Abstract ATM-9632304 Sloan, Lisa C. University of California, Santa Cruz Title: Modeling Anthropogenic Desiccation of the Aral Sea: A Unique Test of the Predictive Capabilities of a Regional Earth System Model The utility of global and regional earth system models (ESMs) for investigating causes of past climate change, and predicting impacts of future human activities, depends on how accurately changes driven by specific perturbations are predicted. This accuracy is controlled primarily by how well feedbacks are simulated. The accuracy of ESM-predicted changes is unknown, because previous model validation has been limited to assessing how well present conditions are reproduced or comparing paleoclimate model results with sparse and potentially ambiguous proxy data. In order to conduct a more thorough evaluation of ESM performance, model- predicted and observed changes resulting from known perturbations over historical timescales will be compared. In this study, anthropogenic desiccation of the Aral Sea will be used as a test of the predictive accuracy of a regional ESM (the RegCM2 and lake model associated with the National Center for Atmospheric Research) that includes an interactively coupled atmosphere and lake. Desiccation of the Aral Sea since 1960, resulting from diversion of water for intensive irrigation in the surrounding basin, has been extensive enough to produce a regional climatic response. Desiccation of this magnitude should also produce a model response. Importantly. meteorological and hydrologic observations that span the interval of desiccation are available so the actual changes can be quantified. The predictive capabilities of the model will be tested by comparing model-simulated changes with observed changes that have occurred between 1960 and 1995. The work consists of three components. First, observations will be used to quantify regional climate and lake hydrologic changes that are likely due to Aral Sea desiccation. These obser ved changes will be used to (i) test model predictions, (ii) directly measure how the presence of the Aral Sea alters regional climate, (iii) directly measure how lake-level variations have impacted the Aral Sea's hydrology, and (iv) quantify regional-scale human impacts. Second, the interactively coupled Reg CM2-lake model's strengths and weaknesses in predicting changes will be assessed, by comparing simulated and observed desiccation-induced regional climate and lake hydrologic changes. In order to gain a more thorough understanding of the causes for the coupled model's successes and failures, the predictive accuracy of independent (non-interactive) versions of RegCM2 and the lake model, driven by meteorological or hydrologic observations, will also be evaluated. Third, the regional climate effects generated by the Aral for seas of different surface area, depth and salinity will be calculated. This will allow: (i) quantification of lake-atmosphere interactions and feedbacks associated with lake-level variations; (ii) prediction of future climate changes driven by further anthropogenic Arals Sea desiccation; and (iii) assessment of how much of the model predicted change between pre- desiccation and -present and present intervals was due to large-scale meteorological variability. Predictions of future climate changes resulting from additional desiccation will be prepared and evaluated in terms of the model's strengths and weaknesses as determined. This research is important because it seeks to test how well coupled regional climate lake models predict the impact of human activities on a large shallow terminal lake and how that feedbacks on the regional and global climate.
