72 - Using GCMs to Project Long-Term Rainfall Deficits

Global Circulation Models (GCMs) have become well accepted tools for projecting rainfall event statistics, as well as monthly and annual rainfall probabilities. What they are generally not trained to do, however, is project long-term (multi-year) persistence of dry conditions, or serially correlated deficits in rainfall. Even without the documented ability to project long-term precipitation deficits, many water suppliers still use GCMs for future drought planning, though the validity of the GCMs can only be expressed as their ability to replicate precipitation probability in any given month, which does not translate into confident predictions of accumulating deficits over long time periods. As part of an internal research and development project, CDM Smith focused on further understanding historic and future drought for water suppliers by developing four simple but innovative statistical tests on GCM ability to reproduce and project cumulative rainfall deficits up to 5 years on a monthly scale. This analysis only considers precipitation at this time and not temperature impacts, as drought forecasting is often hindered by the combined probability of too many factors, resulting in indices which are often unitless and qualitative. Our work focused on understanding the fundamental root cause of droughts, lack of precipitation, more quantitatively through the calculation of cumulative precipitation deficit on a monthly basis for a deficit duration of up to 5 years for historical and future conditions. The final deficit is given in inches of cumulative rainfall below the historical cumulative monthly average. This project also validated if GCMs could replicate historic sustained periods of precipitation deficit. The methodology of calculating historic and future cumulative precipitation deficits and validating the GCMs ability to replicate historic dry periods was applied to 22 locations within the continental United States. The analysis of these 22 locations showed there were four types of results: locations where GCMs can be used to quantitatively assess dry periods, where GCMs can be used semi quantitatively, where GCMs can be used qualitatively, and where the use of GCMs for replicating dry periods need to be further investigated. Approximately 75% of sites tested indicated that GCMs can indeed provide useful insight into future drought frequency, intensity, and duration potential. Other sites suggested that additional work is needed to develop subsets of GCMs that are more capable in aggregate of simulating long-term dry conditions. Additional next steps include testing the method at additional sites both in the US and internationally.