Empirical downscaling of high-resolution regional precipitation from large-scale reanalysis fields

This study describes an EOF-based technique for statistical downscaling of high-spatial-resolution monthlymean precipitation fromlarge-scale reanalysis circulation fields. Themethod is demonstrated and evaluated for fourwidely separated locations: the southeasternUnited States, the upperColoradoRiver basin, China's Jiangxi Province, and central Europe. For each location, the EOF-based downscalingmodels successfully reproduce the observed annual cycle while eliminating the biases seen in NCEP-NCAR reanalysis precipitation. They also frequently reproduce the monthly precipitation anomalies with greater fidelity than is seen in the precipitation field derived directly from reanalysis, and they outperform a suite of regional climate models over the two U.S. locations. With the relatively high skill achieved over a range of climate regimes, this technique may be a viable alternative to numerical downscaling of monthly-mean precipitation for many locations.