Magnitude and significance of observed trends in precipitation frequency over the U.S

Observed trends in the frequency of precipitation and the length of longest dry and wet spells are documented over the U.S. for both wet and dry seasons and on an annual basis using historical daily precipitation data for 774 rain gauge stations from approximately 1900-2009. Stochastic Monte-Carlo simulations based on Markovian precipitation occurrence models are used as a null against which to test observed trends, providing a method to identify significant trends and regional expressions at individual stations without resorting to gridding or area-averaging. Most regions in the U.S. display geographically and seasonally consistent long-term trend patterns that involve an increase in frequency of precipitation and a decrease in the length of extreme dry spells. The Atlantic plain is an exception in terms of long-term trends in precipitation frequency, especially during the wet season, during which it experiences a marked decrease in frequency of precipitation and increase in extreme dry spell length. Geographically, our results are broadly consistent with previous analyses based on gridded data sets, but with significantly larger magnitudes (10 days per century spatial average as compared with 6 days per century). The observed trends are not consistent with climate model projections, which predict decreases in mid-latitude daily precipitation frequency in a warming atmosphere. Trends in the timing of the wet and dry seasons are also evident, particularly over the Ohio and Lower Mississippi (Missouri and Upper Mississippi) river valleys where the Fall (Winter) dry season is arriving earlier (later). These findings have important implications for water management systems, soil moisture, agriculture and ecology, all of which are sensitive to the frequency and seasonality of precipitation.