Predictability and Prediction of Multiyear to Decadal Droughts with a Hybrid Dynamical-Statistical System using CMIP5 Experiments with the MIROC5 Global Earth System Model

There is a substantial, emerging demand by decision makers for decadal scale drought information, especially in drought-prone regions such as Western U.S., Southern Africa, the Sahel, the Nile River Basin, and South Asia. We used four indices of decadal climate variability (DCV), hindcast and forecast by the MIROC5 Earth System Model from 1961 to 2020, in a statistical prediction system to assess multiyear to decadal predictability of the Self-Calibrating Palmer Drought Severity Index (SC-PDSI) from 1961 to 2010 and predict the index from 2011 to 2020 in the above-mentioned regions. The four major DCV phenomena are the Pacific Decadal Oscillation (PDO), the tropical Atlantic SST gradient, the West Pacific Warm Pool Variability, and the ENSO.

Our results indicate that the hybrid prediction system is able to hindcast slow evolution of SC-PDSI reasonably well in Southern Africa, Central India, and the U.S. Great Plains, especially from the 1980s to 2010. These results can be used to develop necessary adaptation measures for water, food, and energy security in the regions where the hybrid prediction system can predict multiyear to decadal droughts with significant skill. This oral presentation will show statistical associations between DCV phenomena and worldwide SC-PDSI, decadal hindcast skill of the DCV phenomena in the CMIP5 experiments with the MIROC5 ESM, and decadal hindcast skill of SC-PDSI. The presentation will also describe and discuss atmospheric teleconnections from the DCV phenomena in the Pacific and Atlantic Oceans to the regions where these DCV phenomena cause multiyear to decadal droughts.