Downscaling Climate Model Data for Energy and Crop Modelling Using Self-Organizing Maps

To make use of future climate projections, they must be actionable at the local level. Here, we develop daily temperature and precipitation climate scenarios for 379 weather stations across a study region centered on Illinois, United States. Self Organizing Maps (SOMs) are used to generate realizations of daily temperature and precipitation from each of 11 models from the Coupled Model Inter-Comparison Project 5 (CMIP5). We use this approach to generate daily data for historical (1976-2005), current (2006-2015), near future (2021-2050), and far future (2071-2100) periods, for low (RCP 2.6), moderate (RCP 4.5), and high (RCP 8.5) greenhouse gas emission scenarios. Estimated greenhouse gas emissions for the period 2005 through 2015 tracked closely with the RCP 8.5 scenario. In the far future period for the RCP 8.5 scenario, the downscaled projections show decreases in average precipitation up to 15% in the southern half of the study region, and increases of up to 15% in the northern half, with an average temperature increase of 2.7C across the region. The pattern of precipitation increasing in the north and decreasing in the south with and a general increase in temperature is found in all three scenarios, with smaller magnitudes of change in the lower emissions scenarios. In addition to the changes in average precipitation, we find an increase in the number of days with no rain, even in the areas with an increase in average precipitation. The SOM method is also used to examine changes in the synoptic patterns in the different scenarios across the region. By comparing the frequency of different synoptic patterns across the region, we can gain insight into the changes in weather patterns that are driving the variations in precipitation across the region.