Radar-to-gauge comparison of precipitation totals: implications for quality control

Hourly precipitation totals from synoptic stations in the United States were compared to two radar-derived precipitation estimates: the digital precipitation array and the stage IV blended radar/gauge analysis. Scores that quantify the joint (gauge and radar) occurrence or non-occurrence of precipitation were used along with binned contingency tables to evaluate the agreement between gauge totals and radar-based estimates. Results suggest that for any given hourly gauge total, a wide range of radar-based estimates occur particularly for smaller gauge totals. In addition, comparison of hourly fractions of 24-hour gauge sums with the hourly fractions of corresponding radar-based values suggest that gauge/radar differences vary considerably even within individual precipitation events. Consequently, for the purpose of assessing the quality of gauge measurements, direct quantitative comparison with radar estimates is likely to be of limited value. On the other hand, qualitative use of radar estimates may be of value when the plausibility of large and spatially isolated gauge totals is assessed.