Rain gages and radar have been used separately and jointly for many years to estimate rainfall. Rain gages measurements are typically used as ground truth measurements for rainfall, despite their known inaccuracies resulting from wind, poor exposures, mechanical problems and their point nature. Radars, though providing better spatial resolution, have known problems related to the nature of the reflectivity-rainfall relationship, the location of the radar beam within the precipitating cloud, and ground clutter. Hildebrand et al. (1979) indicated that for gage densities of better than 1 gage per 250-300 km2, gage-alone and gage-adjusted radar estimates are of similar accuracy when compared to gage rainfall from densities of 1 gage per 30 km2. However, they also indicated that for densities sparser than 1 gage per 250-300 km2, gage-adjusted radar estimates of mean area convective rainfall may be more accurate than gage-alone estimates.

For many purposes, real-time or near-real-time rainfall over a large area such as the central Midwest is desired, but with the same time and point accuracy that is typically reserved for a small areas with many gages. The summertime climate of the central United States is characterized by high spatial variability in precipitation on a daily basis. In this study, county-scale resolution is desired. However, observers reporting in real-time represent only a fraction of all cooperative observers and the resulting gage density may not be adequate to resolve the often large county scale variability.

It is the intent of this paper to compare the county-averaged rainfall estimates employing 1) real-time cooperative gages, 2) quality-controlled cooperative gages, and 3) real-time gridded rain estimates based on WSR-88D radar, real-time gage data, and on a multi-sensor approach. Finally, the accuracy of the county-based rainfall measurements will be assessed for three Illinois counties where dense (1 gage per 100 km2) gage networks are operated.