GC13G-0751:
Global sensitivity analysis of CMIP5 predictions of future changes in precipitation, reference evapotranspiration and drought index (SPEI) over the U.S.
Monday, 15 December 2014
Seungwoo Jason Chang1, Wendy D Graham1 and Syewoon Hwang2, (1)University of Florida, Gainesville, FL, United States, (2)Gyeongsang National University, Jinju, South Korea
Abstract:
Projecting evapotranspiration for estimating future agricultural irrigation demand is uncertain because estimates of future precipitation and evapotranspiration vary significantly depending on the Global Climate Model (GCM), future RCP emission scenario and reference evapotranspiration (RET) estimation method selected. Understanding the relative contributions of these various sources of uncertainty is important for effective long-term water resource planning. In this study variance-based sensitivity analysis (Saltelli et al., 2010) was used to assess the sensitivity of estimated future changes in precipitation, RET and Standardized Precipitation Evapotranspiration Index drought index (SPEI) to 9 GCMs, 3 RCP scenarios, and 11 ET estimation methods over 9 regions of the United States for two future periods: 2030-2060 and 2070-2100. Future changes in precipitation were found to be most sensitive to GCM selection for all U.S. regions and both future periods. For projecting changes in future RET and SPEI, the selection of ET method and GCM were more sensitive than the selection of RCP scenario. In general, changes in ET and SPEI were most sensitive to ET estimation methods in the cold season and to GCM selection in the warm season; however sensitivities differed by region, season and future period. This study underscores the importance of evaluating projections of future agricultural irrigation demand for an ensemble of GCMs and ET estimation methods rather than relying on few GCMs and a single ET estimation method.
