The summertime climate of the central United States is characterized by high spatial variability in precipitation, resulting in high spatial variability in soil moisture and crop stress. Monitoring these conditions and identifying areas of potential crop damage from deficient or excessive soil moisture can be problematic because the existing network of precipitation observations is not of sufficient spatial resolution to identify small-scale variations in precipitation. Further, delay in availability of the quality controlled cooperative rainfall data prevents timely monitoring of precipitation. This study examines the relationship between multi-sensor rainfall estimates and crop yield for a 9-state region (Illinois, Iowa, Indiana, Minnesota, Wisconsin, Michigan, Missouri, Kentucky, and Ohio).

Higher resolution (15-km) precipitation estimates based on a multi-sensor estimation technique employing the National Weather Service WSR-88D radars and hourly raingage data that have been obtained in near real-time from the National Center for Environmental Prediction since March 1997. The multi-sensor based rainfall estimates were averaged by county, summed for the months of May and July, and compared with NWS cooperative observer gage data obtained in real time (for two years) and after-the-fact (for five years). There was general agreement in the pattern of high and low rainfall amounts from the multi-sensor based and quality-controlled cooperative gage based data, for this region, with a correlation coefficient r of 0.78 for a 5-years period, 1997-1999, 2001-2002. Corn yields for 1997-2002, were obtained from the USDA-National Agricultural Statistics Service for each county. These data were normalized using the average yield of the previous 5-years. The corn yield pattern is generally associated with low to moderate rainfall in May and moderate to high rainfall in July. Multiple-regression analysis incorporating May and July rainfall and a number of soil parameters resulted in a correlation coefficient r of about 0.52 between these factors and yield for both data sets. These results suggest that the real-time multi-sensor data is of comparable quality to the gage data for purposes of predicting crop yields. County soil characteristics will be further investigated to ascertain variability in normalized corn yields with similar rainfall amounts.