Storm Precipitation Structure in the Eastern United States: Part II, Temporal Changes

Previous studies of changes in precipitation characteristics have focused on parameters derived from daily data because of their long period of record and good spatial coverage. During a project to modernize the climate inputs for estimating soil erosion rates in the eastern two-thirds of the U.S., a number of storm precipitation structure variables were derived from NCDC 15-minute gauge data (TD-3260). This digital data set starts in 1972 and currently extends to 1999. This finer time resolution data allows for a better definition of storm characteristics such as duration, frequency, maximum 15- and 30-minute intensity, measures of the distribution of precipitation within the storm, and storm erosivity.

This paper will examine the temporal changes in those characteristics at the seasonal (winter, spring, summer, and fall) and annual time scale by comparing the first half of the record with the second half of the record. Differences will be tested for statistical significance using appropriate tests (e.g., chi-square). Spatial clustering of mean storm structure parameters is used to regionalize the storm structures at stations, allowing for the possibility of identifying regional trends while minimizing the effects of a single station. The methodology of comparing the two halves of the record is used rather than other traditional trend analysis because the period of record of the digital data for most of the stations in the dataset is on the order of 25 years. Even so, the analysis should indicate nascent trends in the data.