Downscaling of Seasonal Rainfall over the Philippines: Dynamical versus Statistical Approaches

Abstract

The additional value derived from a regional climate model (RCM) nested within general circulation model (GCM) seasonal simulations, over and above statistical methods of downscaling, is compared over the Philippines for the April–June monsoon transition season. Spatial interpolation of RCM and GCM gridbox values to station locations is compared with model output statistics (MOS) correction. The anomaly correlation coefficient (ACC) skill at the station scale of seasonal total rainfall is somewhat higher in the RCM compared to the GCM when using spatial interpolation. However, the ACC skills obtained using MOS of the GCM or RCM wind fields are shown to be generally—and rather equally—superior. The ranked probability skill scores (RPSS) are also generally much higher when using MOS, with slightly higher scores in the GCM case. Very high skills were found for MOS correction of daily rainfall frequency as a function of GCM and RCM seasonal-average low-level wind fields, but with no apparent advantage from the RCM. MOS-corrected monsoon onset dates often showed skill values similar to those of seasonal rainfall total, with good skill over the central Philippines. Finally, it is shown that the MOS skills decrease markedly and become inferior to those of spatial interpolation when the length of the 28-yr training set is halved. The results may be region dependent, and the excellent station data coverage and strong impact of ENSO on the Philippines may be factors contributing to the good MOS performance when using the full-length dataset over the Philippines.