Uncertainty Analysis of Drought in the Community Atmosphere Model

Over the 21^st century, anthropogenic climate change is expected to increase the occurrence of extreme droughts, leading to significant socio-economic impacts in many regions of the world. The precursors of drought are diverse and still poorly understood. Climate models provide useful tools for studying the physical links between precursor “drought-conducive” climatic states and the characteristics of drought, such as severity, duration, spatial extent and frequency. There are, however, still large uncertainties in model portrayal of the physical processes governing drought behavior. Reducing these uncertainties is a necessary but not sufficient condition for enhancing confidence in model projections of 21^st century changes in drought. Here, we analyze a large (1300-member) perturbed physics ensemble performed with the Community Atmospheric Model (CAM4) to gain insights into causes of the 1998-2002 North American drought. The ensemble was constructed by varying the values of input parameters related to clouds, precipitation, convection, and the boundary layer over allowable ranges of uncertainty. Changes in these parameters can alter the width of the Hadley circulation, strengthen deep convection, and perturb land-atmospheric coupling, leading to different predictions of temperature, precipitation and soil moisture, and, in turn, very different simulations of drought properties. We perform a sensitivity analysis to identify the key parameters influencing drought-related metrics. Using observations and a Bayesian statistical framework, we identify parameter values that yield a better fit to data, thereby identifying configurations that may be more successful in simulating key features of observed drought behavior. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344, and is released as LLNL-ABS675834. It is supported by the Early Career Research Program award to Celine Bonfils.