Model-Data Comparisons of Pan-Continental Drought over North America during the Common Era

Abstract:

Pan-continental droughts, or droughts that simultaneously affect a large percentage of the geographically and climatically distinct regions of North America, present significant on-the-ground management challenges and, as such, are an important target for scientific research. There are, however, two fundamental reasons why a comprehensive characterization of pan-continental droughts, and their causes, proves challenging: 1) Regional hydroclimate in North America is characterized by distinct atmosphere-ocean dynamics; 2) The relative rarity of pan-continental drought and the short (~150 year) observational record mean that there are few events by which to diagnose how these relatively distinct regional hydroclimate dynamics can combine to produce pan-continental drought. The paradigm of model-data comparisons on paleoclimatic timescales is used herein to extend the North American drought record and to simulate potential atmosphere-ocean states during pan-continental droughts, which together provide a more comprehensive understanding of pan-continental drought dynamics. Specifically, six forced transient simulations of the last millennium from the CMIP5/PMIP3 archives are analyzed in conjunction with gridded tree-ring reconstructions of hydroclimate variability from the North American Drought Atlas (NADA). Models are found to simulate pan-continental drought with the frequency and spatial patterns exhibited by the NADA. They do not, however, agree on the modes of atmosphere-ocean variability that produce pan-continental droughts. This is because simulated ENSO, PDO and AMO dynamics, and their teleconnections to North America, are different between models and observations. Despite these dynamical differences, models are also able to reproduce large-magnitude centennial-scale variability in the frequency of pan-continental drought occurrence—an important feature of the paleoclimate record. These changes do not appear to be tied to exogenous forcings, suggesting that simulated internal hydroclimate variability on these timescales is large in magnitude. Results both clarify understanding of the dynamics that produce real-world pan-continental droughts, while assessing the ability of models to accurately characterize future drought risks.