Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 20, 2007
Publication Date: May 1, 2008
Citation: Hardegree, S.P., Van Vactor, S.S., Levinson, D.H., Winstral, A.H. 2008. Evaluation of NEXRAD Radar Precipitation Products for Natural Resource Applications. Rangeland Ecology and Management 61:339-345. Interpretive Summary: The National Weather Service operates a network of radar stations that estimate precipitation over almost the whole lower-48 United States. These precipitation estimates are used for hazardous-event forecasting but are difficult to access and use outside of the specific NWS application for which they were established. In this study, we modified computer programs to make these data more accessible to other users who need spatial precipitation data to run hydrologic and natural resource models. Unfortunately, the NWS radar system underestimates total precipitation for the radar located in Boise, Idaho. We determined that this is because the threshold set for detection of precipitation is set too low to pick up the majority of low-rate rainfall events in the Boise area. Additional processing procedures are suggested that may improve the utility of these data for water balance and natural resource modeling applications where the focus is on total amount of precipitation rather than just extreme-event weather phenomena.
Technical Abstract: The National Weather Service (NWS) operates a network of Doppler-radar stations that produce hourly-rainfall estimates, at approximately 16-km2 resolution, with nominal coverage of 96% of the conterminous US. Utilization of these data by the NWS are primarily for the detection and modeling of extreme-weather events. The usefulness of these data for external hydrologic and natural-resource applications is limited by a lack of programming tools for data-processing outside of the current NWS application. We modified NWS source code to produce decoding and georeferencing tools and used them to evaluate three years of radar precipitation data for the Boise (CBX) radar relative to gauges in the Snake River Plain of southwestern Idaho. During hours when the radar was operational, radar detection underestimated gauge measurements by approximately 60%. The bulk of this discrepancy was attributed to a threshold precipitation detection function (pdf) in the radar processing that set the majority of low-intensity precipitation events to zero. Data processing and analysis tools developed for this study are available from the Agricultural Research Service, Northwest Watershed Research Center ftp site (ftp://ftp.nwrc.ars.usda.gov). Additional modification of NWS precipitation processing procedures may improve accessibility and utility of these data for hydrologic and natural resource modeling applications.