Evaluating Empirically Derived Dew Point Estimation Methods to asses the Effect of Irrigation on Climate in the Southeastern United States

This research investigates the spatial and temporal variation of irrigation induced climate change in the Southeastern United States (SE US). Past research has postulated that increases in dew point temperatures are related to increased intensity of irrigated agricultural landscapes. There is little research on this topic in the Southeast as much of the literature on this topic focuses geographically on California and the Great Plains. Dew point temperature is a vital meteorological measurement that indicates the moisture content of the air but is often not readily available outside of first-order observing stations. As previously indicated, changes in dew point provide an indication of irrigation modified climate. This work first builds a dew point temperature record in a data poor area using regression based and neural network approaches that estimate daily dew point temperatures with daily minimum and maximum temperatures as well as precipitation as its inputs. Data from seven sites within the Georgia Automated Environmental Monitoring Network (GAEMN) were used in the training of the regression based and neural network methods. The two methods (regression based and neural network) are evaluated based on a suite of model evaluation statistics and the method with the best performance is employed to produce a time-series of dew point temperatures in the region from independent data. These time-series are then analyzed to determine any irrigation related changes in dew point. It is expected that the increased intensity of irrigation in the region has led to measurable changes in dew point temperatures.