Comparison of raingage and WSR-88D Stage III precipitation data over the Texas-Gulf basin

Abstract

Precipitation is one of the most important hydrologic model inputs and is characterized by spatial and temporal variability. Traditionally, raingage networks have provided precipitation data for hydrologic studies. Weather radars like the WSR-88D estimate precipitation with high spatial and temporal sampling frequencies, but radar estimates suffer from errors due to several reasons. Assessment of the quality of radar precipitation data is necessary before the application of radar data in hydrologic studies and is important for improving radar data processing algorithms. Accuracy assessment of Stage III precipitation data from the WSR-88D network over the Texas-Gulf basin was performed for the period 1995–99 using 24-h accumulations from 545 raingages. Sampling differences between point raingage measurements and areal radar estimates were ignored. WSR-88D underestimated the five-year precipitation at a vast majority of the raingages. Difference in the total precipitation depth between radar and raingage was within ±500 mm only at 42% of the raingages; the root mean squared difference was more than 10 mm at 78% of the raingages. Such large differences between both rainfall data sources will have significant implications for the hydrologic applications of WSR-88D Stage III data. Radar estimation efficiency was better than 0.50 at 71% of the First Order/FAA raingages and 41% of the cooperative observer raingages. Radar performance varied significantly over the years, and, in general, improvement in radar performance was observed, which is consistent with the on-going developments in WSR-88D data processing algorithms; however, an overestimation trend was identified during 1998–99. Evaluation of the quality of WSR-88D Stage III data and making necessary corrections are important before their application in hydrologic studies.