Examining why trends in extreme precipitation should not be mistaken for trends in extreme discharge

English Abstract:

It is firmly established in the hydrologic literature that flooding is dependent on both antecedent moisture and precipitation. One could possibly phrase this relationship even more directly as “heavy precipitation does not necessarily lead to high stream discharge”, but studies are rarely formulated in such a way as to directly affirm this statement. However, we have observed a number of non-hydrologists mistake trends in heavy precipitation as a proxy for trends in flooding. If indeed the relationship between heavy precipitation and high discharge (or lack thereof) was more often explicitly presented, heavy precipitation would possibly be less often misinterpreted as a direct proxy for discharge. We undertake such an analysis by directly evaluating the frequency in which 99^th percentile precipitation events result in 99^th percentile discharge events for 300 watersheds spread across the contiguous U.S. We found that 99^th percentile precipitation only results in 99^th percentage discharge 30% of the time. However - not unexpectedly - when conditioned on soil moisture, 99^th percentile precipitation results in 99^th percentage discharge 51% of the time during wet periods and 8% of the time during dry periods. This study suggests that especially when one is interested in relating trends in heavy precipitation to hydrologic response, precipitation data should be segregated based on concurrent soil moisture. Taking such an approach for future climate predictions, we found that CMIP-5 GCM projections forecast an increase in concurrence of greater than median soil wetness and extreme precipitation in the northern half of the United States and a decrease in the southern half. This suggests that northern regions may see an increase in very high discharges while southern regions may see a decrease despite both regions having predicted increases in extreme precipitation.