Severe drought is a notable feature of the hydrology of central Southwest (CSW) Asia. Although studies have linked the region's interannual precipitation variability to remote forcings that include East Asia jet stream variability and western Pacific tropical convection, atmospheric general circulation models (GCMs) forced by observed sea-surface temperatures demonstrate little skill in simulating interannual precipitation variability in this region. Here, statistical methods of correcting systematic errors in GCM simulations of CSW Asia precipitation are investigated. Canonical correlation analysis is used to identify model fields related to observed precipitation anomaly patterns. These relationships are then used to predict observed precipitation anomalies. This approach is applied to the ECHAM 4.5 GCM using regional precipitation, upper-level winds and western Pacific tropical precipitation as predictors of observed CSW Asia precipitation anomalies. The statistical corrections improve the GCM precipitation simulations, resulting in modest, but statistically significant, cross-validated skill in simulating CSW Asia precipitation anomalies. Applying the procedure to hindcasts with persisted sea-surface temperatures gives lower, but statistically significant, precipitation correlations in the region along the Hindu Kush mountain range. Copyright (C) 2003 Royal Meteorological Society.
735LCTimes Cited:7Cited References Count:24