Potential Bias in Projecting Future Regional Megadrought Risk: Insights From A Global Data-Model Framework

Abstract:

Megadrought is one of the most significant and costly climate extremes, and one that stakeholders (e.g., water and other resource managers) the world over wish to understand better; in particular, they need estimates of the risk of severe droughts as a function of drought frequency, severity, duration, and atmospheric greenhouse gas concentration. In many dry-climate regions of the globe, megadrought is synonymous with multi-decadal drought. However, in other regions, megadrought can be defined as extended drought, mostly not seen in the period of instrumental observations, and that would have large impacts if it were to occur in the future. New and published paleoclimatic observations allow us to understand the spectrum of drought in many regions of the globe; droughts exceeding 50 years have occurred in recent Earth history in southwestern North America, sub-Saharan Africa, the Mediterranean and Australia, whereas shorter megadroughts have occurred in Monsoon Asia, Amazonia and elsewhere. Data-model comparisons for regions with sufficiently long (e.g., 1000-2000 years) records of observed hydroclimatic variability suggest that state-of-the-art models can provide realistic estimates of interannual to decadal drought risk, but underestimate the risk of megadrought. Likely reasons for this shortcoming are the lack of sufficient multi-decadal variability in simulations of the past and future, plus an underappreciated understanding about how temperature variability and land-surface feedbacks interact with hydrological and ecological drought, as well as the roles played by unusually wet hydroclimatic extremes (e.g., ENSO related) in ending droughts of long duration. Paleoclimatic records also provide the opportunity to estimate how much models underestimate megadrought risk as a function of locale, frequency, severity, duration, and atmospheric greenhouse gas concentration; they also aid in providing stakeholders with bias-corrected estimates of megadrought risk.