Abstract

Global simulations of precipitation from climate models lack sufficient resolution and contain large biases that make them unsuitable for regional studies, such as forcing hydrologic simulations. In this study, the effectiveness of several methods to downscale large-scale precipitation is examined.

To facilitate comparisons with observations and to remove uncertainties in other fields, large-scale predictor fields to be downscaled are taken from the National Centers for Environmental Prediction–National Center for Atmospheric Research reanalyses. Three downscaling methods are used: (1): a local scaling of the simulated large-scale precipitation; (2) a modified scaling of simulated precipitation that takes into account the large-scale wind field; and (3) an analogue method with 1000 hPa heights as predictor. A hydrologic model of the Yakima River in central Washington state, USA, is then forced by the three downscaled precipitation datasets. Simulations with the raw large-scale precipitation and gridded observations are also made.

Comparisons among these simulated flows reveal the effectiveness of the downscaling methods. The local scaling of the simulated large-scale precipitation is shown to be quite successful and simple to implement. Furthermore, the tuning of the downscaling methods is valid across phases of the Pacific decadal oscillation, suggesting that the methods are applicable to climate-change studies.