Described is the impact of the Aral Sea desiccation on the local climate and the impact of its numerical prediction in the region. Presented is the analysis of two series of numerical experiments with the WRF-ARW model (numerical prediction and weather research) with the spatial resolution of 5 km and 28 σ-surfaces along the vertical up to the level of 50 hPa for the Aral Sea region. In the first series of forecasts for January and July 2009, underlying surface parameters from the MODIS database are used. In these series the sea surface mask corresponds to the Aral Sea configuration in the 1970s. In another group of experiments, the characteristics of the underlying surface of the Aral Sea area are replaced by the respective characteristics of the surrounding land. To study the effects of variations of surface properties, air temperature, humidity, cloudiness, precipitation, and wind are analyzed. It is demonstrated that if the Aral Sea is assumed to be absent in the model, this results in the significant strengthening of the continentality of regional climate and in the increase in the forecast skill scores. The supposition is made that the regular (at least each 3–5 years) update of global databases with the description of underlying surface properties due to the climate change can result in the considerable increase in the accuracy of numerical prediction and climate change modeling.