Changes in Spatial Extent of Extreme Rainfall Events Area in the Northeastern United States Using High-Resolution Gridded Datasets

Using several high-resolution gridded daily precipitation datasets, the temporal change in the annual total geographic area (km²) receiving extreme precipitation is analyzed. Extreme precipitation events are defined by unique (separated by at least 10 km) closed 5.08 cm contours as defined by the contour function in Basemapper. The numpy convexhull procedure is used to compute the area enclosed by each contour. Analyses are repeated for three datasets, 1) simple nearest neighbor interpolation of station data, 2) PRISM and 3) the satellite-based PERSIANN-CDR.

There is significant increase in the extreme precipitation area, with quantile regression analysis showing strong positive trends across all quantiles using the station data based nearest neighbor interpolation. With PRISM, the inclusion of radar data after 2002 introduces a discrepancy in the data record that substantially increases the number of events in the lowest quantiles. Prior to this artificial break (1981-2001), trends in the central quantiles (events on the order of 10² kilometers squared) showed significant increases. The results indicate that the spatial extents of extreme precipitation events, particularly in mid-sized quartiles, are noticeably increasing temporally.