Using Neural Networks to Model the Impacts of Climate Change on Water Supplies

Abstract

An artificial neural network methodology is developed to investigate the possible effects on monthly and seasonal surface water supplies in Colorado’s Arkansas River Basin under two transient climate change scenarios, the HAD from the Hadley Centre for Climate Prediction and Research and the CCC from the Canadian Climate Centre. The results show that the decade-to-decade variability is considerably more apparent than any long-term trend or change. Under the HAD scenario, water available for irrigation is expected to increase above the historical baseline in every month of the growing season. However, the CCC scenario predicts constant water shortages in the region and decreased water available for irrigation in almost every month. This wide variation in the predictions from the HAD and CCC scenarios means that there is a large degree of uncertainty on what the future impacts of climate change might be in the region. However, the methodology developed can be used to estimate the impacts of new or updated predictions of climate change.