitude cyclones tracking from west to east are believed to be responsible for the bulk of the trace gas transport from North America to the western North Atlantic Ocean.
Cooper (2002), JGR	Midlatitude cyclones provide the energy necessary for most of the trace gas transport from North America to the western North Atlantic Ocean, even in summer when these systems are relatively weak. The classic midlatitude cyclone is composed of several airstreams such as the warm conveyor belt (WCB), cold conveyor belt (CCB), and dry air stream (DA).
Cooper (2004), JGR	A wide variety of meteorological mechanisms export trace gases from the Asian ABL to the free troposphere, but in this paper WCBs(warm conveyor mechanism) are the most important mechanism for tranport of Asian pollution to North America Despite the large descending motion of WCB over western North America, the magnitude of the descent was not great and most of the air mass passed over the USA in the middle and upper troposphere. So the tranport of pollution from Asia to North America does not necessarily have a large and immediate impact on surface trace gas mixing ratios.
Dabberdt (2004), BAMS	PBL and cloud remain the largest sources of uncertainties in air quality modeling. Current uncertaintainties are presentaly estimated to be about 30% during the day and 100% at night.
Daescu (2003), AE	Define of stiff system (the eigenvalues of the Jacobian matrix differ by several orders of magnitude). 4D-var techniques could be used to find an optimal initial and boundary conditions and emission rates. The application results show that using data assimilation the true emission rate values are successfully retrieved.
Danalatos (1999), AE	Pariculate nitrate as a long-lived species with a ligetime about a week, would not be expected to show significant diurnal variation. The concentration of this specie would be expected to be affected by long range transport. Different sources are operating for HNO3 and NO3-(aerosol phase) and thus a relationship between them would not be expected. The correlation coefficients between gas HNO3 and particulate NO3- is ver small.
Darby (2005), JAM	Cluster analysis in IDL program role of chemistry versus meteorology when sysnoptic-scale winds dominate, O3 is low in Houston because of transport. However, sometimes VOC emission from petrochemical industries could dominate over the meteorology influence. In such case, O3 is still high even tough constant transport.
Darby (2007), JGR	Land surface scheme and PBL scheme may be part of the reason for higher bias of WRF/Chem. The models predict pollutants better at synoptic scale. But the models are weaker on days when small scale circulations (e.g., sea breeze) are prominent.
Dasgupta (2007), AE	The particle phase slightly favored nitrate over chloride at high RH, while greatly favored nitrate at lower RH over chloride.
Dassios (1999), AE	accommodation coef. rangs from 0.8 to 0.5
Daum (1996), JGR	the plumes over ocean were observed in well defined layers at altitudes between 0.3 and 2 km above the surface. chemical species contained in these plumes were not subject to losses to surface by eighter dry or wet deposition processes.
Daum (2000), JGR	During July 11-13, 1995 Tennesse episode, stagnation weather was dominated by a upper level high pressure system. Thie system cause weak pressure gradient, so daytime wind is light. Mean flow is less than 2m/s in boundary layer.
Daum (2004), JGR	High petrochemical industry and mobile source around Houston. 40% of the world's production capacity of low molecular weight alkenes is in the Houston-Galveston area, which will result in high concentration of HCHO, the product of the degradation of low molecular weight alkenes. Unique characteristic of Houston O3: Very high O3 event occur even when background O3 is modest.
Davis and speckman (1999), AE	Houston area has a number of complicating factors which make the O3 prediction difficult. For example the inland penetration of the sea breeze circulation. The lack of knowledge of the vertical concentration of ozone and its downward mixing during the establishment of the daytime convective boundary layer also presents a problem.
Dawson (2007), ACPD	Primary organic aerosol is treated non-volatile. SOA is treated semi-volatile. Biogenic emissions were based on the base case meteorology and did not change with perturbations in meteorology
de Arellano (2004), JGR	CO2 concentration in the boundary layer is reduced much more effectively by the ventilation with entrained air than by CO2 uptake by the vegetation. entrainment ratio for virtual potential temperature is 0.6; for moisture it is -0.25; for carbon dioxide is 2.9
DeBell (2004), JGR	Entrainment of anthropogenic pollutants together with dust plume were observed in northeastern US surface stations. So Asian export of pollutants can impact the entire US.
Debry (2006), JAS	The numerical simulation of the GDE is still a challenging point in current CTM. A key feature is related to the wide range of the characteristic timescales for condensation/evaporation. An alternative approach is based on assumptions related to the equilibrium of condensation/evaporation. For a long time, condensation/evaporation has not been dynamically solved and a bulk equilibrium was supposed. size-resolved equilibrium(apply a thermodynamic model sequentially for each bin) or bulk equilibrium with size-resolved redistribution have also been proposed.
Debry (2006), Atmos. Chem. Phys. Discuss	For cells with a liquid water content exceeding a critical value, the grid cell is assumed to contain a cloud and the aqueous phase module is called instead of GDE model. Closed mode, forward mode, global equilibrium SIREAM has dynamic, equilibrium, and a hybrid mass transfer approach
Debry and Sportisse (2007), ANM	coagulation is a slow process for which no implicit resolution is required (on the contrary to condensation/evaporation)
Dechapanya (2004), AST	Main sources of anthropogenic SOA precursor emissions in Houston area will be point sources.
De Gouw (2005), JGR	There is very limited insight about the fate of organic carbon: deposition or continued chemical reaction leading to CO/CO2 or uptake by aerosol and then deposition. The suspicion is that most gas phase VOC is uptake by aerosol and then deposited. suggestion: the mechanism for POM formation is more efficient than previously assumed.
Delage (2002), BLM	The downward atmospheric turbulent heat flux is underestimated when calculated using a formulation with a sharp cut-off of the fluxes as static stability increases.           Cold bias is largest for higher canopy.             The region where H decrease with stability is intermittent and very stable surface layer.            Decoupling is unlikely if wind speed stays above 5 m/s.
De Leeuw (2003), AE	possible formation of ammonium in the coarse mode Only for particles of radius larger than 10 um the deposition velocity may be larger than the deposition velocity of gas HNO3, so the formation of aerosol nitrate extends the lifetime of nitrate.
Demerjian (2008), JAWMA	At Fresono, nighttime formed nitric acid above the surface layer will be mixed downward in the morning time and form ammonium nitrate. Around Houston SOA is important for rural area. Peak nitrate concentration in the early morning is observed in NY, pittsburgh etc. Summer periods were characterized by early morning peaks in OC, EC and late afternoon peaks in O3 and SOA.
Dennis (1996), AE	What is "generalized coordinate"; A scheme for constructing coordinate systems that allow transormations among various vertical coordinates and transformations among various horizontal coordinates by simple changes in a few scaling parameters, boundary conditions, map origins, and orientation. Photochemical pathway analysis capability Parameterization of clouds for air quality modeling application is deficient
DERBYSHIRE (1999,BLM)	Decoupling is defined as a cessation of turbulent transport between the surface and the atmosphere due to high stability
Dey (1989), Weath. Forecasting	Three principles of subjective meteorological analysis: spatial coherence, temporal continuity, and adherence to dynamic constraints, which are needed to be considered for the objective analysis methodology
Dibb (2004), JGR	Deposition of HNO3 is the dominant removal pathway of NOx
Dirmeyer (2006), BAMS	3 generations of Land Surface Schemes: 1. bucket schemes; 2. SiB and BATS; 3. MOSES2 and ORCHIDEE There is still a great deal of unertainty in our understanding of land-atmosphere interactions at large scales.
Djouad (2002), AE	2nd order Rosenbrock scheme is used to solve the multiphase chemical kinetics: RADM2 (including both gas phase and aqueous phase reactions)
Donkelaar (2007), AE	simple treatment of SOA production from isoprene oxidation products: a yield of 2.0% by mass. (using GEOS-Chem)
Donnell (2001), JGR	Pollutants in the free troposphere have a longer lifetime as there is no deposition to the surface and temperatures are colder. vertical tansport mechanism: advection, convection, turbulent mixing. Transport by advection includes transport due to orographically forced motions, frontal motions, and slantwise convective and nonconvective ascent
Doran (2003), AE	NOx/NOy can be used to indicate photochemical age. For fresh emissions, this ratio is greater than 0.4
Dowell (2004), MWR	Observation volume, data volume from radar.         Produced analysis were sensitive to the method of producing the initial ensemble.       Each data volume is assumed to be collected simutaneously, which may reduce the accuracy of generated analysis.
Dudhia (1989), JAS	Shortwave radiation is just attenuated by cloud and water vapor and Raleigh scattering.  O3 etc are not considered in the Dudhia SW radiation scheme.
Eder (2006), AE	8-h peak o3: forward calculation method: calculation of last 7 8-h peak concentrations including data from next day. Overprediction of O3 from ETA-CMAQ is due to the excessive downward transport of O3 rich air via CMAQ's convective cloud scheme in conjunction with GFS model derived O3 profiles.
Eder (2006b), AE	Full annual simulation (2001) of CMAQ covering the contiguous US. Bad NO3- simulations is partly attributable to an incomplete understanding of ammonia emissions.
Elbern (1999), JGR	To analyzing the chemical composition over a extended domain, kringing (pure spatial approach) algorithm could be used. Also the advanced sptiotemporal methods (namely, the 4D-var data assimilation technique is increasingly used.
Engel-Cox (2004), AE	MODIS AOD data is only quantitative with respect to PM2.5 mass in East US.
Evensen (2003), OD	... Thus, it is possible to use the EnKF for parameter and bias estimation in dynamic models.
Hansen (2003), JAWMA	OM has a more regional character than BC among SEARCH sites.
Edgerton (2005), JAWMA	FRM lost ~90% of NO3-. FRM, by design, does not have a means of correcting ovlatilization artifacts (volatile of NH4NO3, OM). Two approach for correction: FRM equivalent and BE (best estimate). BE is an attempt to represent waht is actually in the atmosphere. Seasonality of PM2.5 in SE US is primarily driven by SO42- cocentration.
Edgerton (2006), JAWMA	Continuous NO3 is slightly higher than filter NO3. Continuous measurements have not evolved to the point where they can be considered independent of filter-based counterparts.
Edgerton (2007), AE	SEARCH sites NH4+, NH3 observation. NH4+ concentrations are more uniform accross the network than NH3, with only slightly higher at urban sites than neighboring rural sites.
Ek (2003), JGR	implementation of Noah in mesoscale model forecast suit Eta-EDAS
Eldering (1991) AE	Annual average HCl concentration 0.39-1.25ppb (or 0.57-1.84 ug/m3). Coarse sea salt aerosol can in part be converted to fine particle chloride. HCl from Chloride depletion process could react with NH3 to form fine particle NH4Cl.
Engvall (2008), ACP	Accumulation mode particle dominate in Spring but Aitken mode dominate in summer. The strong seasonality of solar angle in the Arctic results in the nucleation potential reaching some criticle value about the same time each year.
Erickson (1999), JGR	In remoste oceanic regions, available acidity was insufficient to titrate all sea salt alkalinity, thus, significant HCl was not produced via acid displacement. However, in these regions virtually all HNO3 was scavenged by sea salt. The pH of aerosols, once the alkalinity was titrated, was uniform amongy the different size bins
Erisman (1993), AE	During dry periods, SO2 and NH3 surface resistances are determined by stomatal uptake. In very dry periods, small emission fluxes of NH3 are observed. CO-deposition of NH3 and SO2
Erisman (2005), EP	NH3 flux increase leading to a decrease in Rc of SO2, which support the need for dynamic modelling of deposition process. Dry deposition is more difficult to measure than wet deposition, so it is rarely monitered routinely. We shouldn't account emission twice in CTM( in both emission and bidirectional deposition)
Errico (1997), BAMS	The principal application of adjoint model is sensitivity analysis, and all its other applications may be considered as derived from it. The adjoint operates backward in the sense that it determines a gradient with respect to input from a gradient with respect to output.
Evans (2004), AE	HNO3(g)+NaCl(s,aq) Chloride depletion process sea-salt and CaCO3 can partition gas-phase HNO3 to solid or aqueous-phase NaNO3 and Ca(NO3)2, which will reduce the direct dry deposition of nitrogen to Tampa Bay
Fan and Sailor (2005), AE	anthropogenic heating is found to have little impact in the daytime UHI development, but to play an important role in the nighttime UHI formation.
Fan, Jiwen (2005), JGR	CMAQ is applied to 24-31 Aug. 2000, TX. Model evaluation is conducted using averaged daily PM2.5 variation (mean diurnal PM2.5 variation).
Fan, Yun (2004), JGR	CPC global monthly soil moisture dataset is produced using a one-layer "bucket" model, driven by CPC monthly global precipitation over land and monthly global temperature from Reanalysis
Fan, Yun (2011), JH	So far, most soil moisture datasets used in research or weather and climate predictions are "calculated", they are either from NNRP, R2, ERA-40, NARR or offline runs of land surface models (e.g., NOAH LSM; CPC leaky bucket soil moisture model)
Fang, Guor-Cheng(2000), Chemosphere	Water-soluble particles such as sulfate, nitrate or chloride can easily become cloud condensation nuclei. Summer is more acidic than in winter. Acidic aerosol concentrations are generally higher during the daytime than at night time. Aerosol acidity is stronger in rural areas than in urban areas, because higher NH3 emissions in urban areas resulted in more neutralization
Faraji (2008), AE	in the summer of 2000 in southeastern Texas, the SAPRC mechanism predicts O3 30-45ppb higher than CB04.
Fast (2000), JGR	After 4DDA, the simulated meteorological field were thus a combination of the predicted and observed variables when and where the data occurred. Heteorogeneous chemistry and wet removal were of secondary importantce for ozone and its precursors within lower troposphere. So they are neglected in this study
Fast (2002), JGR	Only those dats not used for FDDA can be used to evaluate the model. CBM-Z extends the original framework (CBM-IV) to function properly at larger spatial scales and longer time periods. Cloud field formation depends on supersaturation in RAMS meteorology model. Emission control strategies in the NE US need to be implemented for the entire region since the significant contribution from the regional transport PEGASUS aerosol is not mentioned
Fast (2006), JGR	The bins are defined by the dry particle diameters, so water uptake or loss does not transfer particles between bins. Transfer of particles between bins is calculated using the two-moment approach. ASTEEM is coupled with MESA to solve the dynamic gas-aersool partitioning over multiple size bins.
Fatogoma and Jacko (2002), AE	A mixing height estimation model is developed to be used in air quality model, especially UAM-IV. At night time, the mixing height is defined as the height where the shear stress is reduced to 5% of its surface level.
Fernandez-Diaz (2000), AE	Improve the partition coefficients method in Jacobson 1994's semi-implicit coagulation scheme.
Fischer (2005), GRL	Ammonia emissions concentrated in midwestern US, it is temperature dependent, peak during summer time. Warm climatic condition in south of Great lakes is related to high nitrate over north eastern US.
Fischer (2004), JGR	High O3 in northeastern US is associated with westly and southwestly flow, downwind of eastern seaboard or midwest. High altitude sites are oftern used to assess background O3 conditions. Free from influence of local emissions.
Fischer (2006), JGR	Aerosol NO3- exhibited a bimodal size distribution with a primary peak associated with sea salt at 4 um and a secondary sub-um peak. Total odd nitrogen(NOy)
Flanagan (2006), JAWMA	The uncertainties associated with the measurements data: analytical measurement uncertainty and sampling uncertainty. Analytical methods: Gravimetry, ion chromatography, X-ray fluorescence
Floors (2013), BLM	Recent studies have shown that even the state-of-the-art mesoscale WRF model often poorly represents turbulent parameters such as the friction velocity u ∗ and sensible heat flux
Foltescu (2005), AE	Bott advection scheme is used in MATCH. Significant difference exists among different sea salt generation functions. especially for size distribution while bulk mass is less different.
Foret (2006), JGR	Dry deposition is the main removal process of mineral particles in the vicinity of dust source areas. The relative importance of wet scavenging processes increases with the distance to source regions. In the initial stage, dust particles are mainly hydrophobic, reducing the in-cloud scavenging efficiency and below-cloud scavenging is the dominant wet deposition process for pure dust particles. Dry deposition velocity and below-cloud scavenging coef. exibit similarity in terms of size dependency.
Forkel (2006), AE	BVOC fluxes into the atmosphere are 10-15% lower than the emission fluxes on branch basis due to chemical BVOC degradation within the canopy
Fortuniak (2006), TAC	for rural winds lower than 1.13 m s−1 urban winds can be stronger than rural speeds
Fountoukis (2007), JGR	Aerosol size distribution, chemical composition and updraft velocity are inputed to cloud parcel model and NS parameterization, droplet number is predicted. Remarkable clousure is achieved for both parcel model and parameterization.
Fountoukis (2007), ACP	ISORROPIA II is evaluated aganst SCAPE2
Frost ( 2006), JGR	WRF/Chem retrospective simulation starting at 0000 UTC each day, and do 36 hour forecasts. Meteorology IC from RUC, BC from Eta. Lack of plume-in-grid treatment in WRF/Chem. The impact of NOx emission reductions on ozone was moderate during summer 2004 because of relatively cool temperature and frequent synoptic disturbances. Power plant NOx emission decreased by 50% between 1999 and 2003, while SO2 decrease by 13%
Fridlind & Jacobson (2000a), JGR	gas-to-particle conversion includes nucleation, condensation, dissolution, and aqueous-phase oxidation of dissolved intermediates. In the two largest size bins, equilibrium NO3- always exceeded measured NO3-, "subsaturated" , "undersaturated" Equilibration of aerosols with water vaper and CO2 is generally fast due to the high concentrations of those gases
Fridlind & Jacobson (2000b), JGR	size-resolved equilibrium. If deviation occur between thermodynamic model and the observation, we can't tell whether if the aerosol is really out of equilibrium or the thermodynamic model is wrong.
Fuentes (2007), JAC	Ordinarily the emissions used in air quality models are estimated from measurements made at the foliage level, however a high level of chemical processing in the canopy may cause the emission at the foliage level different from the actual emission to the regional scale. The chemical processing within forest canopies reduces the mass and chemical reactivity of BVOCS reaching the regional scale.
Galloway (1993), AE	Large Scavenging ratios are associated with Na+ and NO3-. CH3SO3- and nss SO42- become progressively smaller. This is because the size dependency of SRs. The largest particles are scavenged most efficiently
Gao (1993), JGR	For isoprene, turbulent diffusion and emission dominate chemistry in determining isoprene conc. and fluxes at various height in canopy!!!!!! interesting!!! becasue normally it is thought in-canopy chemistry changes isoprene conc.
Garland (2001), WASP	The particles of <10 um diameter are capable of travelling >1000Km in the atmosphere and so are responsible for much of the long-range deposition of particulate pollutants High RH will increase the dry deposition of hygroscopic particles
Gaydos (2005), JGR	ternary nucleation mechanism works well for nucleation in Pittsburgh. Condensation of sulfuric acid alone is often not sufficient to grow nucleated particle to detectable siezes. condensation of organic species, heterogeneous reaction, ion-enhanced condensation also play roles.
Gaydos (2007), AE	PMCAMx equilibrium approach is used even though hybrid and kinetic approaches are available. Three different representative days are available from inventory: an average weekday, satureday, and sunday. EC in urban areas are significantly overpredicted.
Gaza (1998), JAM	Appalachian lee trough and sea-breeze are important for accurately predict location and magnitude of the peak ozone values in New England.
Gego (2008), JAWMA	Major source of NOx in the eastern US were on-road vehicles (36%), followed by power industry (23%). Emissions along Ohio River Valley appeared to be associated with many regional-scale ozone episodes in the Northeast NOx emission reductions have a greater effect on high O3 days than on days with background O3 level
Geron (1994), JGR	The database of Guenther (1994) provide emission rate of ISOPRENE and monoterpenes standardized for birght sunlight and 30oC for dominant forest tree genera in NA, which needs to be adjusted to ambient PAR(photosynthetically active radiation) and temperature to serve as hourly emission.
Ghan (1997), JGR	predict droplet number based on aerosol number conc. is a key uncertainty in estimates of indirect radiative forcing of aerosols through their role as CCN.
Ghan (2001), JGR	The indirect forcing is rather sensitive to the parameterization of the droplet autoconversion
Gilliland (2006), AE	The most comprehensive data currently available for inverse modeling of NH3 emission is NH4+ precipitation.
Gilliland (2008), AE	Dynamic evaluation, which evaluate a model's ability to accurately simulate air quality changes from given changes in emissions. Emissions from electrical generating units decreased by 41% from 2002-2004.
Ginoux (2001),JGR	GOCART is used to simulate single species, such as sulfur, dust? Particles larger than 6 um generally have short life time, thus this study is restricted between 0.1-6 um
Giorgi (2003), Climatic Change	Indirect aerosol effects: type I: increase CCN, reduce cloud droplet radius, increase cloud reflectivity; type II change the lifetime of clouds: several parameterizations of type I effect are available, which relate the aerosol mass concentration to cloud properties. The transport by resolved scale circulation and cumulus clouds, wet removal by resolved and cumulus precipitation etc are included in the prognostic equation. The cloud water content is calculated by the model's explicit and cumulus cloud parameterizations. M ethod of Hegg(1994) obtain number concentration of cloud droplets as a function of the simulated sulfate mass mixing ratio.
Godowitch (2008), AE	CMAQ IC and BC are specified with the same set of time-invariant, vertically varying troposperic background estimates.
Goldstein (2004), GRL	Forest thinning dramatically enhanced both monoterpene emission and ozone uptake, which suggest O3 was lost via gas phase reaction with biogenically emitted terpens.
Gong (1997), JGR	Total mass of sea-salt is mainly governed by the local wind speed. This wind denpendency varies for different locations. Small particles(0.13-0.25) have longer residence time than larger particles(4-8um) because of different removal rate by dry deposition and gravatational settling
Gong (2003), JGR	The global 3D daily averaged fields of OH, O3, H2O2 and NO3 from NCAR MOZART model are imported to CAM for GCM application. In some cloud droplet number parameterizations, the droplet number only depends on aerosol number, or sulphaste mass concentration. Then the cloud droplet number density is fed to a could module with explicit microphysical processes to treat the cloud-aerosol interaction.
Gong, Wanmin (2006), AR	ISORROPIA is abailable to represent inorganic heterogeneous chemistry (gas-particle partitioning)
Grell (2005), AE	O3's precursors (e.g., CO, NOy) are biased high, which may be due to vertical transport scheme. PM2.5 is biased low in WRF/Chem, which may be because of not considering in-cloud oxidation of SO2, and RADM does not account terpene chemistry, so SOA may bias low.
Grossman-Clarke (2005), JAM	refined land use/cover improved the simulated diurnal temperature cycle and wind speeds. Anthropogeniic heating contribute as much as vertical diffusion to the warming of the air during most of the night while its contriubtion during daylight hours is insignificant.
Grini (2005), BER	Box model study. The equilibrium vapor concentrations are determined for each section seperately based on ISORROPIA, Seems the mass transfer process is solved dynamically.
Grover (2005), JGR	50oC conventional TEOM almost can not measure the semivolatile material (NH4NO3, semi-voc). The normalized bias could be 50%. FRM (Federal Reference Method) also have the same problem.
Guelle (2001), JGR	Three generation fuctions of sea salt particles in TM3 global atmospheric tracer transport model submicrometer sea salt competes with sulfate to influence the number of CCN
Guenther (1994), AE	VOC emission for all woodlands in US is comprised of 58% isoprence, 18% monoterpenes and 24% other VOC.
Guenther (1995), JGR	The ocean is supersaturated with VOC with respect to the atmosphere. Oceans and plant foliage are the two dominant sources of nature VOC
Guillaume (2007), Tellus	bulk equilibrium approach with redistribution is used in ORISAM-TM4 for inorganics GP partitioning. Full dynamic solution is not stable with long time step. For organics, empirical partitioning coefficients Kp are used(Odum et al., 1996). Nucleation, coagulation is not activited.
Guimbaud (2002), ACPD	uptake coefficient of HNO3 to deliquescent sea-salt particles around 0.50 In AIM, each "ion" can equilibrate within 3 phases, i.e., either as true ion in the aqueous phase, as a salt in the solid phase, or as an acid in the gas phase
Gustafson (2007), GRL	WRF/Chem sensitivity to CCN treatment. NEI99 emission is updated for 2004 simulation.
Gustafson (2007), BAMS	NARR can be used as a good proxy for the observed diurnal wind patterns since it assimilated near-surface winds.
Hack (1994), JGR	moist convection parameterization in CCM2. In CCM1, moist adiabatic adjustment procedure is used, which adjusts the lapse rate of a saturated conditionally unstable atmosphere to neutrality. Any water mass condensed in this stratification process is immediately precipitated out of the system. It is argued that it should be possible to predict the time evolution of the large-scale fields by describing only the collective influence of the small-scale elements Cloud fraction in CCM2 depends on RH, vertical motion, static stability, and the convective precipitation rate
Hakami (2004), JGR	First-order coefficients, or slopes of the response curve provide valuable information in linear ranges of response. Higher-order sensitivity coefficients provide additional information to describe the curvature of the response curve. HDDM(High-order decoupled direct method)
Han (2008), AE	Pleim-Xiu and NOAH LSM are superior to the slab model
Hanley 2019 QJ	very nice introduction of PBL schemes and mixing scheme for LES
Hara (1999), JGR	Nitrate is observed mostly in coarse mode in Arctic in winter (no or weak solar radiation). In addition to uptabke of HNO3, heterogeneous reactions of N2O5 and NO3 may also contribute to formation of particulate NO3-
Huang, Ho-Chun (2001), JGR	Reference solution from LSODE with RTOL 10-7, ATOL 10-11 ; Comparison between the performance of chemical solvers should be made within the transition stage---the period before quasi-equilibrium
Hamill (2000), MWR	rank deficiency,         resulted in about 1-day gain in forecast lead time.       Hybrid EnKF-3DVAR,    background error statistics are estimated from a linear combination of time-invariant 3DVAR covariances and flow-dependent covariances developed from ensemble of short-range forecasts.
Hamill (2001), MWR	covariance estimate from the large ensemble could be taken as the true covariance.  The optimal magnitude of the inflation was a function of ensemble size, the smaller the size of the ensemble, the larger the inflaction.
Hamill (2005), MWR	Model errors may be caused by many reasons, e.g., insufficient model resolution, inperfect parameterization,  In this paper, model error due to insufficient model resolution, i.e., truncation of model and the resulting lack of iinteraction with smaller scale motion is investigated. Two model error parameterizations, i.e., covariance inflation and additive error are tested. Seems additive error approach is more accurate.      A single inflation factor may not be optimal over all parts of a model domain.
Hanna (1996), AE	UAM-IV, UAM-V, and ROM/UAM-IV are used for regulatory application and all of them use offline diagnostic or prognostic meteorological field. Altough a prognostic meteorological model could be used, in most UAM-IV applications the wind inputs are defined by interpolating surface and upper air observations. Two days for spin-up, to minimize the influence of initial conditions.
Hanna, S.R.(1998) AE	Total model uncertainties include uncertainties in specifying input parameters, deficiencies in model physics and chemistry assumption, uncertainties due to natural or stochastic fluctuations in the atmosphere
Harrison (1994), AE	Particle size control deposition velocity uncertanties associated with atmospheric inputs are greater than other pathways
Hass (1997), AE	Several EUROPEAN regional air quality models (EMEP, EURAD, LOTOS and REM3) use either prognostic meteorology predictions (from NWP or MM5) or diagnostic interpolation of observations.
Hara (2008), ACPD	The bias in aerosol estimation does exist, "however, it is not critical because we are mainly targeting semi-quantitative analysis of dust transport (i.e., a comparison of dust vertical structure and horizontal scale) using the CALIOP measurement."
Hartley (1994), JGR	Gases' indirect effects on climate: through the interaction with other gases and free radicals. How the balance between PBL mixing and convective mixing serves to transport tracers in the real atmosphere is not well known
Hauser (2005), GRL	AOD retrieved from satallites is only for specific time per day. So if you want to compare model prediction with satallite AOD, be careful. Modis AOD is biased high. Modis typically exceeds the AVHRR values by about 0.2
Hayami (2005), AE	in sea area surrounded by China, Korea and Japan, gaesous HNO3 and ammonia were found to exist at low concentration and low fractions relative to the total (gas+aerosol) concentration. Coarse/Fine/Gas in ug/m3 (2.18/2.50/0.21; 3.21/0.94/0.20; 2.49/1.14/0.19; 2.63/1.53/0.20) IF Coarse/Fine in ug/m3/Gas in ppb (2.18/2.50/0.081; 3.21/0.94/0.0776; 2.49/1.14/0.0737; 2.63/1.53/0.0776)
Heald (2006), JGR	Transpacific Asian pollution component is mainly sulfate. The GEOS-chem predicts no significant Asian enhancements of nitrate and OC in the US because of scavenging during Asian outflow. MODIS AOD over land is biased high.
Hegarty (2007), JGR	Relation between O3 and circulation type accross northeast US for summer 2000-2004. The high O3 levels are associated with the map type that featuring the Bermuda High. Some improvements for the analysis in this paper: Circulation pattern be characterized with a nonlinear function.
Held (2004), AE	Ammonium and nitrate distributions at Backersfield exhibit the same mode, qualitative shape, which is totally different from Tampa, FL data. So aerosol at Tampa is affacted by sea spray significantly, but not at CRPAQS
Helmig et al. (2007), AE	Sunrise ODEs are not observed at the elevated site Summit (3212m AGL). which confirmed that the ODEs are confined in boundary layer
Helmig et al. (2007), AE	O3 increase with elevation is generally linked to the transport of O3 from the stratosphere and loss processes occurring near the earth surface. Barrow is incluenced by diurnal sea breeze occurances, which likely affect O3 dynamics. O3 in polar region peaks in winter due to less sinks during the dark periods. A few papers regarding low O3 in marine boundary layer are cited
Helmis (2013, BLM)	Focused on a 5-day period of persistent south-westerly (marine) flow, the analysis of sodar and radiosonde data revealed the presence of a stable layer associated with increased static stability just before the emergence of low-level flow acceleration. As indicated by the Richardson number profiles, the increased stability of the lower MABL suppressed turbulence, allowed the decoupling of LLJ from friction, providing a favourable environment for the development of inertial oscillations. Significant amplitudes of inertial motions, which were confirmed by the application of a Hilbert–Huang transform, are associated with the acceleration at the LLJ’s core, due to the frontal events and the subsequent frictional decoupling, both leading to a modification of the large-scale flow structure.
Henze (2004), JGR	The discrepancy between the trial solution and what is known from observations is measured by the cost function. The optimal model solution and parameters are found by solving the minimization problem. The governing equation of growing/evporating as a result of gas-to-particle conversion is the frequently seen condensation equation.
Hering (1997), AE	Nuclei mode was attributed to fresh combustion sources. Bimodal character of accumulation mode (condensation, droplet) is confirmed in the volumn or mass distribution.
Herner (2006), AST	nighttime particulate carbon concentration is higher than daytime concentration, which indicate its sources are mainly from primary emissions. Coagulation must be considered in simulations of PM size distribution in the SJV during winter episodes.
Hertel (2006), EC	Deposition of NO2 could be also bi-directional. Compensation point may vary by a factor of two throughout the growth of the plants. NH3 flux may vary significantly over the year, this variation is believed to be associated with variations in the canopy compensation point.
Herut (1999), limnol Ocean.	The anthropogenic increase in nutrient input are thought to increase the total amount of carbon fixed in the upper ocean Phosphorus is present primarily in the coarse fraction
Hidalgo (2008), MAP	However, few is known on the dynamics and structure of the urban-breeze circulation in an inland city. Only one study (Wong and Dirks 1978) analyses urban thermally induced motions using observational data for an inland city.
Hidalgo (2008b), MAP	The dynamical perturbation on the ABL due to the rough- ness of the city is only significant up to 50 m of height, the urban breeze circulation being caused by the pressure gradi- ent due to the UHI-induced thermal effects.
Hidalgo (2010), JAMC	Urban breeze is difficult to observe with a simple observational deployment
Hidy (2000), AE	while precursor emissions and atmospheric chemical processes are obviously important, troposhpheric O3 is clearly very sensitive to meteorological condition.
Hodzic (2004), JGR	Simulate the lidar backscattering profiles from the model (CHIMERE) outputs and compare with observed actual lidar signals. The gas phase chemistry scheme has been extended to include sulfur aqueous chemistry and heterogeneous chemistry of HONO and nitrate.
Hodzic (2005), AE	Describe the artifact of TEOM. A routine TEOM dries the sampled air stream by heating the inlet at 50oC to remove water, which could cause the evaporation of semi-volatile species.
Hodzic (2006), AE	A simply mineral dust erosion parameterization! The heterogeneous reaction could be treated either through thermodynamic or through kinetic approach. The nitrate heterogeneous reaction efficiency strongly depends on the uptake coefficient. 0.1 is recommended.
Hofzumahaus et al. (2009), Science	confirm a new OH regeneration pathway, similar with Lelieveld et al., 2008, nature
Hogrefe (2001), AE	With finer grid resolution, air quality predictions might not become more accurate because some model inputs (e.g., emissions, land-use patterns, soil moisture, etc.) are not accurately resolved to the corresponding spatial scales.
Hogrefe (2004), JGR	Possible feedback of climate change on land use and vegetation cover was not considered. MM5 is coupled to GISS in a one-way mode through IC and lateral BC. 3-day spin-up for o3 CMAQ makes a zero-flux assumption for the top boundary, thus stratospheric ozone influx is not considered Change in global atmospheric compositions is approximated by changed BC for regional model. It is found change in BC is the largest contributor to changes of O3 concentration.
Hogrefe (2006), AE	GEOS-CHEM simulation is used as BC for 2001 Annual CMAQ simulation. Scale analysis is used to evaluate Air Quality model simulation.
Hogrefe (2007), JAMC	CMAQ predicted a pronounced double-peak structure for diurnal cycles of PM2.5 and its components. Even tough forecasts of PM2.5 are started in 2004, but it is experimental and not released as guidence products. combination of uncertainties in vertical mixing, magnitude and temporal allocation of emissions, and deposition processes are all possible contribution to the CMAQ overprediction of PM2.5 over New York state.
Hole et al. (2008), EP	Zhang, 2003's big-leaf Rc dry deposition module is used. Improvements are needed to include more processes such as bi-directional exchange of NH3 and co-deposition.
Holland (1997), JGR	nitrogen deposition alleviate nitrogen limitation of productivity in terrestrial ecosystems, thus may contribute to enhanced terrestrial carbon uptake. GCTM, GRANTOUR, IMAGES, MOGUNTIA, and ECHAN are compared.
Holtslag (1995), BLM	Also use maximum of two Kz at the top of the PBL (to increase entrainment exchange?)  Vertical exchange is more pronounced with nonlocal scheme than local scheme because in local scheme no direct account is given for influence of large-eddy motion.       The ratio of the averaged entrainment flux for heat to its surface flux is about -25%.       Local diffusion doesn't show entrainment effect.      Non-local scheme transport more moisture away from surface and deposit the moisture at higher level.     Non-local scheme is less sensitive to vertical resolution than local scheme.
Holzer (2003), JGR	Trajectory analyses typically cannot quantify dilution because of entrainment and mixing en route. MATCH is an "off-line" model that does not integrate prognostic momentum equations
Holzinger (2005), ACP	Chemical O3 loss is linked with biogenic emission of VR-BVOCs (Very reactive); global terpene emission and SOA formation may be larger than previously assumed
Hong (1996), MWR	The countergradient term play important role in transporting the lower-layer moisture upward.    Critical bulk richardson number (Ribcr) could determine the PBL height thus effective mixing.    Precipitation is sensitive to Ribcr.      MYJ scheme is in the strictest sense local diffusion schemes and had a strong tendency to under-entrain in the presence of a strong capping inversion (Ayotte et al., 1996).
Hong (2006), MWR	YSU pbl scheme. improvement over Hong-Pan96: explicit treatment of entrainment from free troposhere. specification of pbl height using a smaller thermal excess and a zero critical bulk Richardson number.
Hong (2004, JMSJ; 2009, JAMC)	no cumulus is used in the 3-km domain.  Cumulus is important to reproduce the heavy rain in US. But it plays insignificant role in reproducing heavy rain over Korean.
Horowitz (2003), JGR	Convective mass fluxes are rediagnosed by MATCH, using the Hack (1994) scheme for shallow and midlevel convection and Zhang (1995) scheme for deep convection. Heterogeneous reaction of N2O5 and NO3 on sulfate aerosols are included in MOZART-2. Soluble species undergo wet removal by in-cloud scavenging. In addition highly soluble species are also removed by below-cloud washout.
Horowitz (2007), JGR	Organic nitrates (mostly isoprene nitrate) could stronly affect boundary layer concentrations of ozone and nitragen oxides. Isoprene nitrates consuming 15-19% of emitted NOx, of which 4-6% is recycled back to NOx and the remainder is exported as isoprene nitrates (2-3%) or deposited (8-10%).
Horvath (2005), AFM	Single-layer Xc model is used even though 2-layer, 3-layer Xc models are available. Right selection of the input parameters such as Gamma_s is important. Above nonfertilized and non-grazed grasslands, generally deposition dominates while large ammonia emission are detected from surface with high N-input.
Houtekamer (1998), MWR	analysis diverge from the real state (filter divergence)           data very dense=> prohibitively expensive => superobservation.
Houtekamer (2001), MWR	In a sequential EnKF, quality of the analysis is almost independent of the number of batches. This supports the used of a sequential algorithm.      A schur product is used to filter the small background error correlation associated with remote observations.     Since the correlation function is smooth and monotonically decreasing, the produced analysis increments should be smooth.
Houtekamer (2005), MWR	An obs are grouped in a time window of 6h centered on the analysis time in data assimilation.   It is assumed that all observation are valid exactly at central analysis time.        Assimilating surface observation doesn't demonstrate a positive impact.             Use a vertical Schur product to limit the vertical extent over which an observation may have an impact, i.e., vertical localization.
Houtekamer (2005), QJRMS	model error realization?    length of increments?    Forecast increment should be larger than the analysis increment.             localization operator may cause imbalance.   Balance operator may help and relaxing localization operator.                    A digital filter finalization.    The impact of observations can be smoothly reduced for increasingly distant grid points using a schur product to permit localization.     Frequent switching between assimilation and model evolution may be costly and detrimental.         Use of the same ensemble for estimation of K and in the estimation of analysis error reuslts in a dependent error estimate, hence, a systematic underestimation of the uncertainties in the analysis.      EnKF provide a Monte-Carlo like estimate of the uncertainty in the analysis.     covariance tranport eq.
Houyoux (2000), JGR	Limited the number of layers used for emissions to 12 (~2000m) because this is a realistic maximum plume rise height. So for NEAQS2004 emission, it is questionable that the emission extends to layer 34!
Huang, X.-L. (2002), GRL	Cloud variability in the GCM is significantly smaller than that in the real atmosphere.
Huang, X.-L. (2005), GRL	Reanalyses is found to have unrealistically large interannual variability in both upper and middle tropospheric humidity comparing to HIRS dataset, while AM2 did good job. So Reanalysis need improvement. Before reanalysis can be used for either diagnositc studies or model evaluation, the fidelity of their products must be tested against observation, e.g., HIRS.
Huang, X.-L. (2005), JGR	AQUA: carries of AIRS and MODIS. AIRS has information for cloud vertical profile. CAM2 is used. Water vapor transport and cloud in climate model is still far from satisfactory.
Huang, X.-L. (2006), JGR	Negtive bias of OLR from AM2 model comparing to ERBE observation can be explained by a too humid layer around 6-9 km in the model AM2. which maybe caused by model deficiencies about water vapor transport.
Huang, X.-L. (2008), JGR	Use Aqua CERES level 2 data product. A forward radiative transfer model, MODTRAN5. AM2 tends to be more humid than the observations in the middle and upper troposphere for all tropical oceans.
Huang, Y et al. (2007)	Menon [2002] concluded anthropogenic aerosols have increased precipitation in southeast China but suppressed precipitation over northeat China. Menon [2004] mentioned that the effect of absorbing BC could be quite variable in both sign and magnitude depending on its verticl location relative to the clouds. Aerosol-Cloud interaction is extremely difficult with large uncertainties. Autoconversion scheme?
Hudman et al. (2007), JGR	The boundary layer NOx data provide top-down verification of a 50% decrease in power plant and industry NOx emissions over the eastern United States between 1999 and 2004.
Hughes (2000), EST	Major nitrate formation parhways were found to be aqueous reduction of NO3 radical and heterogeneous hydrolysis of N2O5
Hunt (2004), J. Phys. Chem	The oxidation of bromide by O3 at the interface may potentially produce several ppt of bromine in the marine boundary layer at night
Husar (2001), JGR	The absence of transported large particles implies that the dust particles above 10 um were preferentially removed by gravitational settling during the 2-3 day atmospheric transport time Yellow discoloration of clouds; Subsidence of the dust layer from the midtroposphere to the surface The fast (>12m/s) trans-pacific transport also implies the dust layer must bave been well above the marine boundary layer The aged Asian dust arriving at west coast had a mass median diameter of about 2-3 um During the dust event, the samples with the highest dust concentration had virtually no sulfur content.
Husted (2000), AFM	Apoplastic NH4+ and H+ concentrations differed between leaf heights but values were relatively stable over time diurnally. But compensation points show clear diurnal variation due to temperature effect. Quantification of NH3 flux is the most uncertain factor in then compilation of a global NH3 budget.
Hutchison (2004), AE	Aerosol products are retrieved by MODIS only during daytime and cloud-free conditions. HYSPLIT: http://www.arl.noaa.gov/ready/hysplit4.html
Hyslop (2007?2008?), AE	The collocated precisions tend to be better for techniques that are performed on the entire filter (gravimetry and IC) instead of just a portion of the filter (XRF and TOR). IMPROVE collocated precision and uncertainty. didn't quite get it
Jacobi et al. (2006), JGR	Bromine explosion the origin and spatial and temporal extent of ODEs are not well understood. O3 depletion was confined to the boundary layer whereas normal O3 levels existed above the temperature inversion. temperature below -20oC coupled with open water favor formation of frost flowers. Frost flowers surface provide optimal conditions for the release of halogens because they are characterized by high slimities. Sea-salt released from frost flowers also contribute to the halogen activation.
Jacobi et al. (2010)	Although we cannot exclude that the O3 transitions to elevated levels are also influenced by the downward mixing of free tropospheric air masses caused by the approach of the lows, it seems likely that in most cases transport from open water areas at lower latitudes is dominating.
Jacobs (1997), JAM	CO2 enrichment=>rs increase=> canopy dries out=> stomata close further=>rs increase further
Jacobson (1994), AE	Semi-implicit coagulation scheme is developed. Fuchs's Brownian diffusion kernel is used.
Jacobson (1995), AE	SMVGEAR II, sparse matrix, vetorized Gear code. In this version, the grid cells in each region of the atmosphere is reordered according to stiffness of the chemical equation to increase the speed.
Jacobson (1996), JGR	When only growth and one particle are considered, gas far removed from the particle diffuses toward and may deposit on, condense on, or dissolve in the particle untill the partial pressure of the gas equals the particle surface vapor pressure.
Jacobson (1996), AE	Online GATOR/MMTD, emission and IC are critical for good model performance. Chemistry took 72% of total computer time.
Jacobson (1997b), AE	Aerosol codes in GATOR, moving-center size structure
Jacobson (1997c), AE	Evaluation of aerosol prediction from GATOR/MMTD. Spinning up the meteorology model may not always improve model performance. O3 strongly depends on IC values, especially of reactive organic gases. Constant aerosol BC over time. sensitivity run: zero BC for aerosol. Seems model is not sensitive on aerosol BC in this 2-day case(emissions and gas-to-particle conversion).
Jacobson (1999), AE	all equilibrium models are notorious for consuming significant computer time in 3-D models. If the model did not work, if data were insufficient or if the system were out of equilibrium, the calculated equilibrium distributions would not match the initial distributions Ca and Mg affect nitrate and chloride partitioning through both charge balance and impact on activity coefficient. So it is hard to tell removal of Ca and Mg will increase or decrease nitrate and chloride
Jacobson (2002), JGR	Condensation vs. Dissolution growth. saturation vapor mole concentration is fixed or not
Jacobson (2005), JGR	OPD (Ocean predictor of Dissolution)-EQUISOLV O is developped to solve air-ocean transfer equation explicitly. There are two advancements of EQUISOLV O over its atmospheric counterpart EQUISOLV II: new method to initialize charge and a noniterative solution to water dissociation equation because that ocean water is only a trivial function of ocean composition.
Jacobson (2006), AE	Jacobson developed the first "online" coupled air quality-meteorological model: GATOR/MMTD
Jacobson (2007), JGR	Both direct and indirect effects are investigated in south CA using one-way-nested global-through-urban GATOR-GCMOM. Approach: baseline simulation with emission-sensitivity simulation without emission.
Jaegle et al. (2004), JGR	Microbial soil emission of NOx in June over southern Sahel in Africa is significant, but not in August
Jaffe (2005), AE	All component of PM2.5 in IMPROVE site of westen US show maximum in summer and minimum in winter, which may due to the precipitation peaks in winter and also enhanced source in summer. Dust crossed the pacific in free troposphere in ~5 days and was brought to surface sites of US by large scale subsidence and orgraphic effects. Limitation of isentropic back trajactory
Jang, J.-C. (1995), AE	Emission and meteorological input data are used at 1-h intervals and linear temporal interpolation is employed to fit these data into the advection time step used in the HR-RADM.
Jang, J.-C. (1995), AE	To improve model accuracy, it is also necessary to have adequate vertical grid resolution
Jang, M (2002), Science	Inorganic acids catalyze carbonyl heterogeneous reactions and consequently lead to a large increase in SOA production, which has not been accounted in previous theories. In the past, SOA formation is addressed primarily by condensation or thermodynamic gas-particle partitioning. Altough ammonia can reduce acidity, the long-range transportation of ammonia can actually result in acidification and eutrophication
Janjic (1990), MWR	Level 2.5 MYJ scheme is used to treat turbulence in PBL and free atmosphere.   Monin-obukhov similarity theory could be used to represent surface layer, but the surface layer can not be as high as 100m for Monin-obukhov theory to be valid.  So MYJ level 2 turbulence closure model is used to represent surface layer, which is valid in the lowest few hundred meters.
Jankov (2005), WF	Predicted precipitation from WRF shows highest sensitivity to the choice of convective treatment, with less sensitivity to PBL scheme, and the least to microphysics. PBL scheme could affect temperature and moistrure profile, which could interact with other schemes such as convective parameterization to influence simulation of precipitation.
Jankov (2007), JH	to quantify the impacts on simulated rain volume due to change in PBL schemes, the factor seperation method was used.
Jazcilevich (2002), AE	Lower temperatures tend to delay and reduce O3 peaks and vice versa when temperatures increase.
JaZcilevich (2003), AE	The convective landing of reactive and non-reactive pollutants can explain sudden significant changes in concentration level in some specific areas. The amount of pollution brought down by convective currents is much more significant than the amount that could be carried down by vertical diffusion.
Jazcilevich (2003), JAWMA	sensitivity of air quality on different land coverings at the future airport.
Jazcilevich (2005), AE	It is very important to predict the confluence line in the right location and time because of the sharp concentration gradients around the confluence line.
Jenkin (1997), AE	A detailed degradation scheme for VOCs is provided.
Jericevic (2009),BLM	a more efficient transport of momentum than for heat in stable conditions.
Jeuken (1996), JGR	The "nudging" method relaxes the model state toward observational data during the assimilation period by adding a non-physical relaxation term to the model equations. The variation of G(relaxation coefficient) in time and space ECMWF data is used for both the assimilation itself and for comparison afterwards with ECHAM model results
Jiang, D-H (2004), AE	Pollution index forecast in Shanghai using an artificial neural network model
Jiang, G-F (2003), AE	The MM5 wind field were "passed through" the CALMET diagnostic wind model with minimal change and combined with the boundary layer scheme in CALMET to provide winds and boundary layer parameters for direct use in photochemical model.
Jiang, G-F (2004), AE	Exactly the episode we are using! Aug 28, 29 were used for "spin-up". RAMS-PEGASUS, aerosol is not considered Most of trace gas emission rate were obtained from TCEQ and SMOKE inventories. Stack emission rates of alkenes from TCEQ were significantly underestimated
Jimenez (2006?2007), Environ. Modelling Soft.	IC may be minimized through a proper spin-up time of 48h, the importance of BC becomes essential and their contribution in O3 concentrations increase with the time of simulation. O3 is long lifetime species (1 month in troposphere, from IPCC)
Jirak and Cotten (2006), JAMC	Increased number of atmospheric aerosols would act to surpress the precipitation process through the creation of a narrow droplet specturm, which has a small collection efficiencies and small collection kernels.
Johnson (2001), GRL	Globally averaged tropospheric ozone in the 2090s was decreased by around 8 ppbv by the effect of climate change mainly through reduction of the net production of ozone in the model.
Johnson (2004), QJ	nice LES shows the decoupling cloud layer during daytime due to cloud radiative heating
Jones (2007a), AE	Rc and Rw increase with increasing of NH3. Night time, stomata are closed so Rc is equivalent to Rw. Very mechanistic. Based on this paper, Rw=1.13NH3+4.6 is used in Walker, 2008
Jones (2007b), AE	Conclusively demonstrated that increases in Rc with increasing NH3 concentration for a whole range of species and vegetation communities in both stomatal and non-stomatal bearing plants. Stomatal compensation point is low comparing to the high atmos. concentration, which leads to unidirectional transport to stomata. At low NH3 concentrations, most deposition to the plant is to the cuticle, with less through the stomata, At very high concentration the NH3 deposition is mainly through the stomata.
Jordan (2003), JGR	The uptake of NO3- and SO4= can be substantial and in some cases may entirely drive off the CO3= in the dust aerosol.
Junquera (2005), AE	Unless the PIG (Plume-in-Grid) module is turned on, emissions from each point source are instantly dispersed into an entire grid cell volume For fires smaller than 0.4km2, the plume rise value is set as top of CAMx layer 3 (170.5m). For Fires between 0.4 and 3.2 km2, it is set as top of layer 4(256.9m)
Kallistratova and Kouznetsov (2012) BLM	Both in summertime and in wintertime, the jet-core height is close to the height of the top of the ground-based or elevated inversion. This indicates the tight link between these two phenomena, so the emergence and properties of LLJs and temperature inversions have to be studied together The jets at the urban site appear more seldom, emerge later, are weaker and located higher than those at the rural site. These differences are likely caused by the storage heat and higher roughness of the urban surface, which influence the thermal structure of the urban ABL the UHI results in a significant increase in turbulent exchange in the urban ABL. Thus the UHI leads to the intensification of the removal of pollutants emitted close to the surface, from the lower part of the urban ABL. A similar positive effect of UHI has been recognized recently in Hong Kong
Kalnay (1996), BAMS	NCEP/NCAR reanalysis project. The NCEP global spectral model is used in the assimilation system. The model includes parameterizations of all major physical processes.
Kanamitsu (1989), Wea. Forecasting	The assimilation system consists of the processing of observed data, objective analysis, initialization, numerical forecast and postprocessing. The spectral method has a number of advantages over the grid point method. The principle of the spectral method is to represent fields by a set of mathematical functions instead of the values at dicrete grid points.
Kanamitsu (2002), BAMS	NCEP-DOE AMIP-II Reanalysis (R-2), which is updated from NNRP
Kane (1994), AE	Wet deposition of NO3- over sea is 3 times of that over the adjacent coastal sites, which imply transfer of nitrate from fine to coarse particle mode.
Kang (2005), JAWMA	discrete-type and categorical-type evluations. Only one of the three models, MAQSIP-RT is better than persistence.
Karl (2003), JGR	surface high pressure system on Aug. 30-31. O3 production is driven by low molecular weight alkene-Nox chemistry. O3 control need to target VOC emission. High molecular weight VOC emission inventories need reevaluation and update.
Karydis (2007), JGR	bulk equilibrium followed by re-distribution is used. EC emission is adjusted by lowing by a factor of two. 0.54 tons/d EC is used in this study.
Kasibhatla (2000), GRL	Measurements do no support the significant impact of ship emissions on the NOx and NOy in MBL predicted by global models. Basically global models overpredicted the impact of ships on NOx. So ship emission may not be that important?
Kasibhatla (2000), GRL	Tuning a model to reproduece characteristics of a particular episode may have the consequence of degrading model performance during other periods. Clean backgroud concentrations of O3 (35ppb) and related species were prescribed at AQM boundaries.
Katoshevski (1999), JAS	The number distribution of remote MBL aerosols exhibit a characteristic bimodal shape, which is seperated around 0.1 um. The minimum around 0.1um is attributed to could processing of aerosol. At pH 5.5, S(IV) oxidation by HOCl is predicted to dominate. Free tropospheric aerosol entrainment tends to sustain particle number concentration while sea-salt maintain most of the mass concentration. An approximate lower activation diamter of 0.1 um in marine stratocumulus. Accumulation mode aerosol is thought to comprise the majority of CCN in marine stratiform clouds. Both nss-sulfate and sea-salt particles will play important role for CCN formation
Kelly (2005), JGR	Thermodynamic equilibrium between carbonates and nitrate acid/HCl. normally carbonate scavenge HCl and HNO3 from gas phase. But under some condition, these forward heterogeneous reaction may be hindered. Even release HNO3 and HCl into the gas phase.
Keene (1998), JAS	model study indicates sea-salt alkalinity is titrated rapidly (within 15 min) by H2SO4 and HNO3 after which PH drops below 5, the oxidation of S(IV) by O3 ceases.
Keene (2002), GRL	Sea-salt may be acidic! which correct our previous misunderstanding that sea-salt is alkaline or circumneutral. Oxidation of SVI by O3 may be less important since it is limited to alkaline condition.
Keene (2004), JGR	intercomparison between the aerosol acidities from direct PH measurements and that inferred from phase partitioning of HCl, HNO3 and their thermodynamic properties. Aerosol in easten coastal air during summer are acidic
Keene (2007), JGR	Based on the observations at a coastal site in New England, Na+ is mostly in super-um size fractions, relative to fresh sea-salt composition, most samples were significantly depleted in Cl-
Kerminen, and Wexler (1995), AE	Coagulation of smaller size particles was more rapid and especially newly born nuclei were believed to coagualte significantly into larger particles under appropriate condition, self conagulation was estimated to be negligible. Explicit treatment of condensation is required during most stages of the plume development
Kerminen (1997), AE	HNO3(g)+NaCl(s,aq)!!!!!!!!!!!!!!!!!!!!!controdict with Ten Brink(1998); Sodium, tracer of sea-salt particles
Kerminen (2001), AE	Less acidic nature of the Aitken mode induce that it can take up semi-volatile, water soluble gases much easier than the acidic accumulation mode. Aerosols are mixture of acidic, netrual and even alkaline particles, so sufficient size resolution is important
Kerminen (2001), JGR	In the absence of clouds and fogs, the principal sink for the vapors H2SO4(g) and MSA(g) were their condensational transport to submicron particles
Kerminen (2002), JAS	Formation of new particles is two-stage process: nucleation and growth to the first size section of the model. An analytical formulae is derived in this paper to handle the second step.
Kerminen (2005), GRL	Significant fraction of cloud droplets activate on Ultrafine (<0.1 um) aerosol particles, especially under clean or moderate polluted environments. In some cases, 2/3 could droplet originate from ultrafine particles. 100 aerosol formation events are observed in a few years in the Pallas station.
Khain, 2005, QJRMS	Under continental condition, more aerosol=>droplet size decrease=>retard collision and rapid raindrop formation=>higher chance to transport to higher level=> large area for sublimation and evaporation and small sedimentation velocity extend the period of fall, fostering mass losses by sublimation and evaporation=> so precipitation loss is high.
Khattatov (2000), JGR	y=A(I(x)). I is the horizontal linear interpolation operator, operator A denotes the relationship between the observation and estimated concentration at the locations of observation. The solution of inverting the eq. is usually called the analysis, xa
Khlystov (2005), JGR	Water absorption by sulfate is enhanced by organics
Kilpelainen (2012), qj	temperature inversion strength and depth were notably underestimated by the model. The underestimation was likely a consequence of excessive mixing in the ABL and deficits in the representation of the surface energy balance in the model
Kim (1990), JCIS	Moveing sectional method uses condensation/evaporation rate as the characteristics of the section boundaries movement. It is unlikely that a multicomponent coagulation and cond./evapor. simulation method will involve moving sections
Kim (1995), AST	include crustal species in SCAPE2. The Pitzer method is chosen as multicomponent activity coefficient estimation method.
Kim, D and stockwell (2007), AE	Meteorology is compared with NCEP and NARR reanalysis data for model evaluation. Meteorology parameters may vary over very short time period. So on online MM5-Chem(gas only) is used for Mountain area.
Kim, K. W. (2001), AE	Reconstructed light extingction coefficient from 6 aerosol components based on IMPROVE proposed equation VS. measurements from transmissometer.
Kim, S.-W. (2006), GRL	In Ohio River Valley, 2004 NO2 columns and NOx emissions decreased by ~34% from 1999 levels indicated by GOME and SCIAMACHY satellite observations. But such trend is not clear in northeast corridor, where mobile source may play more important role.
Kleeman (1997), JGR	internal mixing treatment can distort aerosol size and composition distribution during conditions when significant amounts of sulfate and sodium exist in externally mixed particles and when externally mixed particles containing material with different hygroscopic properties are exposed to high relative humidity.
Kleeman (1998), AE	in the absence of fog conditions, mass transfer is calcualted by AIM. During periods of fog, a fog model is used.
Kleeman (2001), EST	The first-order processes that dominate air quality are emissions, advection, diffusion, and deposition. The concentrations of secondary pollutants such as ozone and secondary particulate matter also are influenced by higher order atmospheric processes including gas-phase chemical reactions, particle pahse chemical reactions and phase-transfer processes (gas to particle partitioning). AIM used to calculate the condensation/evaporation of semivolatile inorganic components was modified to describe N2O5 hydrolysis on wet airborne particles. In the external treatment, particles exist separately from one another and interact only through exposure to the same gas-phase atmosphere. The greatest contribution to regional PM2.5 is associated with the accumulation of secondary ammonium nitrate. external mixture model is computationally slow
Kleeman (2005), AE	Practical levels of ammonia control has little effect on reduction of particulate nitrate in the SJV since it is ammonia rich there. Reactive nitrogen NOy=NOx + reactive products of NOx
Kleinman (2000), JGR	Generally when wind is from W-SW, along the north-east corridor, New York city is subjected to high O3 pollution. while if the wind is from NW, the air is much cleaner. O3 in NewYork is more sensitive to VOC reduction. In this respect, NewYork resembers LA more than it does smaller cities such as Atlanta, Nashville
Kleinman (2002), GRL	High reactive olefins concentration contributed to O3 conc. in Houston several-fold greater than elsewhere.
Kleinman (2004), JGR	The 1999 July episode in Philadelphia was unusual in that the peak O3 conc. were due in large part to stagnation occuring on that morning. It is more usual in this region for O3 episodes to occur under conditions that favor moderate transport along the northeast corridor.
Kleinman (2005), JGR	O3 smog episode=10-30 north hemisphere background + 20-70 regional background + "same-day" O3 Different regions are characterized with different features: high isoprene in Nashville, dry air and low VOC reactivity in Phoneix, emission of light olefins in Houston.
Kleinman (2007), JGR	Organic aerosol/CO is examined as a function of photochemical age marker (NOx/NOy). When NOx/NOy drop to 10%, organice aerosol/CO indicate 75% of total organic aerosol is secondary.
Klemm (2006), AE	Biogenic direct source of HCHO; The application of one-dimensiaonal model to describe vertical bio/atmos exchange process was approriate
Kley (2007), JGR	O3 is quasi-conservative on the convective time scale during which MBL is rapidly transported to upper troposphere where it is detrained, by passing the middle troposphere
Knipping (2003), EST	sea-salt-derived emissions are highly uncertain. Reactive chlorine species are more important on a global scale than on a urban scale.
Knipping (2006), JGR	MM5-CMAQ could not consistently reproduce the transport of tracers emittted from distant individual sources to specific locations. The transport from Mexico contribute sulfate significantly at a U.S. border city.
Koo (2003), AE	The second-generation SOA models assumed that the SOA species form a solution. In this description, the SOA species can partition into the aerosol phase even if their gas concentration are below their saturation levels.
Korhonen (2004), ACP	UHMA: sea-salt and nitrate are not included yet. Hybrid moving-center, full-moving+retrack sectional size structures are tested. Condensation of sulfuric acid and organic compounds are treated dynamically while water and ammonia is trated equilibrium
Korhonen (2008), JGR	It is haze in spring bu cleaner conditions in summer in Artic area. Climate warming is proceeding faster in Arctic than in any other region of the world. The effct of Aitken particles on climate is an important open question. The effect of Aitken particles on CCN depends on updraught velocity. The source of Aitken mode particle in Arctic Area is not completely understood.
Kruit (2007), AE	Derive canopy compensation point from flux direction changes, at this time Xc = Xa. NH4+ has a seasonal trend and H+ remains rather constant during the year. Need more research about the relation between NH4+ and plant development stage and soil N.
Kuang et al. (2008), JGR	Newly formed particles from nucleation events could be an important source of cloud condensation nuclei.
Kuebler (2000), AE	The effect of variations in meteorological conditions can mask the dependence of the long-term trends in O3 on precursors emissions. Detrending technique: A pollutant's "detrended" time series consists of a time series independent of seasonal, weekly and short-term meteorologically induced patterns
Kuhns (2003), AE	SCAPE2 assumes that the aerosol phase is monodisperse and internally mixed. In most portions of the western US, ammonia gas i present in amounts that are in excess of what is needed to neutralize available nitrate.
Kulmala (2003), Science	Nucleation and initial growth may be decoupled under most condition. H2SO4 is involved in nucleation while organic vapor is involved in the initial growth stage
Kulmala (2004), JGR	new particle formation is a two-stage process: 1, homogeneous nucleation of thermodyanmically stable clusters. 2, condensation and coagulation growth of the clusters into quasi-stable aerosol particles (>3nm). If such growth is not sufficient fast, the clusters coagulate with and removed by preexisting aerosol population and no new particle formation. H2SO4 is generally not abundant enough to drive growth of cluster into size > 2 nm. Organic vaper plays the role through nano-Kohler theory.
Kulmala (2004), ACP	The formation and growth of atmospheric aerosols depend on several steps: nucleation, intial steps of growth and subsequent mainly condensational-growth. This paper focus on the initial steps of aerosol growth. Seems the initial growth rate inclrease as a function of size, which is consistent with a recently suggested nano-Kohler mechanism, in which growth is activated at a certain size with respect to condensation of organic vapors. nucleation and the subsequent growth of fresh nuclei are likely to be uncoupled phenomena via different participating vapors.
Kulmala (2006), ACP	derived the cluster activation theory to explain nucleation event. Ion cluster, neutral cluster could all be activated. quite offen, organic vapors are needed for activation.
Kulmala (2007), Science	Neutral nucleation dominates over ion-induced mechasnim, at least in boreal forest conditions. Nucleation meachanism initiating aerosol formation is likely to vary with location and atmospheric conditions.
Kulmala (2008), Tellus	new particles can increase CCN by a factor of more than two over the course of 1 day. so we must understand new-particle formation better to better predict climate
Kuo, Y.-H. (1997), DAO	in OSSE, nature run need to successfully capture the essential dynamics of the atmosphere. In this work, nature run is compared with previous analysis and the nature run is shown to capture the cyclone intensity and position correctly.
Kuo, Y.-H. (2005), GRL	RO soundings are most accurate between the altitudes of 5-25 km. RO soundings are of sufficiently high accuracy to differentiate performance of various types of radiosonde.
Kusaka (2001), BLM	The diurnal range of the roof temperature was larger than that of the road as we expect physically: the entire thermal inertial of the road is greater than that of the roof as a result of its greater thickness, and the radiative cooling of the road is less than that of the roof due to its smaller sky view factor.
Kusaka (2004), JAM	Nocturnal heat island is caused by anthropogenic heating and the following two effects of urban canyon structure: (i) a larger heat capacity due to the walls and (ii) a smaller sky-view factor. Sensitivity experiments show that the anthropogenic heating increases the surface air temperature by 0.67ЊC at 0500 LST. The larger heat capacity due to the walls decreases the daytime temperature and increases the nocturnal temperature by 0.39ЊC at 0600 LST.) The smaller sky-view factor increases the temperature by 0.52ЊC at midnight. The impact of albedo is not as large as the other factors, and its value is less than 0.15ЊC. Anthro- pogenic heating increases the temperature through the day. However, the heating has a greater effect on the nocturnal temperature.
Kusaka (2004),JMSJ	The difference between slab urban model and 1layer UCM are that the single-layer urban canopy model has a larger volumetric heat capacity and a smaller sky view factor due to the existence of a vertical urban surface . Another difference between the two models must be the ratio of the effective aerodynamic resistances to the transfer of heat. The incorporation of the single-layer urban canopy model tends to reduce the diurnal vari- ation of the surface air temperature and to de- lay the phase.
Laaksonen (1998), JAS	The traditional kohler theory only consider the equilibrium growth of aerosol containing only hightly soluble species. But other condensable gases and slightly soluble substances may affect the cloud droplet activation. Kohler theory is reformulated to take into account such factors.
Lane (2007), AE	STN sites are located in more populated areas than IMPROVE sites. The predicted EC concentration are about 3 Tims higher than the STN observed EC, indicating the EC emissions in cities may be overestimated. Non-road diesel dominates Midwest and the north eastern seaboard while its contribution is much smaller in Southeast. Current non-road diesel emission may be too high while the on-road diesel emission may be also high, but not that bad as non-road diesel emission. (2001 CONUS, EC in SEARCH is not overpredicted significantly)
Lane and Pandis (2007) EST	good literature survey for SOA mass yields (0-5%?) , PMCAMx is used to predict SOA from ISOP.
Langmann (2000)	Bott is still used as advection scheme in REMO. Only transport of chemical species is online, there is no aerosol, thus no direct and indirect effect. Online model transport more from lower troposphere to uper troposhpere because of high frequency dynamic motion is considered in online model. Good summary for the advantages of online model
Langner (1992), Nature	MOGUNTIA chemical transport model Only at most 6% of the anthropogenic sulphur emission is available for the formation of new aerosol particle, since half of it deposit, and most of remainder is oxidized in pre-existing particles. sulfate may reduced in south hemisphere because of OH reduction.
Lawrence (1999), JGR	Meteorological data are linearly interpolated between the neighboring time intervals to obtain the values for each time step. MATCH simulates three basic types of transport: advection, vertical diffusion, and moist convection(cumulus transport) provided by cumulus parameterization. Cloud fractions are diagnosed based on the RH. A cloud microphysics module provides formation and evaporation rates for cloud water and precipitation. Transported nontransported species
Lawson (2004), MWR	Two kinds of ensemble method: stochastic filters (e.g., EnKF) (which use random number realizations) and deterministic filters (which use explicit mathematical transformations. )
Lee, Eddie and Chak Chan (1999)	compositions of sea water (or sea salt).
Lee, S.-H. (2008), JGR	Observations indicate efficient new particle formation during the nighttime in the troposphere under low condensation sinks.
Lehning (1998), AE	Instantaneous vertical exchange of trace gases between the ABL and the FT is substantial for the total ABL afternoon budget.The vertical exchange dominates the budget when horizontal pollutant gradiaents across the box are absent or weak.
Lehrer (2004),ACP	Sea salt aerosol is a minor source of halogen sources compared to the sea ice surface. This paper explained why ODEs only occur in Spring
Lei (2004), JGR	CTN: HANK model. Two heterogeneous reactions are included in a hybrid chemical mechanism
Lei (2007), ACP	50% change of chemical BCs didn't show significant effects on O3 and other photochemically active species. No significant entrainments of O3 from above the PBL occur during morning hours.
Lelieveld et al. (2008), Nature	Photochemistry of VOCs over forest recycle OH in a manner that doesn't contribute to O3 formation, not through NO->NO2 Normally in golbal atmopsheric chemistry models, strong biogenic VOC emissions unrealistically deplete OH and isoprene consequently accumulates in the PBL. To circumvent this problem, atmospheric chemistry modellers typically reduce the strength of the isoprene source by a factor of two or more.
Lemone (2010), MWR	1km resolution simulation may have the "double counting" problem associated with partially resolving CBL-scale motions and using a PBL scheme The modeled convection becomes more cellular with smaller grid spacing
Lemonsu (2002), BLM	Some effects, such as the UHI intensity, are easily detectable by a meteorological surface network (provided that at least one heavily urbanized station and a few rural stations are present). However, other effects, such as the enhancement of convection, are less straightforward to document using a surface network. Some effects are difficult to measure, for example the urban area-averaged vertical wind velocity in an urban breeze. Although, these can be measured by aircraft, authorizations to fly above a city are not easy to obtain. For a weak and disorganized synoptic wind below 5 m s−1 , the urban breeze can take place freely in the afternoon, with the simulations showing vertical winds around 1 m s−1
Lemonsu (2009), BLM	The Canadian urban modelling system is run at 250m grid size. At such scale, individual building is not explicitely resolved, but is parameterized using the TEB (Town Energy balance) urban canopy model. a Canadian land data assimilation system has been used to provide initial soil water content and surface temperature. The impact of the city is maximum in the evening and its maximum vertical extent reaches 200m Nose of LLJ: its maximum wind speed
Leung (2003), JC	IC and lateral and lower boundary condistions of the simulation were derived from NCEP-NCAR reanalyses. Cloud amounts simulated by the convection and cloud microphysics schemes oversimulation undersimulation
Leung (2005), GRL	A regional climate model based on MM5 was used to downscale a global climate simulation generated by GISS model. Ventilation coefficient can be used to collectively represent mixing and transport within the BL
Leung (2006), BAMS	Downscaling is the process of deriving regional climate information based on large-scale climate conditions(including dynamic and statistical downscaleing) The statospheric physcis is lack in WRF
Levy (2005), JAS	fine-mode weighting is the fraction of the total AOD at 0.55um that was contributed by small particles.
Li, F.-Y. (2008), JGR	implemented a new online dust module (or dust emission scheme): Ginous et al. [2001].
Li, G.-H. (2008), JGR	A key step in the warm rain process is the autoconversion process whereby large cloud droplets collect small ones and become embryonic raindrops. The critical radius of dry aerosols is calculated from the kohler theory. the kohler theory is used for nucleation of aerosol to form could droplets.
Li, Jia (2003), JGR	Potassium salts and organic particles were the predominant species in the smoke. Excess fine potassium not attributable to mineral dust or sea-salt is an indicator of biomass burning. Potassium and chlorine are organically bound or present in the fluids of the vegetation. Most of the aerosol particles from biomass burning were in the submicrometer size range.
Li, Q (2002), JGR	The north American influence on Europe is driven by the long-range transport of ozone rather than its precursors, and includes comparable contributions from direct transatlantic transport of O3 in the boundary layer and subsidence from the free troposphere over Europe.
Li, Q (2004), JGR	The NOx emitted from fossil fuel combustion has a lifetime of less than a day against oxidation to HNO3 and PAN. We refer to the sum of NOx, HNO3, PAN, and other minor NOx oxidation products as total reactive nitrogen (NOy). view CO as an inert anthropogenic tracer
Li, Q (2005), JGR	Convection is particularly important for ventilating the sourtheastern US in summer, as the WCBs from the midlatitude cyclones generally do not extend that far south. Both MOPPIT and MODIS are aboard the Terra satellite in Sun-synchronous polar orbit with 1030 local time overpass. (another MODIS on board Aqua)
Li, Y.Q. (2006), JGR	The occurence of NH3 spikes was closely correlated with the CO2 spikes, a good marker of traffic exhaust. The NH3 emissions from the traffic exhaust could be a major source of the ambient NH3 in urban areas.
Liang, C. (1997), EST	Gas/particle partitioning of SOCs may not only involves simple physical adsorption, but also involves absorption partitioning when PM already contains some organic carbon.
Liang, X.-Z. (2001), JC	Minor lateral boundary condition errors, may quickly propogate into RCM domain and cause the model to produce unrealistic simulation. So LBC data assimilation technique/buffer zone treatment need to optimally integrate GCM signals into RCM
Liang (1998), JGR	export fluxes from continental boudary layer to global atmosphere. The boundary layer is defined vertically from the surface to the top of model layer3(about 740hPa)~2.6km
Lin, C-J (2005a), JEM	The draft of version 3 of NEI99 includes the emission data submitted by June 2002. MATLAB with netCDF and statistics toolboxes
Lin, C-J (2005b), JEM	CMAQ is formulated using a generalized coordinate system Start simulation assuming a clean atmosphere and allow the model to spin up for 48 hours with continuous emission input
Lin, C-J (2006), AE	Divalent mercury HG(II). Oxidation of GEM and speciation of mecury are very important since the deposition of different species are different. The dry deposition velocity of GEM is often neglected or assigned a very small value. Coarse resolution will reduce dry deposition of Hg since the dilution. current assignment of PHg into fine particles result in negligible PHg deposition
Lin, J-T (2010), AE	It is likely that dry deposition of O3 in summer over eastern US is underestimated by at least 30% on the basis of the Wesely (1989)
Lin, X (1994), JGR	Two ensembles of convective clouds are considered: deep convective cloud and shallow conective cloud. In the middle and upper troposphere, the redistribution is primarily caused by deep convective clouds while in the lower troposphere, both deep and shallow convective clouds will contribute.
Lindberg (2002), EST	Environmental mercury levels are known to be elevated in the Arctic, to generally increase with latitude. (wetland is the important source)
Lin, T-H (2001), AE	For each time step the emission intensity was computed by the deflation module using the surface winds, Temp, RH linearly interpolated from the MM5 output. Yellow sand transported over Taiwan are between 500-1500m
Liu, H. (2008), MWR	Analysis of temperature and water vapor with RO refractivity assimilated using the nonlocal operator is shown to be more accurate than assimilated using the local operater. This is verified against nearby withheld radiosonde observations. But when there are many other data also assimilated, such conclusion may not hold.
Liu, H.Y. (2003), JGR	WCB, deep convection, orographic lifting all play roles in transporting pollution to free troposphere.
Liu, S. C. (1996), JGR	Above 2km the side boundary condition affect significantly the distribution of trace species. O3, SO2, NOy IC and BC at top six altitude levels are set to propotional to PV. Model deficiency in vertical transport is the major source of uncertainty in our analysis of upper troposphere NOy
Liu, Y.B.(2006), JAMC	More vigorous nocturnal mixing over city due to urban forcing could reduce LLJ wind speeds First-order effects of urban forcing can be represented reasonably well with simple treatments of the land surface that do not include urban canopy models
Liu, Y. B. (2008), JAMC	Newton relaxation method is selected. 3Dvar can not take good advantage of high frequency data that are available from various sources. Station nudging, nudge toward observations rather than toward analysis of the data. Java/Web based graphyical user interface.
Liu, Y. B. (2008b), JAMC	The summer forecasts have higher average RMSEs, indicating possible contributions from winds that are not forecasted.
Lo (2008), JGR	From the experience of numerical weather prediction, the skill of limited area models diminished very rapidly with time, become useless for the simulation period beyond a week . The long term continuous simulation with one single initialization and frequent updates of LBCs is currently the most common approach in regional climate simulation.
Lohmann (1999), JGR	Emission of BC are assumed to be 80% hydrophobic and 20% hydrophylic. hydrophylic(soluble), hydrophobic(insoluble)
Lohmann (2000), JGR	Carbonaceous aerosols (organic and black carbon) The maxima of carbonaceous aerosol burden is not over USA because of fewer diesel emissions in USA and the extratropical biomass burning is not considered here. The mode radius of sulfate is much larger than that of black and organic carbon, thus the effect of sulfate in external approach is much smaller than that of carbonaceous aerosols
Lohmann (2005), ACP	Linking aerosol particles to cloud droplet is the weakest point in estimates of aerosol indirect effects. Adirect validation with measurements of aerosol radiative forcing is impossible. Satellite data may help provide some help.
Lorenc (1981), MWR	origin of "super observation"
Lorenc (2003), QJRMS	for a small area, forecast errors will be more influenced by BC than by analysis pertubations and error growth within the area. So ensemble of BC need to be created to reflect incoming pertubation.   A finite ensemble size limits the ability to fit detailed observation, so we need to get "superobs"?    model foorecasts are now as accurate as radiosonde observations.
Loridan (2010), QJRMS	Canyon roughness length and zero plane displacement height are now parameterized as a function of the Canyon geometry and removed from the list of inputs in the WRF/UCM model WRF/UCM shows a high sensitivity to parameters characterizing roof properties in contrast to a low response to road-related ones
Louie (2005), AE	In spring HK is mostly influenced by marine air mass, which is considered to be clean. South eastern China and western Taiwan are potential PM2.5 source region only in winter and fall.
Lu, Miao-Ling (2008), JGR	The life time of continental shallow cumulus: 30min up to 1 hr or so. Entrianment effects must be considered in subgrid cumulus parameterization.
Lu, Rong (1997), JGR	The SMOG modeling system represents a fully coupled mesoscale meteorological model and a gas, aerosol, transport, and radiation model (the GATOR code). While homogeneous nucleation is a source of new particles, heterogeneous nucleation represents a transformation from one type to another(mixed aerosol). Using these vapor pressures, the rates of condensational growth or evaporation at each particle size can be solved, and the overall mass transfer for the ensemble of particles can be determined. Means dynamic approach? The meteorological model solves the equations of fluid dynamics and thermodynamics..
Luecken (2006), AE	Even though also annual 2001 continental US simulation, the grid cells are slightly different from Eder and Yu(2006). BC of HAPs are set to zero because continental emissions are assumed to dominate trans-oceanic transport.
Lurmann (1997), AE	Neglecting gravitational setting, coagulation, chemical reaction and nucleation, while keeping condensation/evaporation term in aerosol general dynamic equation treatment for aerosol size distribution: bulk equilibrium+ allocate the mass to different size sections using a simplified mass transfer algorithm.(idea from Pandis et al. (1993))
Lynn (2009), JAMC	The most effective strategy to reduce the 2-m surface air temperature was to increase the albedo of the city surfaces (contrary to the results from Zhou and Shepherd, 2010). MM5/NOAH is modified: two roughness length are introduced, a global roughness and a local roughness effective surface temperature is proposed. The calculation of effective surface temperature indicates UHI's impact on city inhabitants during noon time is as large as or larger than the nocturnal heat island.
Lyons (1995), AE	Within sea breeze frontal zones, entire plumes can be vertically translocated aloft, where vertical motions may be of the same order of magnitude as the horizontal wind components. Within sea breeze retun flows recirculating plumes may bifurcate into distinct branches.
Ma, Y.(2003), JGR	Biomass burning tracers (K+, CO). EC associated with potassium has also been used as a tracer for biomass burning plumes. Incomplete combustion is a global source for CO. CH3Cl is a fairly long-lived gas, while K+ is water-soluble and readily scavenged by precipitation.
Macdonald (1998), AE	A new method to estimate surface roughness from the geometry of obstacle is provided. This method is used in the latest WRF model.
Madronich (2006), AE	The initial urban chemistry is well-understood, but not the intermidiate chemistry because of the much larger number of intermediate species produced by the photooxidation of the initial compounds. So a very complicated box model (~2000 species, ~5000 reactions) is tested in this paper.
Magliano (1999), AE	Secondary NH4NO3 is the largest contributor at all sites during winter, suggesting a regional influence. Secondary NH4NO3 is primary cause of diurnal variations observed at two rural sites.
Mahowald (1995), JGR	Cumulus parameterization: moist convective scheme. The convective shcemes vary in the closure used to derive the mass fluxes, as well as the cloud model formulation. under the unstable conditions, the counter gradient term in nonlocal PBL schemes is dependent on the surface flux of the consituent, vertical velocity scale and the PBL height
Mahowald (1997), JGR	The ability of the inverse method (defined as the combination of an optimal estimation scheme, the transport model, and the observational network) to produce useful estimates of emission is dependent on 3 factors: the accuracy of the transport in the CTM, the validity of the optimal estimation scheme, lastly the location, size, and accuracy of observing network.
Mahowald (1997), JGR	boundary layer mixing and moist convective mixing MATCH model simulation indicate that inclusion of an accurate parameterization for mixing due to moist convection is important to simulate accurately the vertical distribution of trace gases in the upper troposphere
Mahrt (2014), ARFM	In extremely stable conditions, the depth of the surface-based boundary layer, if definable, may be thinner than the height of the roughness elements inflection point instability
Makar	Neglecting chemical losses while making isoprene emission estimates was found to underestimate emission rates up to 40%. Is this contradicting with Forkel (2006)?
Mallet (2004), ACPD	Brief discreption of 3D chemistry transport model Polair. Polair has tangent linear mode and an adjoint mode, which could be used for sensitivity analyses and data assimilation.
Mallet (2006), JGR	ensemble modeling regarding different choice of physical parameterization, input data, etc. Seems there are high uncertainty associated with tubulence closure and chemical mechanism.
Mallet (2007), JGR	Different operating splitting method. Splitting sequence: advection, diffusion, and then chemistry according to the rule of putting the stiffest operator at the end. Choice of advection scheme is important. The uncertainties associated with numerical schemes may not be as big as those due to coarse input data or physical paremeterization.
Mallet (2007), ACP	Overview of structure of modeling system: Polyphemus, which even includes data assimilation and ensemble forecast.
Malm (1996), JGR	Light scattering by gases in atmos is described by Rayleigh scattering theory. The only gas that absorbs light is NO2. In most instances, particle scattering and absorption are primarily responsible for visibility reduction.
Malm (2002), JGR	While in most region sulfate concentration decrease during the past decade, it did increase in some other region e.g. west Texas. Highest concentrations along Ohio River Valley and central Tennessee.
Malm (2004), JGR	Fine nitrates tend to be highest around california and the Midwest. Nitric acid vapor has a deposition rate that is about 10 times greater than particulate nitrate. During spring in many areas of the western US, fine soil can contribute as much as 40% of fine mass.
Mao, HuiTing (2004a), JGR	Observation and model simulations suggested the nocturnal low-level jet and orgraphically induced flow during the daytime play important role in producing elevated daytime mixing ratio of O3.
Mao, HuiTing (2004b), JGR	NOY=NO+NO2+HNO3+PAN+RNO2+ aerosol NO3- High O3 in summer in NE occurred mainly in southerly wind quadrants, indicating substantial long range transport from upwind source regions
Marguilis (2004), BLM	Entrainment ratio of heat flux 0.4. days with higher winds generally have higher entrainment due to mechanically generated turbulence. The relationship between entrainment ratio and Bowen ratio, mixed-layer humidity and ABL height need further exploration.
Martilli (2002), EMS	In decoupled meteorological and chemical models, the transport error is not due to the time step of advection scheme, but to the updating frequency of input meteorology field.
Martin (2006), JGR	During summer 2004, extensive wildfires burned in Alaska and western canada, which contributed a significant source of BC, NOx, O3 in the central North Atlantic lower FT.
Massad (2008), EP	"Gap" between modeled and measured NH3 concentration partially due to emission from vegetation. The models underestimate by 30%. Empirical parameterization of gamma do not consider the effects of agricultural practice, plant developmental stage, plant nitrogen cycle on gamma.
Mathur (1992), AE	This paper demonstrate the importance of background concentration of key pollutant species on final deposition patten and quantities
Mathur (2003), JGR	Extended RADM for reduced nitrogen simulation. no distinction of aerosol size was retained since inorganic particles over the continental region are primarily in the fine aerosol size fraction. Unlike most gas phase pollutants, which are consistently deposited, NH3 is both emitted from and deposited to land and water surfaces: air-surface exchange of NH3
Mathur (2005), JGR	MAQSIP uses modal approach. MAQSIP is a prototype for CMAQ MM5 is run with FDDA and reinitialized every 5 days representation of process (e.g. chemical mechanism) in the free troposphere and its exchange with the boundary layer need improvement
Mathur (2008), JGR	Improving linkage between dynamic and chemical model calculation would reduce model error! solution: on-line model! CMAQ overpredicts during 2005 winter. The overprediction are pronounced at urban locations. Uncertainties in secondary NO3 formation may partially cause the nitrate overprediction. heterogeneous pathway may be overestimated.
Mathur (2008b), JGR	Assimilate remotely sensed AOD information into CMAQ. The 2004 fire caused enhancements up to 140 ppb in surface level daytime-average CO. PM2.5 is enhanced by 30%.
Mattsson (2002), PP	Plant NH3 emission is influenced by external N supply.
Mattsson (2008), BD	NH3 emission were frequent during a warm, dry period, while in a wet, cool period deposition dominate. Management practice affect ammonia emission potential.
Maykut (2003), EST	Mn, Fe are major components of disesel fuel additives. So we should see high concentration on week days since disesel vehicles are more on week days
Maxwell-Meier (2004), JGR	If accommodation coefficient of HNO3 is 0.1, HNO3 should be scavenged by fine-particle alkaline mineral dust very fast. This contrast to the observations may suggest that the initial accommodation coefficient for acidic gas (e.g., SO2, HNO3) reactions with mineral dust are orders of magnitude below 1. The formation of CaSO4 can occur by direct deposition of the acid (gaseous or particulate H2SO4) through CaCO3+H2SO4(a)->CaSO4+H2O+CO2
Mckeen, S. (2002), JGR	To keep the MM5 simulation close to reality. ECMWF is used for the grid nudging. Emissions of NOx and NMHC from the fires are introduced as molar ratios to CO. NOx/CO and NMHC/CO ratios are major uncertainty in the model. O3, CO, NOy are initialized propotional to potential velocity.
Mckeen, S. (2005), JGR	Ensemble of seven real time ozone forecasts Hourly output from WRF/Chem 36-hour forecast are started at 0000 and 1200 UT of each day using a horizontal grid spacing of 27 km
Mckeen, S. (2007) JGR	The equilibrium assumption requires that all particles have the same chemical composition for all species involved in gas-particle equilibrium. The cold and moist conditions of 27 Km WRF/Chem allow incorporation of ammonia and nitrate So that NH4 can exceed the ammonium sulfate ratio of 2. Night time sulfate in WRF/Chem is also very low due to the decoupling between the surface layer and layers aloft where sulfate is formed. All models overpredict total sulfur. The causes could be transport, deposition, rainout or meteorological biases STEM-2k3 got both lateral and top boundary conditions from global CTM WRF output is snapshot at the top of each hour. Only CMAQ gives average concentration.
Mckendry (2001), JGR	The mineral dust from tranpacific transport was incorporated into the planetary boundary layer as a result of strong subsidence. (meteorological mechanism by which mineral aerosol was transported to the PBL) If unusually high concentrations of PM10 in late April 1998 were associated with local anthropogenic sources (i.e., transportation sources), it might be expected that O3 and CO concentrations whould show increases proportionate to those of PM10. But apparently O3 and CO concentrations didn't show such increasing. This leaves either a local crustal source (e.g., spring ploughing) or an alternative nonlocal source of PM.
Mckendry (2007), JGR	Several baroclinic systems need to cooperate in order to generate trans-pacific transport events Dust is lofted into mid-troposphere near the source region in the low-pressure center then cross the Asia and pacific and then gradually subsided in a ridge along the west coast of North America
McNaughton (1986,1989),BLM, EAE	Slab Mixed-layer model; negative feedback between evaporation and height growth: higher H=> rapid growth=> more dry air entrained=> higher saturation deficit=>Higher E=>Lower H
McNaughton (1991), AFM	There is little point in using multi-layer canopy models for regional studies of the energy balance
Mebust (2003), JGR	Eventhough CMAQ overpredict SO2 by 100%, it did very good for sulfate. CMAQ did well on days with high O3 conc. but less well when O3 obs. were low. We could determine visibility degradation by Mie extinction or an empirical mass reconstruction approach.
Medina et al. (2007), JGR	CCN is calculated using measured aerosol size/composition distribution coupled with Kohler theory. The CCN closure study suggests "simple" Kohler theory can be used to predict CCN concentration. CCN clousre, i.e., comparison of predictions with observation of CCN concentrations.
Meehl (2004), Science	Parallel Climate Model predicted more intense, more frequent, and longer lasting heat waves in the 21st century. Heat waves are generally associated with specific atmospheric circulation patterns represented by semistationary 500hpa positive height anomalies.
Melamed (2008), JGR	measured SO2 power plant emission is consistant with reported SO2 emission.
Mellor and Yamada (1982), RG	The negative flux overshoot near the inversion height is underestimated by a factor of 3 and 4.  means not enough entrainment process!!!!!
Mena-Carrasco (2007), JGR	Anthropogenic emissions were taking from NEI for the base year of 1999 (NEI1999). NEI 2001 has lower emissions of CO, NOx and SO2 than NEI 1999. NEI 2001 NOx emissions maybe still high. Sea-salt emissions, dry deposition rate are among the significant uncertainty factors. WP-3 flow at lower altitude for 2004 NEAQS.
Meng (1995), AST	include carbonates in SCAPE2, pitzer method performs better than K-M and Bromley.
Meng (1996), AE	Inorganic substances account for 25-50% of the taotal aerosol mass. Equilibrium is hold for water, Treatment of mass transfer flux for solid particles, which is used in Aero0.F
Meng (1998), JGR	SCAPE2, "Most previous efforts in modeling condensation/evaporation of volatile species have assumed instantaneous equilbirium" means "Condensation/evaporation" = " Mass transfer"
Meng, Z.-Y. (2007), MWR	Mutischeme ensemble, EnKF shows different improvements for different cases because that error growth dynamics are scale and flow-dependent.  Model error may be caused by numerical inaccuracy ,truncation error, inadequate parameterization of subgrid physical process.
Meng, Z.-Y. (2008a), MWR	EnKF is shown to outperform 3DVAR, which is partially because the use of ensemble mean in EnKF, while single unperturbed run in the 3DVAR in RMSE verification.  Ensemble based 3DVAR outperform the standard 3DVAR. EnKF outperform 3DVAR through its flow dependent background error covariance but also its ensemble based static estimation.
Meng, Z.-Y.  (2008b), MWR	Best data to do data assimilation, rawinsonde data?  Since only marginal benefit might be gained through assimilation of additional data.  RMS fit of the verified field to rawinsonde observation is referred to as RMSE.  A large portion of improvement from EnKF appears to come from using an emsemble mean and may be due in part to the ensemble smoothing effect.
Mesinger (2006), BAMS	NARR does not assimilate surface (2-m) temperature over land. Neither does NNRP global reanalysis. but the surface temp and winds are closer to station observations than those of GR. The improvement of wind and temp over the whole troposphere is also demenstrated.
Metzger (2002), JGR	solve gas/aerosol equilibrium partitioning noniteratively through parameterization of activity coefficient to speed up the equilibrium model, which is warrent urgently by the global modeling Gas/aerosol partitioning only occurs for volatile and semivolatile aerosol compounds. For ammonium-sulfate-nitrate-water aerosol system it is only relevant to the ammonium nitrate salt. All other ammonium salts and nonneutralized sulfate are treated as nonvolatile. The equilibrium model assumptions(thermodynamic equilibrium; internal mixing) are valid for inorganic salt compounds if the aerosol modeling is limited to the ammonium-sulfate-nitrate-water system.
Metzger (2002),JGR2	During winter and at night, aerosol loading is higher than without accounting for gas/aerosol partitioning because HNO3, NH4 partition completely into aerosol phase to yield ammonium nitrate. Where limited ammonia is available, So2 reduction is not propotionally effective in reducing aerosol concentration becuase NH4NO3 wil replace (NH4)2SO4 to be in aerosol phase
Metzger (2006), ACP	Small particles have shorter equilibrium time-scales than larger particles due to a larger surface to volume ratio. importance of mineral cations and organics in gas particle partitioning, which are both considered in EQSAM2
Metzger (2007), ACP	In classical equilibrium thermodynamics, water is held constant, also the activity coefficient used is measured under laboratory condition. So the classical equilibrium thermodynamics is defined for laboratory condition, but applied to atmosperic modeling. In this paper, thermodynmaic is simplified by reformulating equilibrium to consistently include water. The equilibrium can be solved analytically and non-iteratiely. neutralization order: Anions: PO3− 4 >SO2− 4 >HSO−4 >NO−3 >Cl−>Br−>I−>CO2− 3 > Cations: Fe3+>Mg2+>Ca2+>Na+>K+>NH+4 >H+
Meyers (1998), JGR	Multilayer model, discretizing the vegetated canopy into 20 layers. Single-layer model (i.e., big-leaf model)
Meyers (2006), AFM	NH3 compensation points depend on many factors, such as leaf age, plant water stress and availability of nitrogen in the root zone. The complex source/sink distributions of ammonia within agriculture ecosystem preclude the application of the deposition velocity concept.
Miao, J.-F. (2007), JAMC	no systematic temperature bias for all experiments (NOAH, with MRF or MYJ, PX), while diurnal cycle intensity is underestimated.   PX is sensitive to soil initilization.   NOAH simulate urban effect well while PX doesn't.
Miao, J.-F. (2008), MAP	For clear day, ETA and MRF (both coupled with NOAH) consistently underpredict temp.   Model (MM5) performance is better during the nighttime than during the daytime for all schemes.      The model perform better for low pressure system.
Miao, S.-G. (2011), JAMC	Urban effect lead to the breaking of the squall line into convective cells over the urban area
Michalowski (2000), JGR	The removal of the BrO reaction with HBr has no impact on O3, while the BrO reactions with aldehydes significantly quickens the onset of O3 depletion
Mickley (2004), GRL	Mid-latitude cyclone frequency is expected to decrease in a warmer climate. So the severity and duration of summertime regional pollution episodes in the midewestern and northeastern US will increase.
Mignacca (1999), JAWMA	Retention of semi volatile species is a big concern for TEOM PM measurements.
Ming (2002), AICHe Journal	A new thermodynamic equilibrium model is developed. In this mode, the phase equilibrium is based on the PSC approach to electrolytes(inorganic ions) and UNIFAC approach for organic species.
Mircea et al. (2002), Tellus	very nice paper! Applied the modified Kohler theory of Shulman (1996) to 3 types of aerosol distribution with different composition. Total CCN at differnt supersaturation are calculated. The relationship between total CCN and supersaturation is derived. The classical power law doesn't apply for the rural and Urban aerosol type.
Mitchell (2002), MWR	A perturbation of streamfunction can be used to generate perturbations of u,v,T and p, which are quasigeostrophically consistent.    Realizations of the model error will be generated and added to the pair of ensembles of first-guess fields prior to the assimilation of observations.         There can be substantial imbalance in the analysis when localization is severe.   Imbalance decreases as the localization is relaxed.       For a given ensemble size, there is an optimal value of localization parameter.
Miyazaki (2003), JGR	Wet deposition of PAN is negligible since its insolubility The WCB and convective outflow(COF) can transport NOx over long distances in the form of PAN to remote regions Most HNO3 is removed by precipitation during transport in lower free troposphere in WCB and COF air mass The WCB is an ascending airstream ahead of a surface clod front, traveling into the middle to upper troposphere, which transports trace species from the boundary layer to the free troposphere
Moody (1996), JGR	Two peaks in the O3 profile Significant amount of natural O3 transfer to upper troposphere through exchange mechanisms occuring with tropopause folds. Natural O3 exchanged into the upper troposphere has the potential to reach lower altitudes as air parcles undergo subsidence.
Moody (1998), JGR	Back trajectory analysis shows the regional transport from urban corridor and midwestern contribute to pollution episode in NE.
Morris (2005), JAWMA	CMAQ assume that all of the secondary PM is in aitken and accumulation modes. O3 differences between predictions by different mass transfer approach are negligible. nitrate needs more research.
Morris, G.A. (2010), ACP	In the presense of storm, O3 loss may occur initially. NOx photochemistry begin to favor production of O3 at the dissipation stage
Morris, G.A. (2010), AE	O3 in the morning redidual layer resemble that in the free troposphere in Houston during TexAQS II
Moya (2001), AE	The bulk aerosol approach implies that all modeled particles have the same chemical composition. This assumption thus allows the mixing of acidic with alkaline aprticles which may introduce errors in predicting aerosol behavior SEQUILIB, SCAPE2 and ISORROPIA treat the equilibrium problem by solving a set of nonlinear algebraic equations (reaction equations) simultaneously with electroneutrality and mass conservation equations using Newton0Raphson and bisectional methods.
Moya (2002), AE	size resolved equilibrium model is not good for low RH, because metastable assumption is not valid. Bulk equilibrium approach assume all modeled particles have the same chemical composition
Moya (2003), AE	in dry season, accumulation mode is centered at 0.32 um, which is result from condensation of gas phase precusors. during rainy season droplet mode centered at 0.56um is probably result of aqueous-phase reactions
Moya (2004), AE	The ionic composition of an aerosol determines to a large extent its acidic/basic characteristics, influencing the partitioning of semi-volatile compounds between the gas-phase and different size particles. Correlations of NH4+ in the accumulation mode with the gas-phase precursors (NH3 and HNO3) presented an inverse relation suggesting the gas-phase to PM conversion.
Murphy and Allen (2005), AE	Daily emissions in Houston-Galveston area could vary significantly from annual average(The industry emission could be episodic). both observation and simulation show that the event emission could result in 50 ppb or even higher addtional O3 formation.
Myles (2007), ERL	The deposition of ammonia, nitric acid and particulate sulfate was controlled by aerodynamic and quasi-laminar layer resistances and that sulfur dioxide is relatively dependent upon stomatal conditions.
Myhre, Metzger (2006), ACP	Mass transfer is done in 2 steps: 1st fine mode is equilibriated with gas phase. 2nd, coarse mode is equilibriated with gas phase. So thermodynamic equilibrium is solved twice. So by doing so, two modes are seperated. Na2SO4 is not allowed to form, which would have been predicted if all species are treated together in one equilibrium calculation.
Nair et al. (2011),AE	Nair et al., 2011, AE also observed the anti-correlation between O3 in the marine boundary layer and PBLH. A useful information. water vapor plays role in destroy O3 in the marine boundary layer. O3 in the marine boundary layer remains low during nighttime partly due to insignificant nighttime intrusion from the free troposphere.
Nallapareddy et al. (2011), JAMC	The nocturnal warming events are due to the downward tranposrt of warmer air by the strong and gusty winds; wind shear induced strong turbulence? Warming evensts were mostly associated with stronger initial temperature inversions. "mixing out"="leaky" ?
Nam (2006), AE	additional emissions are added to the TCEQ prepared emissions. Only small percentage of event emission (0.5%) result in significant O3 formation. The timing, duration, composition, meteorology will determine the eventual O3 formation.
Neirynck (2005), AE	Simple resistance model is used Ra+Rb+Rc. The compensation point is modulated by leaf temperature but also depends on nitrogen status, plant development stage etc. In contrast with fertilized crops, semi-natural vegetation is generally regarded a perfect sink for NH3.
Neirynck (2008), EP	Net-emission occurance is widely accepted, but its origin(stomata, cuticle) remained speculative. Evidence was found that the emission potential from stomata might vary accross season. Sutton 1998 AE's models are tested. The static model (Single layer canopy compensation point-cuticular resistance model) didn't work out because Rw parameterizations were not robust enough.
Nemitz (2000), AFM	Dry deposition schemes to be used in operation CTM should be easy to calculate and based on as few variables as possible. Accuracy vs. simplicity and applicability. Live leaves were a net sink as leaf compensation points were smaller than air concentrations. While the 3-layer model provides the better mechanistic representation of the exchange process, 2-layer model provide a simplified generalized approach for application to CTM.
Nemitz (2001), QJRMS	For implementation in ATMs, simple resistance approaches are needed which is computationally efficient and their input parameters are available. Simple Rc approach is only applicable to (semi-) natural vegetation where only deposition occurs,( i.e., no emission of NH3). However, NH3 volatilization from vegetation and fertilizers contribute significantly to global total NH3 emission (23%). If exchange of NH3 is dominated by the foliage, then one-layer Xc model is adequate. But over spare canopy, it may not work. The two-layer Xc model treat bi-directional stomatal transport, cuticular deposition as well as emissions from at ground surface.
Nemitz (2004a), ACP	Given the much faster exchange rate of gaseous species if compared with their pariculate conterparts, the phase partitioning between gas and aersol can have a significant effect on local budgets and atmospheric transport distances. Even the deposition of normally alkaline ammonia gas and neutral ammonium aerosol can cause acidification depending on their fate within soil NH3 fluxes show mainly deposition, with some periods of significant daytime emission. NH3 exchange could be reproduced by toth Rc model (deposition fluxes only) and canopy compensation point model. Ammonia emission from plant foliage can be the result of either of a stomatal compensation point or desorption from leaf surface. Observations confirmed stomatal exchange dominates.
Nemitz (2004b), ACP	In general particles deposit more slowly than reactive gases, they often dominate long-range transport and transboundary fluxes of air pollutants. Fomation of ammonium sulfate is irreversible
Nemitz (2004c), ACP	Apparent emission of particles with a diameter <0.18 um and deposition of larger particles over heathlands and forests.
Nenes (1999), AE	ISORROPIA solves 'forward' and 'reverse' problem (serve for the dynamic approach).
Nenes (2001), Tellus	Three mechanisms may lead to kinetic limitations on cloud droplet formation: inertial mechanism, deactivation mechanism, evaporation mechanism. Under such condition, the equilibrium assumption (with ambient water vapor concentration) may not stand.
Nenes (2003), JGR	aerosol activation parameterization is developed. Kinetic limitations on droplet growth is introduced. First generation of aerosol activation approach is based on sulfate mass or total aerosol number. aerosol activation parameterizations still rely on some assumptions, which may not hold under some conditions. Large uncertainties about the effect of organic species on droplet activation.
Neu et al. (1994), BLM	O3 mixed down from RL contributes 50-70% to the maximum concentration near the surface on the following day. The rest come from chemical production and possibly advection. The relative importance of various processes to surface O3 is still unsolved problem
Neuman (2003), JGR	The vertical gradients in the observed HNO3 and particle mixing ratios, but not in other trace gases(co, o3) indicated the equilibrium time for ammonium nitrate aerosol formation was faster than boundary layer mixing time. Air mixing in BL is around 10-20 min. Time scale for fine particulate ammonium nitrate formation is about minutes, which is quicker than other time dependent changes in the atmosphere.
Neuman (2006), JGR	Over the Gulf of Maine, a strong temperature inversion isolated the air in the MBL from that above. Vertical mixing is suppressed by the highly stable stratification, and high concentrations of pollutants can be transported long distance. Transport of HNO3 is important for ocean plant life, which is sensitive to the location and magnitude of nitrate deposition to the ocean surface.
Nielsen-Gammon (2007), JGR	There are bay sea breeze on Aug. 30, 31, 2000 TX.
Nielsen-Gammon (2008), JAMC	Seperate instruments are found to measure PBL heights with gret variability under some conditions (e.g., no strong stable PBL top present, clouds present near the top of boundary layer).
Niyogi et al. (2011), JAMC	Liu et al. (2006) slab model is used to simulate the impact of Indianapolis. Storms split closer to the upwind urban region and merge again downwind
Noh (2003), BLM	Added explicit representation of entrainment rate to TM1986, which turns out to be critical for improvement.        Thinking: no entrainment in MYJ?
Nolte (2008), AE	dynamic approach correctly predict that nitrate is predominantly in the coarse mode in Tampa, a coastal site. CMAQ assume the equilibrium between gas phase and fine PM, while interaction between gas phase and coarse is not simulated. In other words, coarse particles, including sea salt, are assumed to be inert.
Nolte (2008), JGR	With SAPRC mechanism CMAQ significantly overpredict Max 8hr average O3 (by 10-15ppb) (Warm temperature bias may partially contribute to this overprediction) Extended O3 season and higher O3 in future USA is predicted despite decrease in O3 boundary condition(backgroud O3). Limitation: absence of feedback from atmospheric pollutants on the simulated climate.
Nopmongcol (2006), AE	Wood-smoke mediated sulfate formation: A fraction of impingement of SO2 on wood smoke particles resulted in sulfate formation. In addition to the sulfate formation, the reaction of SO2 with carbonaceous particles can convert hydrophobic particle to hydrophilic, which could further initiate the acid-catalyzed heterogeneous reaction. TexAQS2000 is significantly impacted by forest fires.
Nowak (2006), JGR	assumption of equilibrium is reasonable for July, Aug. 2002 Atlanta. The discrepancy of NH3 from CMAQ and measuremnet is likely due to missing NH3 source in the model, for example: inhomogeneous small scale soil NH3 emission.
Odman (1992), AE	Gear's method with automatic step and error control can provide solutions of high accuracy. LSODE is one of Gear solver. Gear method (Gear, 1971b; Lambert, 1973) is actually a family of methods of various orders. Both Deltat and the order are varied to satisfy the error criteria most efficiently. The methods are implicit multi step methods An error analysis is performed on the hybrid and QSSA solutions using the LSODE solution as a reference
Odman (2004), Fuel Processing Technology	The sensitivities are defined as local derivatives of concentrations with respect to emissions. As such, they are only first-order accurate approximations to how the model would respond to actual emission perturbations.
O'Dowd (1997), AE	Buffering capacity The paimary condensation sink for the marine aerosol is the sea-salt jet drop mode with nearly three times greater potential for condensation associated with this mode when compared to the accumulation mode distribution. So don't think accumulation mode always has the most condensed mass.
O'Dowd (2000), JGR	determine the amount of water associated with aerosol using Kohler equation. Idea solution, that is one which obeys Raoult's and Henry's laws.
O'Dowd (2001), AR	In marine air masses, surface area is generally dominated by super-micron particles. in polluted air, surface area is dominated by sub-micron sizes.
O'Dowd (2002), Nature	Under marine condition, the new particles can form from condensable iodine-containing vapours, thus the marine iodocarbon emissions have a potentially significant effect on global radiative forcing
O'Dowd (2008), GRL	During winter sea-salt dominates sea-spray mass while in summer OC contributes more than 60% submicron sea-spray mass.
Odum (1996), EST	70% of organic aerosol comes from secondary origin under peak photochemical conditions. Yearly average this percentage is 20-30%. For organic species, even products whose gas phase concentrations are below their saturation concentration will partition a portion of their mass into the condensed organic phase.
Ogoli (2003), Energy and Buildings	Materials with high thermal mass have long time lag and moderating effects to temperature swings
Olszyna (1994), JGR	An air mass was defined as being photochemically aged when NOx/NOy ratio was less than 0.4. Here aerosol NO3- is included in NOy
Oshima (2004), JGR	Two type of uplifting process: cumulus convection, WCB. Outflow of Asian Pollution was caused by limited numbers of episodes rather than persistent circulation, depending on the occurances, location and times of vertical transport processes. Such vertical transport processes vary from year to year. enhanced CO at altitudes between 5-10km
Otte (2005), WF	Currently Eta model is the North American Mesoscale (NAM) model. WRF is expected to substitute it. Eta-CMAQ only provides next-day forcast for O3, not PM yet. Aerosol and heterogeneous chemistry processes are omitted for now. Gas-phase and aqueous chemistry are included.
Otte (2008a), JAMC	Impact of MM5 dynamic analysis (using nudging)/forecast on CMAQ simulation. No nudging of mass field at the surface or with in the PBL.
Otte (2008b), JAMC	CASTNET measurement in the western US are less representative of average conditions within a 36*36 grid cell. So those stations are omitted.
Ottley (1992), AE	Over the sea, the flux of ammonia could be either upward or downward. Unkown direction of ammonia flux would affect the nitrogen deposition input to the sea. Average concentration of NO3- 8.57ug m3-, nss-SO42- 6.62 ug m3-(source from Europe)
Owen (2006), JGR	Emissions rapidly transported to the upper troposphere, where many loss processes are less effective and wind speeds are high, are more likely to have detectable large-scale impacts, but may be less likely to directly affect downwind surface concentrations. Transport at lower altitude can also occur, but the time for dilution and photochemical tansformation is greater. So enhancements in mixing ratios well downwind are expected to be smaller. Uplift mechanism: warm converyor belt, convection.
Pagowski (2005), GRL	In summer 2004, only few observation exceed 85ppb for max 8-hr average concentration. No exceedance of 125 ppb 1hr average conc. is recorded. So it is not good to evaluate the model in terms of categorical statistics such as false alarm rate etc.
Pai (1999), JGR	zero dry deposition for Hg(0). The largest uncertainty in deposition velocity is surface resistance The aquesous phase chemistry is activated only if the grid cell is occupied by clouds. Hg(0)+O3 is dominantly important in gas phase chemistry
Palecki (2001), BAMS	It is quite possible that some future increases in heat wave frequency, intensity, and duration may occur due to natural climate variations, or due to a broader global climate change.
Palmer (2003), JGR	Top-down constrains from atmospheric concentration measurements, intepreted with CTM, can be used to improve the bottom-up estimates through an optimal estimation methodology (inverse model). The fundamental idea of an optimal estimation inverse method is to minimize a cost function. Subgrid variability of observations (representation error).
Pandis (1993), AE	assume saturation vapor pressure for organic speices. size distribution of primary aerosol and accomadation coef. of condensable species are most important to pridicted SOA size distribution.
Pandis (1995), J.Phys.Chem.	The stratospheric aerosol is an aqueous sulfuric aicd solution. Because of their long lifetime, stratospheric aerosols are much more spatially and chemically homogeneous than aerosols in the troposphere. Particles can be activated in the presence of water supersaturation to become fog and cloud droplets, these particles are termed CCN
Pankow (1994), AE	importance of absorption and adsorption played in phase partitioning
Pankow (2003), AE	Multiple liquid and solid phases in aerosol PM.
Park, R. J. (2004), JGR	Export and trans-pacific transport of Asian aerosol pollution is expected to be far less efficient than O3 because the lifting of Asian air to the free troposphere involves wet process that scavenge aerosols with high efficiency. Anthoropogenic sulfur is emitted as SO2 except for a small fraction as sulfate(5% in Europe and 3% elsewhere) The cloud volume fraction in a given grid box is specified as a empirical function of RH Wet deposition accounts for 80-90% of total deposition. desciption of IMPROVE CASTNET and NADP Sulfate transported to US relatively more efficiently than other species since SO2 partly escape scavenging during wet lifting processes. Scavenging of SO2 is limited by H2O2 availability transpacific transport of sulfate cause a depression of nitrate since their competing for NH3
Park, S.-K. (2006), AST	Fine particle number and surface area concentration are more directly related to disease. Ion-mediated nucleation mechanism can produce new particles in very low H2SO4 concentration that are unfavorable for binary homogenous nucleation. From CMAQ, number conc. in Aitken mode is underpredicted by a factor of 10-1000. One possible cause is that putting little emission in Aitken mode. or say Assumed size distribution of emission is not accurate.
Pathak (2003), AE	When HongKong was under the influence marine air masses, the measured concentrations of sulfate, ammonium and nitrate showed a strong correlation with the atmospheric mixing height. In total, about 40% of the sulfate and ammonium of PM2.5 in HongKong measured in this study was from continental air masses
Pathak (2004), EST	A notable example of an interparticle interaction is the mixing of ammonium sulfate and sodium chloride particles, which can account for some of the chloride depletion in sea-salt aerosols in field observations
Pathak (2004), AE	Strong acidity [H+]=2[SO42-]-[NH4+] Most atrong acidity existed as bisulfate ions in the aerosol in HK Free acid concentration and PH can represent the acidic nature of aerosols more precisely
Pavlovic (2006), AE	Secondary PM account for 50-60% of fine PM mass in eastern Texas. no major fires were underway between 22 Aug. -31 Aug. CAMx is applied.
Peltier (2007), ACP	OM is relatively more spatial uniform, and the fraction of OM to PM was largely influenced by source of sulfate, which is consistent with Quinn (2006). Ohio River valley has many power generation facilities. nitrate is rarely present above limit of detection and was found to associated with biomass burning plume. In PM1.0 surfate account for 35% and OM 55%
Peng, Li (2006), Chemosphere	Globally, the largest source of CO in troposphere is photochemical production (the oxidations of methan and non-methane hydrocarbons). CO concentrations are high in winter and spring.
Penner (1998), Climate Dynamics	most sulfate is formed within cloud droplets and does not cause an increase in CCN, whereas carbonaceous aerosols may add CCN proportional to their emissions
Penner (1998), JGR	In the lower stratosphere, time scales for transport are much longer than upper troposphere. So chemistry there approaches steady state between NOx and the other forms of NOy. IMPACT use meteorological field from the Goddard Data Assimilation office (DAO) statospheric model models may place aircrafe emissions within different regions(lower statosphere or upper troposphere),which will make huge difference for the concentration profiles
Penner (2004), Nature	Aerosol indirect effect is simulated using a could parcel model + radiative transfer model based on observed aerosol properties. The indirect effect is verified by the observed radiation properties.
Perez (2006), AE	Inputs to the neural network model are concentration on the present day plus measured and forecast values of meteorolgical variables(temp., wind etal.). Output are expected maximam concentration for the following day. A good choice of input variables is very important to neural network model
Peters (1995), AE	First and second generation models are off-line. Third generation model should be on-line. Gas-phase machenism is quite advanced relative to other component in models (e.g., cloud microphysics and precipitation scavenging) Cloud formation occurs by heteorogeneous nucleation of water vapor on aerosol particles at super saturation less than 1% Convective clouds, which are sub grid phenomena, can rapidly transport species from boundary layer to free troposphere. The simplest approach is to enhance the vertical eddy diffusivities in area of convective activity
Peters-Lidard (1998), JAS	Soil thermal conductivity exhibits a significant impact on the partitioning of surface energy fluxes and the prediction of soil and skin temperatures.
Pfister (2005), GRL	CO is major sink for OH. Two different approaches for estimation of emission: bottom-up and top-down estimates(i.e. inverse modeling). The inverse technique includes: a priori estimate of emission, a forward model, measurements, an assimilation scheme, an inverse modeling approach.
Pham (2001), MWR	The existence of an attractor: the system state will approach a special subset of the phase space regardless of the initial state.    The sequential data assimilation is also known as filtering.       The EnKF has "linearity aspect", in that the ensemble members are "corrected" by the addition of a linear function of the observations.       The Monte Carlo technique has a notoriously slow convergence rate.    Filter stability problem:  the filter work well for a while and suddenly commit large errors and then can lose track of true system state almost completely.
Phillips (2004), AE	Upward flux of NH3 could occur each season, mostly during mid-day (11am - 3pm), which is also reported in (Anderson, 1999).
Phillips (2006), AE	Comparisons between modeled and interpolated spatial patterns for 2001 annual continental US domain simulation. Ammonium has maximum in summer?
Pierce (2006), JGR	Sub-micron sea-salt particles are much more efficient at contributing to CCN than super-micron sea-salt. Convective clouds are assumed to have a supersaturation of 1% and large-scale coulds are assumed to have a supersaturation of 0.2% clear explain of "factor"
Pierce (2007), JGR	20-30% NOx emitted from US is exported to the global atmosphere as NOy
Pierce (2007), ACP	The uncertainty of CCN generated per unit mass of primary particles may exceed an order of magnitude because of the uncertainty of diameter of particles. For soluble particles, the activation diameter at 0.2% supersaturation (representative of stratiform clouds) is around 100 nm. Ultrafine sea salt may contribute significantly to CCN
Pilinis (1987), AE	Composition in Los Angeles: SO4(20%), NO3(25%), EC (10%), OC (20%), Soil, metals and water (25%)
Pilinis (1987), AE 21(4)	An urban multicomponent aerosol model. Assume that aerosol sulfate and organic levels are controlled by gas-to-particle conversion(condensation and nucleation), while particulate amonium and nitrate concentrations are governed by thermodynamic equilibrium Size and composition distribution are predicted
Pilinis (1987), AE 21(11)	SEQUILIB. Stands for Sectional Equilibrium model? Size distribution is alse handled. SIZE-resolved equilibrium(Debry, 2005). Thermodynamic models underpredict the nitrate concentration(including Russell(1988)'s model). no NaCl is one of the reasons. 0.1 mole H2SO4 in file mode, 0.2 model NaCl in coarse mode, what will bulk equilibrium model predict? pure Na2SO4
Pilinis (1999), Global Nest	The mechanism for developing the accumulation mode is still unclear, the prevailing hypothesis is that this section is the result of cloud formation and evaporation cycles. SCAPE, SEQUILIB implement a domain-oriented solution algorithm so thermodynamic equilibrium is achieved between all the particles and the gas phase.
Pilinis (2000), AST	New expression for mass transfer flux for solid particles are developped, which is more advanced than that of Meng (1996)
Pinder (2004), AE	Manure application is largest contributor to ammonia emission in fall and spring, while housing in winter. To reduce ammonia emission, we may need to increase in cow nitrogen efficiency that need to reduced manure nitrogen and decrease in application emissions
Pinder (2007), EST	Continued reduction in NOx and SO2 emissions are likely cost more than the past and less effective at reducing PM2.5. Ammoniua controls may be more effective, especially in winter.
Pinto (1997), JGR	The Dudhia SW radiation scheme must be used with cauthion over highly reflective surface particularly for clouding situation in which multiple reflections between the surface and the cloud layer occur.     Also absorption by O3 etc is not modeled.
Piot and von Glassow (2008), ACP	Cloud may play several roles for ODEs: 1. lead to bromide loss due to deposition, uptake of bormine and other species; 2. enhance vertical mixing uptake of HCHO to cloud reduce production of HOBr
Piringer (1999)MAP	The lifting of the jet across the city is probably due to the enhanced roughness over the city
Pirjola (2003), JGR	Monodisperse aerosol dynamics module: all particle in a size mode have the same radius and the same chemical composition. In order to calculate the other properties of aerosol than msss concentration(e.g., number conc. and size distribution), aerosol dynamic processes have to be described
Pirjola (2004), GRL	A faster growth is required for newly nucleated particle to survive the scavenging by pre-existing particles through coagulation. Nucleation is important to be included in regional/global models.
Pirrone (2000), JGR	Elemental mercury is relatively insoluble where as a number of Hg(II) compunds are very soluble. It is unlikely that adsorption to particulate matter will significantly change the gas phase concentration of elemental mercury. In the mass transfer section of GASPAR, therefor, the concentration of mercury in the gas phase is treated as a constant
Pieterse (2007), Tellus	Wu (2003)'s deposition model is incorperated. Dry deposition is also reversible for most gas species.
Pleim (2007a), JAMC	ACM2 reproduce the temperature gradient near suface better than the original ACM1 due to including local scale turbulent (or eddy) diffusion component. to the original nonlocal transport in ACM1.            For stable or neutral conditions, fconv is set to zero for pure eddy diffusion. This explain ACM2 is close to MYJ during night time.
Pleim (2007b), JAMC	wind profiler data do not provide best PBL evaluation because they are usually more influenced by model dynamics than PBL mixing.
Plessow (2005), AE	In March and April, the secondary ions contributed more than 90% to the PM2.5 fraction. The relative importance of N-species increased with the reduction of SO2 emissions
Poissant (2004), JGR	The vegetation might uptake RGM and release elemental gaseous mercury during daytime
Potukuchi(1995), AE	Strong acids such as sulfuric acid absorb water continuously at almost all relative humidities and so these particles grow continuously and smoothly with increasing relative humidity. The presence of solid phases can be observed only in weakly acidic aerosols at low relative humidities, whereas particles whose main components are strong scids will remain in the aqueous phase under most conditions
Potukuchi (1997) , AE	predicting surface vapor pressure from the aqueous-phase composition using neural networks
Pouliot (2005), AMS proceding	significantly reduction of large point source of NOx from 2001 to 2004.
Poulos et al. (2002), BAMS	The flux generated from one intermittent turbulence event can significantly affect the vertical structure of NBL despite its short lived existence. Such fluxes appear to be significant even when compared to the strong fluxes during convective daytime conditions.
Prakash Bhave, Kleeman (2002), EST	Source-oriented model prediction are evaluated by comparison with aerosol measurements of single-particle size and composition .
Pryor (2000), JAM	whether the transfer of HNO3 to particle phase to form sodium nitrate increase or decrease the deposition velocity of nitrogen, it depends on wind speed.
Pryor (2008), STE	data from Pacific 2001 shows that NO3- is largely associated with sea-salt, indicating chlorine depletion process. Partitioning of total NO3- between sea-spray and NH4NO3 is highly sensitive to the amount of sea spary present.
Pszenny (2004), ACP	Turnover rate? Increasing sea-salt mixing ratio by 25% led to a decrease in O3 of 16% Aerosol PH as a function of particle size was inferred from phase partitioning and thermodynamic properties of HCl. Kinetic models suggest that most sea-salt alkalinity in most remote marine regions is rapidly titrated and near equilibrium pH is established within seconds to tens of minutes after aerosol formation. Corresponding lifetimes for most sea-slat aerosols against dry deposition are many hours to a few days.
Pudykiewicz (1997), Tellus	Improve O3 simulation by using highly accurate nonoscillatory semi-Lagrangian advection transport algorithms Subgrid scale processes are important, which is reflected by the varibility of the model results is smaller than the variability of experiment data.
Pul (2004), AE	Model underestimate the measured NH3 concentration by ~30%, which is probably due to large uncertainty in dry deposition of NH3 (using a scheme of resistance)
Pun (1999), AE	A conceptual model is a qualitative compilation of the physical and chemical processes that govern the formation of PM, which, to the extent possible, is supported by quantitative information.
Pun (2001), EST	PM formation in SJV is HNO3 sensitive, HNO3 formation is oxidant sensitive, and oxidant formation is sensitive to reduction in VOC emissions.
Pun (2008), EST	Satisfactory model performance does not guarantee that the models can correctly predict the effects of changes in pollutant emissions on PM concentrations because the ability of a mathematical model to calculate a state variable does not necessarily imply the ability to predict its derivatives with respect to input variables. A dynamic evaluation is needed to provide information on the ability of the model to reproduce the response of concentrations to changes in emissions. Dynamic evaluations are typically not feasible because of a lack of observations associated with the emission sensitivity scenario. The response is subject to less uncertainty than the base or sensitivity simulations, because biases in both the base and sensitivity scenarios will cancel to some extent. Although good agreement among the models is not necessarily a guarantee that all models would predict the response correctly, large discrepancies among the models will necessarily imply incorrect responses predicted by at least some of the models.
Qian, Yun (1999), JGR	Aerosol effects are largest at regional scheme since highly localized nature of emissions and the relatively short sulfate aerosol lifetime. In Holtslag's PBL scheme, the PBL height is based on a bulk Richardson number. Vertical diffusion and vertical transport by cumulus clouds are treated seperately Sulfates are not directly emitted by anthropogenic activities, but are products of SO2 oxidation
Qian, Yun (2003), Tellus	The aerosol concentration is simulated by the MIRAGE model and used as input to the MM5(which has been adapted as a regional climate model) MIRAGE is composed of two fully coupled models: Global chemistry model(GChM) and CCM2 One of the most important factors contributing to the radiative properties of aerosol is water uptake
Qin, Y. (1997), Science of Total Environment	The spatial variations of SO42-, NH4+ and NO3- were even, suggesting that long distance transport, rather than local emission sources are dominant. The spatial variations of the concentrations of crustal elements show the impact of road dust, construction and reclamation works on atmospheric aerosols
Quinn (1998), JGR	The majority of the sea-salt aerosol mass is found in the coarse mode but has been reported on occasion to dominate the mass concentration of the accumulation mode size range as well. Low gas phase NH3 concentrations resulted in only partially neutralized nss SO4= aeroosl. The predicted equilibrium concentration of NH3(g) by AeRho agree well with the 3 to 30 pptv values measured. So NH3 appears to have been in equilibrium between these two atmospheric phases.
Quinn (2005), JGR	Sea salt dominates the sub- and supermicron aerosol mass in remote marine atmosphere. So we may re-consider source function of sea-salt. High POM mass is found in NE US plume! useful for NEAQS2004. POM=1.6*OC NO3- is almost entirely associated with coarse mode sea salt.
Quinn (2006), JGR	Close to the source region, the aerosol was neutral with an equivalence ratio near one, and POM mass fraction is high. As distance from the source region increased, SO2 was converted to SO4=. With no additional NH3 to buffer the acid uptake. So aerosol become acidic and POM fraction decrease with distance.
Quinn (2007), Tellus	K+ tracer of biomass burning Mg2+, Ca2+ tracer of dust have maximum concentrations in winter and spring indicating long range transport from continent to Arctic. Arctic haze in winter/spring.
Quinn (2008), ACP	Arctic temperatures have increased twice the global average rate over the past 100 years (IPCC, 2007). The largest single source of methane is natural wetland. Wetland, permafrost methane emission in Arctic, sub-Arctic. Methane being globally well-mixed, contribtues to increase in background tropospheric ozone level.
Raatikainen (2005), ACP	In non-ideal solutions, activities (a) represent the effective concentrations which should be used instead of the real concentration (c). The relation between the effective and the real concentration is given by the activity coefficient (r): a=r*c
Raga (2000), AE	Both aerosol particles and ozone peaks close to the top of the mixed layer in Mexico City
Ramanathan (2005), PNAS	Atmospheric brown cloud (aerosol) radiative forcing cool the surface, stabilize the atmosphere, and reduce evaporation and rainfall. More cooling due to aerosol in Northern Hemisphere relative to the southern Hemisphere weakened Hadley cell.
Randall (1980), JAS	Cloud top radiative cooling tends to drive convection in the cloud layer. This largely explains why cloud layers are almost always turbulent
Rao (1999), MWR	For mesoscale modeling in the Cape Canaveral area, nested grids with horizontal spacings less than 1 km is necessary.
Rappengluck (2008), JGR	Regional transport, take a look at 850 mb field. Conc. above PBL mostly reflect regional transport. Ozone decrease above the inversion layer due to its "capping" effect, and wind speeds increased and wind direction change.
Rasch (1997), JGR	Short timescale features present in a GCM or the atmosphere are absent in the archived data and hence from the simulations in off-line CTMs A recent version of MATCH forms one of the components of MOZART The errors in off-line model can be made small when the sampling interval is order 6 hours or less
Reale (2007), GRL	NR realism can be investigated through statistical techniques by determine how the means and variability in the NR compare with reanalysis or operational analysis. In this work the NR realism is investigated using a phenomenological approach targeting one specific region and specific category of events.
Riedo (2002), EM	dynamic cuticle absoption-desorption model; 3-layer model are over complex. Nemitz's 2-layer Xc model is combined with a grassland ecosystem model. For the cuticular exchange pathway, both instantaneous resistance models (e.g., Nemitz, 2001) and dynamic models have been developed.
Rife (2004), MWR	Anomaly correlation (AC). Different models with different resolutions show similar performance in terms of convectional verification metrics. In some locations increasing resolution will have little positive impact where the diurnal power in spectrum is very small compared with the power in the superdiurnal band (synoptic scale) and in the subdiurnal band (poorly observed and therefore largely unpredictable small scales) In contrast to convectional measures, the AC illustrates that the high resolution model has more skill where the local diurnal forcing is greatest.
Rife (2005), MWR	A set of verification metrics is designed that is increasingly tolerant of timing errors for temporal changes of wind
Rizzo (2003), JAWMA	summarized a lot of literature showing that TEOM undeerestimate PM compared to reference filter based measurement.
Roberts (2002), JGR	Inorganic salts only accounts for 15% of wet-season aerosol in Amazon Basin, most of which is sulfate. Sulfate may originate from biogenic emissions of reduced sulfur compounds. Aerosol number distribution and surface tension effects are more important to CCN activity than the total amount of soluble content.
Rockel (2008,2009?), JGR	Interior nudging gives better results for large scales but at the expense of a reduced variability at smaller scales. BC forcing is also nuding. Dynamic downscaling is to add value to resolve the smaller scale features which have a greater dependence on the surface boundary.
Rosenfeld (1999), GRL	The coalescence efficiency of cloud droplet into raindrops is greatly reduced when the radius of the largest cloud droplets is smaller than about 25 um, which is equivalent to an effective radius threshold of 14 um. Smoke shut off warm rain-forming processes.
Rosenfeld (2008), Science	CCN decrease precipitation from shallow clouds and invigorate deep convective rain. Cloud droplet coalescence into raindrops (i.e., autoconversion) . Weekly cycle of rainfall driven by weekly cycle of aerosols
Roy (2007), AE	Use MODIS observed active fire data to reallocate (or redistribute) biomass burning emissions. MODIS instruments from polar satellites (Terra Aqua)
Roy (2007), JGR	CMAQ AOD captures regional scale distribution as well as day-to-day variability arising from mesoscale/synoptic activity.
Rozoff (2003), JAM	RAMS/TEB (Town Energy Budget) is applied over St. Louis Momentum drag over the city regulates the strength of the UHI
Rupakheti (2005), AST	High concentration of sulfate in the afternoon, evidence of their formation connected with the photochemical oxidation. The presence of higher concentration of nitrate was favored by condition of low light, low temp, high RH.
Russell (1988), EST	The apportionment of TN between the gas and aerosol is important because of the different depositional rates of aerosol and reactive gases. Bulk aerosol phase! Size distribution is not simulated, just as Russel (1983)AE
Russell (1993), EST	HNO3 deposition velocity 1.71cm s-1, NO2 0.28cm s-1. The dry flux of NOx-derived species is dominated by HNO3and NO2 . The average ratio of deposition velocity of aerosol nitrate to the deposition velocity of HNO3 is 0.182
Russell (2000), AE	Lagrangian models could have significant errors. Typical air quality models are off-line to date. only very few are online Plume models: after the plume is sufficiently diluted, the pollutants are then mixed into the appropriate grids. Accuracy tests indicate that the methods used to integrate the chemistry introduce little error as compared to the advection algorithms. Sensitivity analysis: brute force; DDM
Russell, KM(2003), JGR	Nitrogen input via wet deposition are well quantified but dry-deposition fluxes are uncertain. Phase change of HNO3(g) to NO3- in the presence of sea-salt result in modest increase in the overall deposition of total NO3
Russell, L. M. (1994), JGR	Generally CCN linearly depends on DMS flux under typical remote marine conditions. But this linearity does not hold for low DMS fluxes, under which conditionheterogeneous conversion of SO2 on sea salt particles will inhibite the creation of new particles. Most of aerosol particles in the MBL are in the submicron regime. Nucleation rate scaling factor and sulfuric acid accommadation coefficient are very uncertain.
Russell, L. M. (1996), JGR	In marine boundary layer, low concentration bimodal distributions in cleaner air masses and higher concentration single mode distriubtion in air masses with apparent recent continental influence.
Russell, Matthew (2004), AST	FRM has negative sampling artifacts, particularly for ammonium nitrate. TEOM estimates even lower mass than FRM filter-based mass. Annual mean FRM mass in Southeast TX is close to annual NAAQS of 15 ug/m3 and tend to be higher near the urban and industrial core of Houston. PM2.5 mass concentration and compositions are generally spatially homogeneous throughout southeast Texas. Two-peak PM2.5 pattern around Houston.
Russell, Matthew (2005), JGR	SOA formation in southeast TX is dominated by the reaction of alfa-pinene with O3. In this model simulation, ambient measurements of total particle phase OC is used as a basis to estimate the partitioning of semi-volatile compounds to the particle phase.
Russell, M (2005), JGR	Organic compounds comprise 30-40% on average of PM2.5. During the Ozone season, secondary organic carbon was predicted to be roughly equivalent to primary organic carbon.
Russell, P (1999), JGR	Increase of Carbonaceous fraction with increasing height.
Rodriguez (2004), JGR	The major purpose of the thermodynamical module is the computation of Cs,i
Ryan (1998), JAWMA	The most extreme O3 events in the Baltimore-Washington metropolitan area occur in multi-day high O3 episodes. Significant regional-scale transport of O3 and its precursors occur during extreme O3 events of the standard type in the B-W area.
Ryerson (2003), JGR	Substantial and rapid O3 production in a single day in Houston differ from other US cities where the highest O3 mixing ratio result from slower accumulation of O3 over period of several days. Alkenes(C2H4, C3H6 dominate OH reactivity, contribute most to O3 formation, other VOC's contribution is small). Houston area is VOC-limited. Alkenes emission is substantially underestimated.
Sacher (2008), MWR	The approach of filtering: all of the information is used up to and including the considered analysis time.     Monto Carlo filter has two kinds: stochastic (require perturbed obs)  and deterministic (which does no require perturbed obs).      When spread is "too small" the filter believes it performs better than it does in reality. This behavior is called filter divergence.   The analysis spread among the ensemble members always underestimate the mean squred difference between the ensemble mean and the true state and the filter is therefor naturally subject to divergence.
Salamanca et al. (2011), JAMC	Four urban canopy schemes (i.e., BULK; a single-layer urban canopy model; BEP(building effect parameterization); BEP+BEM(building energy model)) are compared in WRF BULK scheme tends to overestimate T2 while other schemes underestimate it Anthropogenic heat was responsible for an increase in the air temperature of up to 2 oC in the densest urban areas
Salamanca et al. (2011), Int. J. Climatol.	used a horizontal diffusion coefficient of 300 m2s-1 to filter the roll-types structures for high-resolution (sub-km) WRF simulations
Salmond and McKendry (2002), AE	The increases in surface O3 during nighttime are linked to three reasons: changes in mesoscale wind flow regimes, LLJ, breaking of gravity waves. Such increases in noctunal surface O3 due to vertical mixing are unlikely to pose a significant health threat to local populations.
Salzen (1999), AE	In general, dissolved ionic species are favored and neutral species are of negligible concentration in secondary aerosol. Instead of explicit condensation equation for water vapor, the water content of aerosol is modelled from the equilibrium approach by use of ZSR. Accommodation coefficietn 0.11(HNO3), 0.01(HCl), 0.09(H2SO4), 0.097(NH3)
Sander (1996), JGR	Eventhough the aqueous phase(sea-salt aerosols) is highly concentrated solution, Henry law is used to calculate the equilibrium concentration. MOCCA is developped, which includes aqueous phase, gas phase and mass transfer between them and heterogeneous reactions.
Sandu (2003), Applied numerical math.	Growth processes include condensation, evaporation, deposition, and sublimation. Time splitting scheme introduce hard-to-quantify splitting errors
Sandu (2003), AE	The most popular and efficient techniques for sensitivity studies are the direct-decoupled and the adjoint sensitivity methods. Given a chemical mechanism, KPP generate the code for its forward integration, and the direct-decoupled/adjoint sensitivity analysis.
Sartelet (2006), AST	dynamic mass transfer approach is developed in a Modal Aerosol Model (MAM). Some of its treatments are borrowed from MADM(Pilinis et al., 2000), which is very like MADRID. Also a equilibrium version of MAM is developed, in which bulk equilibrium followed by redistribution is used, which is the same as that used in CMAQ4.3. MAM-eq's coarse nitrate and chloride prediction may differ by 40% from MAM. So we could see the importance of using dynamic approach!
Sartelet (2007), JGR	The impact of nucleation is negligible in Summer. Hybrid mass transfer approach is used, in which thermodynamic equilibrium is only assumed for the smallest mode (total 4 modes).
Sartelet (2008), AE	hybrid approach may predict 12% less of nitrate than that predicted by equilibrium approach.(My guess: hybrid predict nitrate in coarse mode, which has higher deposition rate?) Sulfate is not sensitive to mass transfer approach since it is non-volatile.
Sarwar (2007)	Chlorine emissions and chemistry enhanced O3 formation in Huston and New York area.
Sarwar (2007? 2008?)	CB05 performs better in high altitude condition and rural areas for O3.
Saunois et al. (2008), JGR	A 2-D model simulation shows that LNOx case increase O3 by 10-20 ppbv in the mid and upper troposphere while NOx from soil also increased O3 in the layer 9-12 km by 20-25 ppbv over west Africa
Sauvage et al. (2007), GRL	LiNOx may have contributed to O3 production in upper troposphere (9-12km?); Ozone minimum in upper troposphere is associated with tropical convection over Africa
Schaap (2004), ACP	ISORROPIA is used in LOTOS
Scheffe (1993), AE	The diffusion break corresponds to the top of a mixed layer. In general, two approaches are used to condense an explicit kinetic mechanism: the lumped approach and the carbon-bond approach. UAM computes the average concentrations as well as the instantaneous concentrations.
Schell (2001), JGR	coupling of MADE and SORGAM
Schichtel (2001), AE	On high-ozone days, the transport conditions in the central and south regions of eastern US are indicative of poor air mass dispersion The O3 exceedances in the central and southeastern US are predominately "Homegrown", Eastern seaboard is a reciepient of regional scale ozone transport.
Schubert (1993), BAMS	Assimilation system contains AGCM(atmopheric general circulation model) and optimal interpolation scheme. The penetrative convection originating in the boundary layer is parameterized using the relaxed Arakawa-Schubert scheme
Seaman (2000), JAM	mesoscale meteorological structure associated with high O3 episode in Urban corridor, New England area: Appalachian lee trough, etc. MM5 setting: parameterized convection for 4-km domain is turned off since the explicit microphysis could resolve the deep convection at this scale. Also the subgrid scale convection at mesh size below 10km is problematic because some assumption break down. No FDDA was applied below 850mb to allow surface forced feature to develop.
Seaman (2003), Environ. Inter.	The time dependency of the meteorologyical forcing can be a significant factor when rapidly evolving features are present, such as thunderstorms. Thus, direct coupling of the meteorological and air quality models into a sigle system on identical grids can eliminate time and space interpolations.
Seibert (2000), AE	The PBL height is chosen at the height where Rib reaches a critical value (e.g., 0.25). Wind profiler PBL height data is not reliable enough for operational use so far.
Seinfeld (2004), BAMS	Mineral particles are transported aloft through boundary layer convection and verticlal motions accociated with frontal boundaries. The depolarization ratio indicate the amount of non-spherical particles(e.g. dust) An elevated river of dust situated above a sulfate rich pollution plume located near surface the deposition of H2SO4 occurs on both accumulation and coarse mode in case of dust Strongly hygroscopic aerosol(e.g. NaCl)
Shao, Yaping (2002), JGR	Three dust emission mechanisms: Aerodynamic entrainment, saltation bombardment, Aggregates disintegration. Saltation bombardment is more efficient than aerodynamic entrainment.
Shen (2002), AE	SMPS-APS is reliable for continuous PM2.5 measurement and offer several advantage over other continuous and time-integrated monitors.
Sickles (2002), AE	Single-sampler weekly measurements from CASTNET may bias low for semi-volatile species comparing with composite day-night weekly sampler results, which is suggested to be more accurate, which is used before 1989. After 1989 single-sampler weekely sample is chose due to funding limitation.
Sievering (1999), JGR	NSS (SO4=-0.0603Na+) in nmolm-3. The production of NSS by the O3 oxidation of SO2 in SSAW(sea-salt aerosol water) is quenched once the buffering capacity is consumed. Thereafter H2O2 oxidation will continue to produce NSS.
Sihto et al. (2006), ACP	New particle formation rates correlate with sulpuric acid concentration to the power from 1 to 2. Power 1 corresponds to the activation theory. Power 2 corresponds to the kinetic type mechanism. There are large uncertainties associated with nucleation mechanism and growth rates of newly nucleated clusters.
Sillman et al. (1993), JGR	transport of pollutants over ocean where vertical mixing and deposition are suppressed contribute to O3 accumulation in northeastern U.S.
Simon et al. (2008), JAWMA	Sea salt account for 6-15% of total PM2.5 emission. In NEI fugitive dust emission may be overestimated. Uncertainties in primary emissions for PM2.5 are substantial.
Simpson (2007), AE	Observed PBL heights are derieved from profiler data using daily atmospheric sounding data. AERMET predicted PBL heights agree well with observation, which only use routine meteorological observations and an early morning atmospheric sounding to estimate evoluation of PBL height.
Singh (2006), JGR	The summer of 2004 was one of the largest fire seasons on record in NA because of persistent wild fires in the boreal forests of Alaska and Canada resulting from exceptiongally warm and dry conditions. So HCN and CH3CN, tracers of biomass combustion, were elevated everywhere in the troposphere.
Singles (1998), AE	Ammonia deposition should be considered together with its emission in air quality models. One process (e.g., emission) can't be treated twice.
Smedman (1995, AE)	Strong temperature inversion is critical for the formation of near-coastal LLJs in high latitudes
Smith (2000), AE	Most current generation of dry deposition models follow the "big-leaf" approach, in which the vegetation structure is conceptually treated as a single leaf representing the entire canopy and covering a large area.
Smith and Roebber (2011), JAMC	WRF/UCM with NLCD 2001 is applied to Chicago Limitation of green roof approach: moisture increase, thus the apparent temperature only has a minimal reduction
Snyder (2003), MWR	Splitting supercell storm.   EnKF could be used to improve small scale simulation, e.g., convective system.
Solomon (2000), AE	Specification of observationally based BC for air quality models is critical to their performance. Models should be evaluated not only with high O3 conditions, but also with a range of O3 concentration and meteorology conditions.
Solomon (2008), JAWMA	SMPS-APS size distribution data can be used to estimate PM2.5 mass concentration, but not ultrafine or coarse particle mass.
Song (1999), AE	SO2 and sulfuric acid deposit onto particles
Song, C. H. (2001), JAC	IDEA: take a look at correlation between nitrate and calcium The mineral dust contribution to total cation mass in the middle and upper troposphere is estimated to be greater than 50%, reflecting the long-range transport of dust in the free troposphere. Alkalinity increases with altitude, indicate greater capacity for HNO3 to partition into the particle phase in the middle and upper trposphere.
Song, J. (2008), AE	vertical mixing schemes affect surface concentrations significantly, resulting in differences of as much as 270% in modeled isoprene concentrations. Biogenic emissions of hydrocarbons, primarily isoprence, dominate the VOC emission inventory in eastern Texas.
Sorbjan (2009), BLM	entrainment flux/surface fluxes  = 0.2
Sorooshian (2006), AST	Some papers argued that equilibrium is valid for sub-micrometer particles. But some other papers said this assumption may not be valid under some condisitons.
Sorteberg (1996), AE	Emission of NH3 from stomata was found to be in the order of 0.1% of the total NH3 emission over Europe, but may give large emission contributions on certain days over crop land and where the average dry deposition velocities can be influenced. Crude assumptions are made for [NH4+]/[H+]
Spackman (2010),ACPD	BC mass loadings were positively correlated with O3 in ozone depletion events (ODEs) for all the observations in the ABL suggesting that BC was removed by dry deposition of BC on the snow or ice because molecular bromine, Br2 , which photolyzes and catalytically destroys O3 , is thought to be released near the open leads in regions of ice formation.
Spengler (1990), EST	Generally acid aerosol concentrations are higher during the day than at night. Acidity may be higher in rural than in urban areas because of higher NH3 emissions in urban areas that result in more neutralization.
Spindler (2001), QJRMS	The static 'canopy compensation point-cuticular resistance' Xc-Rw model is used. Sensitivity test shows Gamma = 1000 gives minimum model bias.
Spokes (2000), Tellus	The presense of sea salt aerosol result in a factor of 3 increase in the flux of nitrate to coastal waters through enhancement of both wet and dry deposition. non-volatile sodium nitrate.
Sportisse (2007), CG	Future direction of Air Quality model development: increase resolution, coupling of organic and inorganic thermodynamics. Two way feedback between meteorology and chemistry. Current models are overtuned. They are extensively tuned to meet acceptable model-to-data error statistics for few species at ground. It does not ensure that the results are satisfactory for 3D fields and many other trace species. An alternative way for plume-in-grid is using fine resolution. the order of splitting sequence. condensation/evaporation (i.e., mass transfer).
Spracklen (2005a), ACP	Sensitivity to the amounts of primary sulfate aerosol (0-5% of total sulfur) Change the order of operations or timestep change the prediction less than 5% UT and FT are the dominant source regions of new sulfuric acid particles due to the low temperature, which accelerate the nucleation rate
Spracklen (2005b) ACP	CN are defined as particles larger than 3 nm diameter. This size corresponds to the detection limit of current instrumentation
Spracklen (2006), ACP	Nucleation events greatly enhanced the number conc. in remote area, since there is low existing particle surface, which acts to scavenge condensable gases and clusters during growth.
Spracklen (2008), GRL	PBL particle formation is an important global source of CCN. But uncertainties in particle formation and growth rates must be reduced before accuracy of predictions can be improved.
Springston (2005), AE	Power plants are the largest contributor to the fine particle aerosol loading because of its SO2 emission.
Stanier (2004), AST	Water concentration varies from 0-20microg/m3 (0-50% of the hydrated particle mass)
Steeneveld (2006), BLM	Increase of stratification may lead to a collapse of turbulence so that the actual SBL decouple from surface in some cases (positive feedback).            Large-scale model tends to decouple and remain decoupled in an unphysical sense. This can result in a serious cold bias.   So some model apply enhanced mixing formulation to circumvent this problem.
Stein (2000), AE	HY-SPLIT-Chem development In Lagrangian model, only first order chemical reactions can be treated rigorously. For higher order reactions, a simple superposition of Lagrangian parcels is not strictly valid. Dry removal is assumed to be a first-order process.
Sterk (2013), JGR	For low wind regimes, the model sensitivity is larger for surface coupling and radiation, while for high wind speeds, the largest sensitivity is found for the turbulent mixing process.
Stevens (2000), MWR	An important effect of non-local flux is its ability to modulate the diagnosis of PBL height and thus the determination of the entrainment flux at the top of the PBL.        non-local flux term play important role in ventilating the surface layer.
Stier (2006), JC	ECHAM5-HAM: nitrate and treatment of semivolatile species are neglected. Microphysical aging refers to growth and modification of aerosols by coagulation, condensation, and cloud processing
Stockwell (2003),AE	Reductions in nitrogen oxide emission rates would be expected to increase aerosol concentrations since OH will increase after reducing NOx. The gas-phase reaction of N2O5 with water is negligible but it reacts with liquid water on aerosol particles, fog or cloud water droplets to produce nitric acid.
Strauss (2019), QJ	evaluate 3 PBL schemes in cloud, in terms of considering the countergradient flux, very nice paper! 2 methods to represent non-local turbulence: countergradient term, mass flux the MOENG approach could be agood candidate to generate clear air turbulence, which is a major issue for aviation forecasts
Streets (2003), JGR	Emission inverntory in Asia, however neither are primary PM10 and PM2.5 emissions reported The paper has apportioned annual emissions to daily and monthly emissions using a variety of techniques.
Streets (2007), AE	1st CMAQ application in China?
Strong et al. (2002), AE	The change of potential temperature closely followed the variation in O3 in the ODEs. The vertical transport mechanism replenish Arctic atmospheric boundary layer with O3 following the depletion episodes.
Strong (2004), AFM	When estimate active scalar flux, active chemistry need to be considered in addition to footprint transfer function and source strength distribution In fully leafed canopy, isoprene is significantly destroyed because of long residence time (comparing to its life time)
Stroud (2005), JGR	It is important to include escape efficiency estimates for sesquiterpenes within biogenic emission processing systems for regional air quality models; Since biogenic escape efficiencies were only moderately sensitive to anthropogenic pollution levels, a simple parameterization of escape efficiencies in regional air quality model may be possible.
Stutz (2002), GRL	Activation of chlorine and bromine is not observed in marine environments, while salt lakes release a large mount of reactive halogens. O3 is anticorrelated with halogen oxide level.
Sugawara (2001), JMSJ	The effective thermal property parameter for a town-scale urban area was found to be two to four times larger than that of the surface material component. The explanation for this, shown by several model simulations and other radiometric observations, was total surface area increased due to urban canyon structure. The thermal property of the urban surface has been set to be the same as its component material in many models. This study demonstrated that thermal inertia should be a few times larger than that of the component material in a bulk model that does not deal with the canyon shape explicitly.
Sun (1998a), AE	Only condensation remains significant for aerosol size/composition distribution . Nucleation, coagulation can be eliminated . Justification for the splitting of the condensation equation. Integration time step for condensation equation(growth term) must be small . Bott's scheme is used for solving shift term. acid equilibrium assumption, coupled condensation. Steps for implementing dynamic mass transfer
Sun (1998b), AE	Hydrogen ion in aerosol is in equilibrium. So condensation of NH3, H2SO4, HNO3,HCl is coupled. Under foggy condition, a heterogeneous module runs instead of aerosol module.
Sun, Yele (2004), AE	The contributions of mineral aerosol from outside Beijing were first estimated with a newly developed element tracer technique: (Mg/Al)aerosol = m x (Mg/Al)local + n x (Mg/Al)non-local Yao(2003) speculated that cloud process and non-cloud heterogeneous process were responsible for the formation of sulfate in summer and in spring, respectively
Sutton (1998), AE	The usual resistance model quantifies the affinity of the pollutant for the surface using a canopy resistance Rc, but is essentially only suitable to describe deposition. To simulate bi-directional exchange, a frame work is required to treat both source of NH3(stomatal emission, soil emission) and sinks (cuticular uptake, stomatal uptake). The simple Rc model may suffice for upland ecosystems where emission of NH3 is rare, Canopy compensation point-cuticular resistance model (Xc-Rw) model is suited for long-term emstimation of NH3 fluxes. Single-layer/big leaf model.
Sutton (2000), AFM	Apoplastic [NH4+] and pH of leaves show no diurnal patterns. Decomposing litter is a significant NH3 source, explaining nocturnal NH3 emissions and larger emissions following cutting. 2-layer model is recommended to be used in ATMs, because of its balance between simplicity and adequate process description.
Sutton (2001), PS	2-layer model provides a balance between detail and ease of applicability. The detailed NH3 exchange models need to be generalized to a form that can be applied widely with suitable input data to be used in CTM. While livesttock dominate point source NH3 emissions, exchange with vegetation play a major role in regulating both air concentration and the extent of long-range transport. NH3 is emitted from intensive agricultral ecosystems, while semi-natural ecosystems act as NH3 sinks.
Sutton (2007), EP	Very nice figure for Nemitz(2001)'s 2-layer compensation point model. Nemitz's model may be sensitive to different plant growth stage, on which the compensation point will depend.
Sumner and Shepson (1999), Nature	HCHO play important role in ODE in Arctic? HCHO helped release of bromine from sea salt, which then destroy O3? HCHO in atmosphere is in equilibrium with the HCHO in snow?
Tabazadeh (1998), GRL	Global chemistry-transport model overestimate HNO3 in upper troposphere, which will affect O3 concentration over there. considering mineral and biomass aerosols may mitigate the overestimate since they may scavenge HNO3.
Tagaris (2007), JGR	Future pollutant concentrations are more sensitive to the expected change in emissions than to the expected change in climate change (e.g., temperature) Emissions of SO2, NOx are projected to decrease by ~50% in 2050 considering all facters including climate change, growth in human activity, and emission controls. O3 and PM2.5 concentration will be lower in 2050 comparing to 2001. OC will dominate PM2.5
Takami (2005), AE	The size distributions of ammonium, sulfate and organics match well with each other throughout the entire observation period, indicating these specises were internally mixed.
Talbot et al. (2005), JGR	short time period of high NO suggest a NOx limited enviroment that restricts local O3 production
Tan, Q (2002), JGR	20% of anthropogenic SOx emitted over East Asia is exported into midtroposphere overlying the North Pacific Ocean, where sulfate tends to be longer-lived and stronger zonal winds favor more rapid tranposrt. RegCM is based on MM4 with several physical parameterization modified for long-term climate simulation.
Tanaka (2000), EST	Paraffinic ozone precursor species that otherwise would be inert to OH attack can be rapidly activated by Cl and contribute to ozone formation. So Cl2 enhances ozone production when alkanes dominate the hydrocarbon mixture.
Tanaka (2003), AE	HCl is directly produced by the abstraction of hybrogen from hydrocarbons by Cl Chloride chemistry enhances ozone formation primarily in the late morning hours is suggested by the regional modeling using CAMx
Tang, Youhua (2004), JGR	Aerosol growth due to coagualtion is only considered for accumulation mode (the 1st 2 sections). Fresh dust doesn't contain gypsum. So all sulfate associated with dust surface come from gas phase. Heterogeneous chemistry is handled by SCAPE. Calcium in the dust affects the gas-aerosol equilibrium by shifting the equilibrium balance to an anion-limited status, which benefits the uptake of sulfate and nitrate, but reduces the amount of aerosol ammonium
Tang, Youhua (2007), JGR	Most regional chemical transport models have used fixed concentration profiles as their BC, thus the corresponding specific variability in the regional simulation mainly reflects the contributions of emissions, transport and chemical processes within the model domain. Using global model BC can bring time-varied external signals to the regional domain, and reflect certain event informations such as biomass burning, stratospheric intrusion, and Asian air mass inflow. O3 below 3km is not sensitive to BC, but beyond 3km, o3 is dominated by BCs. CO is impacted by BCs for all the layers.
Tarasick and Bottenheim (2002), ACP	increase of Mercury is a direct result of the increase in occurrance of depletion episodes. ??
Tarasick (2007), JGR	For inorganic heterogeneous chemistry (gas-particle partitioning of H2SO4, HNO3, and NH3), a thermodyanmic model based on ISORROPIA is used. AQM show considerable skill at forecasting boundary layer ozone but have serous discrepancies in the free troposphere. This may due to BC and stratosphere influx.
Taubman (2004), JAS	Appalachian Lee Trough and the nocturnal low-level jet often associated with air pollution events in the eastern US. In PBL the lift times of most precusor species are relatively short because of deposition. So after escaping into lower free troposphere the lifetime is protracted. So air parcel there could be more photochemical aged. So more O3 and larger particles were observed in low free troposphere.
Taubman (2006), JGR	Air mass from northern Ohio River Vally results in high pollution in New England area. Slightly different conclusion from Kleinman (2000), which think the air mass from W-SW, along the north-east corridor, will results in high pollution in New England area.
Ten Brink, Harry M. (1998), JAS	Nitric acid does not react with dry NaCl-aerosol
Tesche (2006), AE	For most species there is higher bias in the winter and lower bias in summer suggesting potential issues related to mixing ( too little in winter ) in CMAQ
Textor (2005), ACPD	Wet removal could be due to scavenging by convetive and stratiform rain.
Textor (2006), ACP	There is high diversities among models for vertical aerosol dispersal, which are particularly relevant for aerosol radiative forcing. (Xiaoming: Thus, Vertical aerosol profile is important for AOD. For regional model that use constant BC. The vertical aerosol profile may not be accurately predicted. )
Theeuwes (2013), JGR	Nice flat UHI during nighttime. the single layer urban canopy model (Kusaka et al., 2001, Chen et al., 2011) with adjustments from Loridan et al. (2010) is used, which is coupled to the NOAH land surface scheme (Ek et al., 2004) The only deviation from these default values is that no anthropogenic heat flux is prescribed. Generally, the warmer air in the city rises and transports in the air from the rural surroundings into the city itself; this effect is called the urban breeze and has been described a number of times before (e.g. Hidalgo et al., 2008). However, this phenomenon has not been discovered in this modeling study. This is caused by a relatively high background wind speed in this case, which is around 5 ms-1 during the day and 2.5 ms-1 at night.
Thompson (2007a),JGR	Stratospheric contribution to tropospheric column O3 23% in July-August 2004 over northeastern North America
Thompson (2007b), JGR	Typical summertime stagnation-pollution was rare in JJA 2004. Precipitation and cloudiness were 25% greater than normal. Surface O3 is in lowest 35% of values recorded during JJA 1996-2004. A persistent low-pressure system led to exchange between stratosphere and troposphere.
Tian (2008), EST	Prescribed fires are intentionally ignited . Air Quality impacts from prescribed Forest fires.
Tian, X.-J. (2008), JGR	Two advantages of 4DVAR: dynamic constraint and assimilation of multiple time observation.  However adjoint or tangent linear model is required, which are difficult to obtain.    This paper merged Monto Carlo method and proper orthogonal decomposition into 4DVAR to avoid using adjoint model.     EnKF can provide flow dependent background error covariance but lacks the dynamic constraint as in 4DVAR.
Tie (2003), JGR	considering clouds, the global OH concentration will increase, O3 concentration will also increase in the middle and upper troposphere, overall O3 will increase. Since OH increase, CO will decrease.
Tie (2006), AE	Since there is less VOC in Eastern China than in the Eastern US, O3 formation rate and O3 concentration in summer is lower in Eastern China than in the Eastern US. Biogenic emissions of VOC contribute substantially to the production of O3 in the Eastern US.
Tie (2007), AE	O3 is produced via oxidation of alkenes and aromatic HCs during the day and removed by NO at night. O3 production in the metropolitan area of Mexico City is HC limited.
Tippett (2003), MWR	The EnKF and other ensemble data assimilation methods belong to the family of square root filters. After assimilation of observations, the analysis ensemble can be constructed stochastically or deterministically.    Schur products can be used to filter ensemble covariances and effectively increase covariance rank.      Covariance inflation is one simple way of accounting for model error and stabilizing the filter.
Tiwary (2007), EMS	If the residence time of air parcel is long enough (longer than the life time of isoprene to OH and NO3), then chemical reaction may need to be considered in the model. Chemical process of isoprene in canopy is not considered in this study
Tjernstrom (2009), JAS	The RL turbulence could be as strong as in the underlying stable boundary layer. The gravity waves generated by flow over the low-relief terrain may be responsible for the locally enhanced turbulence in the residual layer. This may explain why models require enhanced vertical mixing in the residual layer (like that produced by the "long-tails" stability function)
Tolocka (2001), AST	On average in Phoenix crustal component accounts for 5% of total PM2.5, which is the highest among the 4 sites.
Tombette (2006?2007?), AE	3-D application of size resolved aerosol model SIREAM with equilibrium, hybrid, kinetic mass transfer approaches.
Tong (2005), AE	NOz=NOy-NOx The HYSPLIT4 model was chosen to calculate air mass trajectories for cluster analysis Process budget analysis using MAQSIP model
Tong (2006), AE	CMAQ continental US simulation for year 1996. Vertical O3 distributions are better simulated when using MOZART2 prediction as lateral boundary condition than using predefined LBC in CMAQ.
Topping (2005), ACP	ADDEM (Aerosol Diameter Dependent Equilibrium Model) use a combination of mixing rules and other potentially more accurate techniques to calculate the water content. This model is based on the method of direct minimisation of the Gibbs free energy.
Torn (2006), MWR	Ensemble boundary conditions must be perturbed to prevent an ensemble forecast from losing variance as lead time increases.          Background-error covariance control how observational information affects model variables.
Tratt (2001), JGR	Only small particles (diameter < 4 um) are subject to long range transport The plume transported to USA consisted of up to three well-defined layers distributed throughout the free troposphere
Trebs (2005), JGR	4 thermodynamic models are used to study equilibrium using observed data in Amazon. The thermodynamic model EQSAM2 includes organic acids! (not explicitly modeling the thermodynamics of organic acids tough) In the tropic site in Amazon Basin, NO3-, Cl-, SO4= were balanced predominantly by mineral cations(particularly K+). NH4+ is largely neutralized by LMW polar organic acids. So in such case, to simulate the gas/aerosel partition, the equilibrium model'd better consider both mineral species and organic acid.
Trebs (2006), ACP	Dry deposition of NH3 is inferred using sutton et al (1998)'s dynamic canopy compensation point model. Grass surface appeared to have a strong potential for daytime NH3 emission, owing to high canopy compensation points. The exchange of NO, NH3, HONO and NO2 between surface and atmosphere may be bi-directional. The predicted bi-directional NH3 flux are most sensitive to the H+ and NH4+ concentrations in the apoplastic fluid.
Trier et al. (2011), WF	YSU produce less precipitation in high plains due to less moisture in PBL induced by stronger vertical mixing. The more rapid initiation in simulations with strong surface exchange results from a more rapid removal of negative buoyancy beneath the level of free convection, which arises primarily from greater PBL warming.
Tulet (2005), JGR	Gaseous species that interact with aerosol phase are from two different categories: mineral and organic species. Mineral thermodynamic balance is predicted by ISORROPIA or ARES. Meso-NH-C model is online model.
Tulet (2006), JGR	speed up of aerosol model ORILAM-SOA: speed up of CACM and partitioning of organics. Thermodynamic partitioning typically account for Raoult's law equilibrium between the vapor and a liquid phase and Henry's law equilibrium between the vapor and an aqueous aerosol phase.
Turner (2008), JAWMA	Secondary formation, which is often regional in nature, drives fine particulate matter toward high intraurban spatial homogeneity while black carbon, which is dominated by primary emissions within the urban area, tend to exhibit greater spatial heterogeneity. Low intersite temporal correlation is and indicator for spatial heterogeneity.
Turpin (1991), EST	Secondary organic aerosol appears to have contributed roughly half of the organic aerosol in Pasadena during mid day summer conditions.
Uno (2004), JGR	Dust loading is transported with the meandering of synoptic-scale temperature field at 500-hPa level. Dust transport was trapped with in the typical potential temperature ranges
Uno (2006), JGR	Reliable surface land use conditions and soil/surface information are more important than the complexity of the dust emission scheme or model horizontal resolution.
Van de Wiel (2012), JAS	diffusion time scale 10-15 min; geostrophic wind dictate the existence of turbulence two limiations for heat flux: neutral limit, stability limiation
van Heerwaarden (2009), QJRMS	CBL regulates how intensely the atmosphere can extract water from the land surface (thus, LE). Bett (1994) showed increase in EF by entrainment? land surface conditions, atmos stability may not be the dominant facters in surface heat fluxes. To reproduce realistic surface flux and CBL properties, it is fundemental to have accurate moist profile and represent dry-air entrainment properly.
Van Leeuwen (1999), MWR	ensemble size of at least 100 members are necessary to obtain reliable converged error variances.  At this size, the different between EnKF and double EnKF is negligible.
Van Loon (2007), AE	The diurnal cycle of Ox (NO2+O3) is better simulated than O3 diurnal cycle, showing that titration of O3 by NO is not well simulated
Vautard (2001), AE	First guess (6-h forecasts) provide better diagnostic quantities such as cloudiness than analysis. Boundary condition of PAN and formaldehyde are critical for O3 prediction in the model domain. Half of forecast error may be due to the errors in BC estimation. A hybrid statistical-deterministic chemistry transport model: boundary conditions come from a statistical model based on back trajectories. The CTM is forced by ECMWF forecasts. Model is initialised at 00UT on Day-3, providing forecasts at lead times from Day-1 to Day+3.
Vayenas (2005), JGR	Nitric acid vapor has a much shorter residence time in atmosphere than aerosol nitrate, so partition correctly is important for control strategies.
Venkatram (1999), AE	The electical analyogy does not apply to modeling dry deposition of particles since tranport of particles cannot be represented in terms of a concentration gradient as in the case of particle settling.
Vignati (2001), JGR	Nucleation is only important in very clean environments. Impact of sea-salt aerosol on corrosion of materials are significant. Atmospheric transport and dry deposition have been indicated as an important pathway for eutrophication of coastal seas.
Villani (2005), BLM	delta theta is set arbitrarily in slab model; stratification of air above PBL is key factor in tetermining PBL development, so be careful when prescribe the vertical temperature gradient. Slab model generally underpredict PBLH?
von Donkelaar (2006), JGR	estimate surface PM based on MODIS and GEOS-CHEM
von Glasow (2002)	Roughly 30% of the total chemical O3 destruction in the Marine boundary layer in the model was due to halogen reactions
von Salzen (1999), AE	Since in general dissolved ionic species are favored and neutral species are of negnigible concentration in secondary aerosol, only transformations between gas molecules and ions are included in the aerosol model.
von Salzen (1999), JGR	2-D application under offshore wind condition. aerosol model SEMA is used to solve mass transfer, in which metastable aerosol assumption is assumed.
Walcek (1997), AE	Aqueous-phase reactions reduce free radical concentration in clouds, so reduce the O3 produce rate, so under typical urban conditions, O3 formation rates are reduced by 30-90% when aqueous-phase reactions are considered.
Walker (2006), AFM	Over a soybean canopy, emission fluxes were consistently observed during the late morning and early afternoon. Uptake via the leaf cuticle dominate stomatal uptake as a deposition process. Near the source of NH3, dry deposition is high, while the flux is bi-directional further downwind owing to the presense of compensation point. Cuticular resistance was large due to dry condition and accumulation of NH3 on leaf surfaces.
Walker (2008), AE	Nemitz et al. (2001)'s two-layer canopy compensation point model is applied. Cuticular uptake is the primary deposition process though stomatal and ground uptake also play some roles.
Wang, M.-X.(2001), Advances in AS	Dust is transported to the upper layer through the daytime vertical turbulent diffusion in the PBL and the upward current in front zones. The horizontal transportation of dust general occurs in two layers: in lower layer (below 4 km), dust is moved by the cold front; and in upper layer (above 4 km), dust is transported in the westerly belt, which is very efficient and faster. The sulfur transport is largely limited to the 2-6 km level of the atmosphere.
Wang, Jun (2003), GRL	MODIS AOT can be used to qualitatively descern air quality. AOT have a good positive correlation (0.7) with surface PM2.5 mass.
Wang, Tao (1997), JGR	CO appeared to be the best chemical indicator of the relative intensity of the outflow from the continent. SO2 and NOy correlated reasonably well indicates that elevated SO2 and NOy may have due to sources of common origin
Wang, Tao (2001), AE	Cleansing agent OH, pollution-laden continental air Much higher levels found in the marine trajectories suggest the impacts of regional and/or sub-regional emission CO a tracer of anthropogenic pollution with medium lifetime of 1-2 month
Wang, Tao (2003), JGR	The data are segregated into several air mass groups based on the concentration of 22Rn and wind direction. The 22Rn is more representative of air masses that have had recent contact with land, whereas the variation of CO is more stongly impacted by combustion pollution. Cold Surge
Wang, X. (2009), AAS	WRF/Chem-UCM is used with 2004 MODIS data
Wang, Zifa (2000), JGR	Walcek advection scheme is used for transport. Multiple layers transport for dust, but not for gaseous pollutants. Most coarse particles are in surface layer since the large settling velocity prevent them rising to higher levels
Warneke (2006), JGR	Acetonitrile (CH3CN) tracer of forest fire emission and biomass burning. Chloroform tracer of an thropogenic emission. Over NE, during summer 2004, 30% of the CO enhancement is attributed to the forest fires in Alaska and Canada. 70% is attributed to urban emissions of mainly New York and Boston. Biomass burning caused enhancement between 3-5km. On some days the forest fire emissions were mixed down to the surface.
Washington (2000), CD	PCM scales well up to 64 processors. The atmospheric component in PCM is CCM3 at T42 resolution, which is a spectral model for the basic dynamics.
Watson and Chow (2002), AE	Conceptual model for SJV PM formation: Nitraic acid is formed in the upper layer at night, and is prevented from deposition by surface inversion. So the nitric acid is available over non-urban areas with high ammonium emissions to rapidly form NH4NO3, which will reach the surface next morning because of vertical mixing.
Watson (2006), JAWMA	nanoparticle (3-10 nm), ultrafine (10-100nm) Although SO2emissions in california are low, the small residual amount is sufficient to initiate nucleation event. PM2.5 concentrations were poorly correlated with nanoparticle number. Observed nucleation rates are often orders of magnitude higher than can be expected from binary nucleation mechanism.
Webster (2007), AE	Industry emission of VOC from non-electricity generating units have been traditionally assumed to be continuous and at constant level for photochemical modeling purpous. Impact of O3 formation of variablity of routine industry emission that is not enough to require reporting as an emission event is investigated.
Weckwerth (1999), MWR	The formation of convective rolls depends on the magnitude of buoyancy flux; when -z/L exceed some critical value, the rolls will decay
Weisman (1997), MWR	4k resolution is sufficient to explicitly simulate squall-line behavior (one of the mesoconvective systems)
Wert (2003), JGR	ethene and propene alone could explain the high O3 levels in Houston. Ethene and propene emissions from 1999 TNRCC inventory are underestimated in the emission iventory. There is one of the largest power plants in the nation outside of Houston.
West (1999), JAWMA	Nonlinear PM2.5 response to SO2 reduction. However in summer, when PM2.5 is often highest, SO2 reductions will generally be effective.
West (2004), JGR	Emissions of VOC and CO are often underestimated. Proved to be true for Mexico City.
Wexler (1990), AE	different treatments of mass transfer flux for solid and liquid particles. Time scale for single particle and aerosol population. Water content is in equilibrium. Unity accomadation coef. on aqueous phase particles. For solid particles it is much smaller. under atmospheric condition the concentration of the associated species is considerably less than the concentrations of the dissociated species. Give mass flux for both aqueous phase and solid particles: These flux can be affect by surface heating due to the latent heat of condensation and motion of the particle. But they are negligible. Have I touched Dr. Braham's question?
Wexler and Seinfeld (1991), AE	In AIM, equilibrium composition is calculated by directly minimizing the Gibbs free energy. Kusik and Meissner is used to predict activity coefficient. At low RH conditions, mass transfer calculation is challange since small change of hydrogen ion will result in large change of ph due to small water. so surface vapor pressure need to be evaluated frequently.
Wexler (1992), AE	If the aerosol is in solid state or osmotically dominated by ammonium nitrate, the surface equilibrium concentration can't change, so tp is infinite. Aerosol is hygroscopically externally mixed, ammonium nitrate only transport to a hydrophilic fraction of the particles. So use of the overall number distribution to estimate gas equilibrium time tends to overestimate the surface area available for transport (condensation)
Wexler (1994), AE	up to 70% of organics and 90% of inorganics can be of secondary origin. Water is in equilibrium. accommodation coef. 0.01. Condensation if more important comparing with Coagulation, nucleation when considering the mass concentration computational efficiency of condensation solver is not important since the time spend by thermodynamic model dominates time step is small enough that the condensation rate is constant when integrate the condensation equation. So it must be too computational expensive. is this the reason Lurmann move to equilibrium approach in the companion paper?
Wexler (2002), JGR	extend AIM to include double salts and to low temperature condition, which could be potentially applied to stratosphere. Double salts may form in some coastal cities. So the extended AIM may be useful for such cases.
Whitaker (2002), MWR	If the same obs. and same gain are used to update each member of the ensemble, the ensemble will systematically underestimate analysis error covariance, which may lead to filter divergence.     In EnSRF, the traditional gain is used to update ensemble mean but a reduced gain is used to update deviations from ensemble mean, while in EnKF these two gains are the same.         Techniques to boost ensemble variance are almost always necessary in ensemble data assimilation to prevent filter divergence.
Whitaker, 2008, MWR	express model state vector as an ensemble mean and a deviation from mean.  The EAD analysis yielded a 24-h improvement in geopotential height forecast skill relative to the NCEQ 3DVAR system. Among three parameterization of system error, additive inflation performed best in this test. EnKF should become more and more attractive relative to 3DVAR as forecast model improve and amplitude of model error part of background error covariance decrease.
White (2007), JGR	Wind conditions and PBL height favor the high O3 accumulation in 2004, but 2004 experienced the least O3 exceedance days during the past 10 years. Why? extreme climatic conditions "Cloudier cooler wetter". In contrast, southeast US and central valleys of california always suffer high pollution when PBL height is low, which is caused by subsidance inversion. Possible transport from industrial midwest and the urban megalopolis along the east coast to NE area
White (2009), JAMC	The impact of such episodic warming events on the closure of budget of CO2 is an concern While it is known that some warming events are foreced by synoptic and mesoscale fronts, there are many cases in which weak gravity waves propogating on the inversion surface would be a plausible mechanism The intermittent turbulence event may be able to transfer a large portion of the scalars that accumulated during the calm period (Acevedo et al., 2006, BLM)
Wilczak (2006), JGR	Bias-corrected ensemble provide better skill. In some models greater horizontal resolution alone may not improve model performance.
Winkler,sandra (1996), EST	Transport refers to the advection and diffusion of air pollutants. Gear-based codes such as LSODE accurately solve stiff systems of chemical equations but with excessively long computing times Using inaccurate numerics in the modeling of a complicated system may hide potential errors in input data and model assumptions
Woo (2003), JGR	Different regional CO/CO2 ratio could be related to the regional economic development. Differences between fossil fuel and biomass burning could be reflected in the BC/OC ratios.
Wood (2004), JC	Deeper marine boundary layer tend to be more decoupled than shallow ones
Wu, S (2008), JGR	Decreasing cyclone frequency as a result of greenhouse warming appears to be a robust result from both model studies and long term observations. O3 in the south east US is insensitive to climate change, reflecting compensating effects from changes in isoprene emissions and air pollution meteorology
Wu, W.-S. (2002), MWR	3DVAR in physical space allows inhomogeneity and anisotropy for background error statistics, while 3DVAR in spectral space doesn't                 Analysis variable defined on the grid: Streamfunction, unbalanced part of velocity potential. etc.
Wu, Y (2003), JGR	Good description of Meyers et al (1998)'s Multilayer model. Upgrade of Meyers (1998)'s multilayer model. It is designed for use in dry deposition networks. The model only need routine meteorological and chemical concentration measurements.
Wyers (1997) AE	The deposition of submicron aerosol has been considered relatively unimportant as compared to deposition of gaseous compounds. The enhanced nitrate evaporation near canopy would cause a low concentration of aerosol nitrate near canopy and apparent high deposition velocity of aerosol nitrate.
Wyers (1998), AE	Preceding flux to the leaf surfaces may lead to accumulation of NH3 and to an increased resistance to deposition and emission potential. So dynamic NH3 exchange model is needed to take into account of such memory effects. Rw is not a simple function of RH, but also influenced by other factors such as accumulation of NH3 and other pollutants on leaf surface.
Xia (2008), JGR	Weekend effect, AOD during weekday is larger than during weekend in US and Europe, but not China and India. Surface aerosol may not be a good indicator of column-integrated value (i.e., AOD) in some cases.
Xiao, Hui (1997),JGR	Of the total sulfur emitted, 33% is removed by dry deposition. The remaining 67% of the sulfur emitted is transported out of the region. The region of maximum total sulfur outflow during this 2-week period is located at height between 2 and 6 km
Xiao, Q.-N. (2007), MWR	background error statistics for WRF3DVAR are computed using ensemble forecasts of the same case.  More radar data are assimilated, the more positive is the impact on the QPF skill.  cv3, cv5 background error statistics options in WRF-3DVAR?        control variable vector contains component of the streamfunction, velocity potential, unbalanced pressure, and total water mixing ratio.
Xu, Xiaohong (2000), AE	In SAQM, the dry deposition velocity for HNO3 were adopted for Hg(II) since the aqueous solubilities of these two species are considered to be similar. Bi-directional air-surface exchange of elementary mercury
Yamaguchi and Randall (2008),JAS	Nice minimal liquid water potential temperature at the boundary layer top due to cloud processes in their 2012 JAS papaer, but barely see that in this earlier 2008 paper
Yamaguchi and Randall (2012),JAS	Nice minimal liquid water potential temperature at the boundary layer top due to cloud processes
Yao, Xiaohong (2001), EST	Description of the sampling in May 1998, HK. Seawater [Na+]/[Cl-]=0.85 The sampling site is principally affected by sea salt aerosols and is slightly polluted by regional pollution ammonium nitrate is not expected to be present prior to collection mainly because of the ambient high temperature. Size-segregated samples can redue the sampling artifact of chloride depletion as a result particle-particle interactions.
Yao, Xiaohong (2003a), AE	droplet mode centered 0.7 um was attributed to in-cloud processing. The evaporation of cloud water results in droplet mode sulfate aerosols. Another mode centered 0.45 um was ascribed to non-cloud heterogeneous processes
Yao, Xiaohong (2003b), AE	Chloride depletion in fine sea salt particles can be ascribed to heterogeneous reactions and in-cloud processes. ambient concentration of sulfate in the droplet mode under the existence of low cloud were evidently larger than those without low clouds.
Yao (2006), AE	Since NH3/NH4+ equilibrium for overall PM2.5 was probaly not completedly achieved, SO42- was not completely neutralized even when (NH3+NH4+)/SO42->2.0
Yao, Xiaohong (2006-7?), AE	Strong acidity [H+]+[HSO4-] The in situ PH decrease with increasing submicron particle size, which may indicate the netrlization rate of smaller submicron particles is faster. Equivalent ratio of [NH4+] to [SO42-] is less than unity, suggesting the possibility of incomplete gas-aerosol equilibrium
Yao (2007), AE	Fresh soot particles are mainly ~0.1um. The on-road 0.2um mode particles are consideered to be the result of coagulation and condensation growths of soot particles.
Yeatman (2001), AE	NH3 is generated primarily from the decomposition of agricuture livestock waste and to a lesser extent, fertiliser production. Shift from fine mode to coarse mode could result in enhancing both dry and wet deposition Dissolution/coagulation shift NH4+, NO3- to coarse mode
Yu, F (2006), ACP	Ions are generated in the atmosphere by galactic cosmic rays and other ionization sources. The second generation ion-mediated nucleation.
Yu, F (2008), ACP	The ion-mediated nucleation is sigificant in the tropical upper troposphere. Activation theory of new particle formation remains to be further investegated.
Yu (2005), JGR	NMEs of predictions of nitrate from CMAQ are from two to three times larger than the NMEs for sulfate 4 possible reasons for the discrepancies between the model predictions from themodynamic model and observations
Yu, SC (2006), JAWMA	a 48-hr ETA-CMAQ forecast is needed based on the basis of 12-UTC initializeation to obtain the desired 24-hr forecast period. mobile sources (or urban plumes) SO2/NOy < 1. point sources from power plants SO2/NOy > 1
Yu, SC (2007), JGR	O3 lateral BC above 6Km is augmented using GFS forcast. overpredict O3 concentration in areas of cloud mainly caused by unrealistic convective cloud scheme. Accurate description of forntal passages, cloud cover and wind fields are key to forecast regional air quality. Underestimate CO because of transport of biomass burning, underestimate NO because of exclusion of aircraft and lighting emission
Yu, SC(2007), EST	underprediction of both OCsec and OCpri is due to 1). missing source of OCpri in summer emission inventory and 2) oxidation of isoprene and sesquiterpernes are not included in CMAQ
Yu, SC (2008), JGR	CMAQ BC in this study: static vertical profile. Overestimate of SO42- lead to underestimate of NO3-. Need improvement of SOA formation process.
Yuan (2008), JGR	Cloud droplet effective radius may show positive relationship with aerosol loading, which is measured in terms of AOD in some area.
Zaveri (2005), JGR	During mass transfer calculation, repeated evaluation of activity coefficient is computational expensive. Some previous multicomponent activity coefficient models: KM and Bromley are based on mixing rules. PSC is a mole-fraction-based ion interaction model. MTEM combines the accuracy of PSC model and efficiency of Metzger et al. (2002) to estimate activity coefficient.
Zaveri (2005b), JGR	good review of thermodynamic model. GFEMN and AIM are regarded as the most accurate thermodynamic models. MESA only works for low troposphere so far.
Zaveri (2008), JGR	Carbonaceous aerosols include sooty particles and organic compounds. Gas/particle partitioning's theoretical formulation is well established. The difficulty primarily lies in its numerical solution. CBMZ is implemented in a way that the entire mechanism is partitioned into 4 submechanism. It is important to obtain accurate solution in the initial stage for dynamic mass transfer approach because the transport-chemistry time splitting interval is short (5-30 min). A numerical integrator must take time step on the order of shortest time scale of integrated process. MOSAIC solve condensation of non-volatile species at first, then solve the dynamic mass transfer of volatile species. The bias of predicted aerosol size distribution could be cuased by the errors in the size distribution of emitted aerosols.
Zhang C. L. et al. (2009), JGR	Method of Liu (2006) is used to simulate the impact of urban using MM5. Urbanization leads to less CAPE (more convective inhibition energy) due to less water vapor in the boundary layer
Zhang, D.-L. (2004), JAM	MYJ predict cooler and shallower PBL than other PBL schemes (including MRF) due to less upward transfer.    T2=0.45Tg + 0.55Ta.     All the schemes contain nonlocal treatments of the unstable PBL developments, which include TKE closures(e.g., MYJ), countergradient heat fluxes (MRF). ?????????
Zhang, D.-L. (2009), GRL	2001 NLCD data is used for simulations of year 2007
Zhang, D.-Z. (2000), AE	sulfate dominate fine mode and nitrate dominate coarse mode, which contians significant amount of soil (mineral elements). Soil particles were more suitable fore internal mixture of sulfate and nitrate than other kinds of particles.
Zhang, D.-Z. (2004), GRL	interaction with sea salt would change dust particles' size/composition distribution.
Zhang, F.Q. (2005), MWR	EnKF could estimate flow-dependent background error covariance Some other data assimilation schemes assume stationary isotropic background error covariance.     The structure of mesoscale error covariance is shown to be flow depedent and highly an isotropic.
Zhang, F.Q. (2006), MWR	The truth simulation is chosen from 50 different random realizations to compare most favorably to observations of this event. EnKF is most effective for pressure perterbation becasue it has the strongest larger scale component among all variables while it is least effective on vertical motion and moisture field which has small scales.   MM5 3DVAR background error covariance is used to generate initial conditions.      Larger ensemble spreed could prevent filter divergence.
Zhang, F.Q (2007), JGR	ensemble forecasts for TxAQS2000 case using large sample of dynamically consisten IC and BC. The paper demostrate the importance of accurate representation of meteorological condition in the high O3 event.
Zhang, F.Q. (2008), BAMS	Simutaneous state and parameter estimation through state augmentation could treat parametric model error.     EnKF performs better than 3D variational system over data sparse regions where observation spacing is comparable or greater than characteristic correlation lengths.
Zhang, Lin (2007), ICCS 2007, Part I, LNCS	Different approaches for data assimilation: Kalman filter technique, variational method. Data assimilation use obs to obtain better IC, BC, and emission etc. for model simulation to constrain model prediction.
Zhang, Yang(2006), AE	Gas/partical partition is solved by ISORROPIA, the condensational growth of nitrate is not treated in CMAQ, and also the aqueous phase kinetic reactions leading to nitrate formation. A kinetic mass transfer approach is recommended. NET96 does not account for seasonal varability. NET99 inculded the temporal varibilities
Zhang, Jian (1998), JAM	Vorticity of the flow increase on the leeside of the Appalachians, and a leeside trough forms usually. A nighttime low level jet near the top of stable layer overland is evident. Leftover pollutants in the residual layer from the previous day will be entrained downward into the mixed layer in the gumigation process. The vertical mixing process might be the primary source for the observed rapid increase in ground level ozone conc. during the early morning
Zhang, Jian (1999), JAM	Vertical mixing contribute to O3 increase in the morning, decrease in the afternoon because of dilution. O3 increase rate is higher in the morning and slower in the afternoon because of mixing dilution in the afternoon. In the morning both vertical mixing and chemical production contribute to O3 buildup.
Zhang, J (2002), JGR	Thermodynamic equilibrium may or may not be a reasonable approximation for PM2.5. ISORROPIA is used to study aerosol equilibrium based on high time resolution measurements. equilibrium prediction based on average conditions over a sampling period may not reproduce the actual partitioning
Zhang, K. Max(2002), AE	Gravatation settling velocity is part of deposition velocity, but it is only appreciable when Dp>1microm; accomadation coefficient 0.6
Zhang, K.M. (2005), AE	sea-salt generation functions for open-ocean is propotional to U103.41 For surf zone sea-salt emissions can be 1-2 orders higher than the oceanic background
Zhang, K.M.(2006), AE	ATS is developed to solve the growth term in the condensation equation. Aerosol thermodynamic states are typically near equilibrium because the perturbation timescale is longer than aerosol equilibrium timescale.
Zhang, K. M. (2008)	AIM is too slow to be incorporated into large-scale atmospheric models. The detailed phase state is not necessary for gas-particle partitioning. A simplified thermodynamic model is used in CMAQ-UCD. The scavenge of unactivated particles by falling raindrops is not considered. "shift" term is solved using one from Dhaniyala and wexler, 1996, while the "growth" term is solved by ATS integrator.
Zhang, L (2003), ACP	Simple models have the advantage of requiring fewer assumptions regarding input parameters, which avoid additional uncertainties. Bi-directional exchange is not included due to lack of information over broad land types. Non-stomatal resistance formulation is improved.
Zhang, Meigen (2003), JGR	To evaluate the impact of the anthropogenic emissions on the distributions of trace gases and aerosols, the IC and BC were chosen at the lower end of their obs range. Stratospheric influence on tropospheric ozone is parameterized by specifying the IC and BC at the top three layers of the model to values proporional to potential vorticity. Highest outflow concentrations are in lower troposphere while maximum outflow fluxes are in free troposphere (3-6km)
Zhang,Qiang, C. Zhao (2006), AE	The strong vertical gradient between the PBL indicates that the PBL plays an important role to prevernt the dust particles to be transported onto the surface, and fine particles from the surface to the free troposphere
Zhang, Yang (2008), AE	A bi-directional deposition scheme is needed to represent the flux of NH3 in AQMs over fertilized agricultural crop lands and grazed grasslands, where the NH3 aurface-atmosphere exchange is bi-directional.
Zhang, YongXin	MSM-LSM was initialized everyday, was run for 36-h simulation. Vertical extent of sea breezes generally ranges from 1.0 to 1.5 km
Zhao, C.S. (2005),JMSJ	Due to its large radius, sea-salt particles are activated into cloud drops in the initial cloud development. Sea-salt activation decreases supersaturation by consuming water vapor and suppresses nss-sulfate activation. Nss-sulfate indirect forcing may be overestimated in some conditions because of the presence of sea-salt particles.
Zhao, C.S. (2006), GRL	The reduction of precipitation over eastern China is strongly correlated to the high concentration of aerosols. Stability is shown to be increased due to the increasing loading of aerosol.
Zhao, T. L. (2006), JC	The frontal cyclone uplifted dust aerosol in the free troposphere for trans-pacific transport. Trans-pacific dust transport peaked between 3 and 10 km in the troposphere. ~3% of Asian dust aerosol was carried to the North American continent via trans-pacific transport. Over the Asian dust source regions, dry deposition is the dominant removal process, while on the pathway, wet deposition exceed dry deposition by a factor of ~20.
Zhong et al. (1996), MWR	The pressure gradient force and the geostrophic wind speeds were considerably smaller outside the jet core than inside. This suggests that a strong east-west synoptic-scale pressure gradient is crucial for the formation of a strong LLJ. Without such a large-scale pressure gradient, or the "basic flow" on which the frictional stress could operatre diurnally, there would not be the strong diurnal LLJ so characteristic of the Great Plains
Zhou (2010),NH	UHI is more prominent in spring, which is related to cloud fraction. Increasing Atlanta's Albedo by 100% was not an effective way to mitigate UHI effect while when Albedo is tripled it is effective! WRF-NOAH is used to simulate UHI effect
Zhu L (2009), Energy Buildings	the thermal mass wall does have the ability to store heat during the daytime and release it back at night massive building envelope components delay and flatten thermal waves caused by exterior temperature swings The low mass test chambers closely followed outdoor conditions and did not offer any significant thermal storage, while high thermal mass was very effective in lowering indoor maximum temperatures
Zhuang, Hong (1999), AE	sulfate and ammonium were dominant in fine particles which nitrate in coarse mode. Sampling period: winter
Zhuang, Hong (1999), AE, 4223-4233	The extend of chloride depletion was found to decrease with increasing particle size in the coarse mode. The reaction of HNO3 with sea-salt particles favors small particles because of their much larger surface area distribution and larger atmospheric residence time.
Zilitinkevich et al. (2009), BLM	Internal gravity waves contribute to the turbulence production and eventually enhance mixing intead of separating the turbulent and the laminar regimes, the Energy and flux budget turbulence closure model separates two regimes of essentially different nature but both turbulent: strong turbulence and weak turbulence
Zupanski (2005), MWR	Data assimilation used in atmospheric applications can be viewed as an effort to approximate the Kalman filter/smoother theoretical framework.