Characteristics, precursors, and potential predictability of Amu Darya Drought in an Earth system model large ensemble

The socioeconomic stability of the Central Asian Republics in the Amu Darya watershed is sensitive to drought. Activities related to agriculture employ a large fraction of the population and are responsible for at least one-fifth of the gross domestic products of Afghanistan, Tajikistan, and Turkmenistan. Toward building a predictive understanding that may be applied to drought early warning practices, the characteristics, precursors and potential predictability of agricultural drought in the Amu Darya watershed are examined in a large ensemble of Community Earth System Model version 1 simulations during 1920-2019. Agricultural drought is examined over Upper and Lower regions of the Amu Darya watershed, which have different mean hydroclimates, and is defined by 1-m soil moisture deficits lasting three or more months. The likelihood of drought onset and demise is phase-locked with the seasonal cycle of precipitation of each region, but with some notable differences. For the Upper region, drought onset and demise are three times more likely to occur during Autumn and Spring than other seasons. For the Lower region, drought onset and demise are three times more likely to occur during November-April than during Summer. Precipitation anomalies drive drought onset and demise during the climatological wet periods of both regions while temperatures play a smaller role. The probability of drought onset and demise is modulated by La Nina and El Nino, which control the interannual variability of precipitation over the Central Asian Republics during their wet seasons, indicating that the state of the El Nino Southern Oscillation serves as a key predictor of agricultural drought phase changes.