Hydrology Models for Climate Change Assessment: Inter-decadal Climate Variability and Parameter Calibration

摘要

In this study, the calibration of hydrologic model parameters is investigated under different phases of low frequency climate variations, with a case study conducted in the Connecticut River Basin. "ABCDE style" hydrology models of a tributary on the Connecticut River are calibrated in a Bayesian framework over two different time periods dominated by dry and wet meteorological conditions, respectively. Uncertainty in parameter values is characterized by parameter distributions that are produced from the Bayesian calibration. T-tests and other statistical analyses are used to determine any differences in parameter distributions between the two separate calibration periods. For each calibration climate phase, parameter distributions are then used to predict flows in two other time periods, each also dominated by dry and wet climate conditions, respectively. The performances of both hydrology models in the two new verification periods are compared to determine whether the performance of a hydrology model depends on the climate phase during which it was calibrated. Results indicate whether the climate phase under which hydrology models are calibrated influence climate impact studies on hydrology and water resources.