NWC REU 2020
May 26 - July 31



Photo of author

Radar Rainfall Estimates Versus Station Rainfall Data

Michael Lavallee and Bradley G. Illston


What is already known:

  • Radar estimates have a multitude of difficulties with accurate estimations from the weather, terrain, or obstacles in the radar’s view. As well as difficulties from characteristics of the radar beam itself.
  • Previous studies have shown that radar estimates have a large bias towards underestimation, up to an underestimation of 80% in some studies.
  • Summer generally has a larger margin of estimation errors when compared to other seasons.

What this study adds:

  • This particular radar estimation product has a slight bias towards underestimation with around 57% of the months investigated having radar underestimates.
  • Locations that are close to the radar site and wetter have the lowest absolute errors. Also, locations with higher rainfall totals have greater radar estimate accuracy when compared to locations at equivalent distances that are drier.
  • May, the month where most convective precipitation occurs, has the largest underestimation errors.
  • Overall radar estimation accuracy could be improved by removing instances when the radar estimated precipitation when gauges recorded none.


This study investigates the Arkansas-Red Basin River Forecast Center’s P3 radar precipitation estimator accuracy with select Oklahoma Mesonet gauge locations. The data collected comes from the NEXRAD radars located within ARBFC’s forecast area as well as rainfall measurements recorded by the Oklahoma Mesonet stations. Radar estimation errors were analyzed from a temporal and spatial standpoint to determine if errors developed specific trends in estimation accuracy. Greater radar estimation accuracy was observed at locations close to radar sites and at locations that were climatologically wetter. The fall season was also shown to have the smallest margin of radar estimation errors.

Full Paper [PDF]