Item

Abstract

We develop a framework for assessing the drivers and impacts of drought events, built upon a Markov Random Field-based drought identification method. The method uses a precipitation threshold for drought, while also considering the drought state of neighboring grid points, and identifies contiguous and distinct droughts that propagate through space and time. We apply it to North American precipitation from observations and a multi-model ensemble of 67 historical simulations to produce a repository of 25,156 identified droughts. The framework uses an observed drought for comparison and we choose the 2011-2016 drought in the western United States, which is among the most severe and persistent droughts in recorded history. Simulated droughts are ranked by how well their location, size, and spatiotemporal characteristics match those of the observed drought. As we narrow these matching constraints, we quantify if the local-scale hydrological and ecological impacts (evaporation, leaf area index, soil moisture, and runoff) and large-scale drivers (atmospheric circulation, SST, ENSO, AMO, PDO) of the simulated droughts also narrow. Our findings suggest that ecological impacts are not predictable even for simulated droughts that closely match the spatiotemporal characteristics of the observed drought. The drought drivers are also not predictable, as simulated modes of climate variability do not increasingly correlate with observed modes as we narrow the matching constraints. Together these results suggest that drought drivers and some drought impacts have limited predictability, even for the most persistent and severe of droughts, although additional work is needed to quantify the role for structural uncertainty.