Causes and Predictability of Drought in the US High Plains

Droughts in the High Plains, an important wheat-growing region in the United States, can exert substantial impacts on local agriculture and ecosystems. Taking the 2017 northern High Plains drought as an example, the exceptional dryness combined with the unusual heat that occurred over Montana and the Dakotas caused agricultural losses of $2.5 billion and contributed to one of Montana's worst wildfire seasons on record. Mitigating the societal impacts of such droughts requires an advanced scientific understanding of the processes that drive them, as well as a proper incorporation of these processes in forecast models in order to provide reliable drought prediction and early warning systems. This study investigates the causes of past High Plains droughts in the context of a warming climate and assesses drought prediction capability in the operational Subseasonal Prediction Experiment (SubX). The results are based on an integrated analysis of observations, reanalyses and an extensive suite of NASA GEOS modeling experiments, as well as a process-level investigation of drought predictability and prediction skill in SubX. We have found that dry conditions in the High Plains are often associated with a zonal wave train (of roughly wavenumber 5) in the Northern Hemisphere midlatitudes that resembles the leading patterns of upper-level circulation variability within the jet waveguide; the connection to SST is weak overall. Historical global warming led to no appreciable increase in the risk of precipitation deficits but an increased risk of heat waves in the region. The increased risk for heat waves may have increased the likelihood of agricultural (soil moisture) drought in the region, thereby exacerbating recent High Plains droughts. The prediction skill of High Plains droughts in SubX is overall modest; enhanced prediction skill can nevertheless be gained from forecasts of opportunity offered by recurring patterns of subseasonal atmospheric circulation variability. The capability of the SubX models in representing the observed drought-inducing processes will be discussed.