Identification of sub-urban scale extreme precipitation variability using a geospatial algorithm

Established urban hydrometeorological literature shows the enhanced rainfall occurs generally near and downwind of urban centers. This information is often too generalized for operational use. Anomaly analysis in a spatial framework similar to operational forecast zones may provide utility in prediction of extreme events like localized flash flooding. This paper introduces a spatial analysis algorithm deploying 4-km NCEP Multisensor Precipitation Estimates Stage IV data to determine specific areas, or sectors, of anomalous precipitation across three southern United States cities (Atlanta, Houston, and Raleigh). Rainfall frequency and intensity are quantified during the afternoon and early evening across a 21-year period. Hotspot analysis and sector-based comparison of anomalies show that specific areas within each city are prone to isolated extreme rainfall events more often than other areas. Isolated precipitation in central and eastern Atlanta, specifically, has a high rate of recurrence and intensity during convectively active hours. 700 hPa wind deviations from the zonal mean are mapped to domain sectors as rainfall anomalies exceed ten standard deviations. These results are an application of the provided geospatial algorithm which combines gridded raster data with subdivisions defined by polygon features to assess the clustering of variables (environmental or otherwise) within domains of interest.