Downscaling Relationships for Summer Heavy and Widespread Rain Events in the U.S. Midwest

Anthropogenically-caused changes in the regular occurrence of widespread summertime rainfall, and local heavy precipitation events, are major areas of concern in the Upper Midwestern United States. For example, the periodic occurrence of summertime rain events are critical to agriculture, while heavy rain events can lead to tens of millions of dollars of flood damage. Statistical downscaling techniques are needed to relate climate change predictions to changes in summertime convective rainfall because of the large grid resolutions of current GCMs and the mesoscale nature of summertime convective elements. The development of such techniques has been hampered by limitations in the temporal and spatial resolution of surface rainfall observations.

This study utilizes the NCEP precipitation analysis based on radar-derived rainfall rates to develop statistical downscaling relationships with atmospheric factors that can be simulated by GCMs. The focus is on the areal coverage of moderate precipitation (over 6.25 mm da^-1) and heavy precipitation (over 25 mm da^-1) averaged over 225 km² grid areas. Daily data for four summer months (May to August) over a three-year time period (1997 to 1999) were available for this study.

Atmospheric factors examined included stability indices (CAPE, Lifted Index), afternoon boundary layer depth, and regional surface pressure patterns. Daily surface and 850 hPa analyses were also examined. Stability indices and boundary layer depth appeared to be poor predictors of the coverage of moderate rainfall, but were somewhat better predictors of heavy precipitation coverage. A large-scale weather feature (such as a cyclone or frontal system) was almost always present when rainfall covered large areas of the study region. Large-scale analyses, based on NCEP/NCAR reanalysis data, reveal a tendency for heavy and widespread rainfall to occur with low pressure to the southeast of the sample area. An index was developed to create statistical relationships using large-scale pressure fields.