Two distinct methodologies for empirical prediction of precipitation on timescales of months to decades are discussed. In the first, the relationship between ENSO and wintertime rainfall is used to hindcast periods of meteorological drought and deficient reservoir inflow for Mexico's Rio Yaqui Basin. Although much more rain falls during the summer (JJAS) than during the winter (NDJFMA), wintertime rainfall is over three times as variable relative to the climatological mean and is strongly related to both reservoir inflow and ENSO. Analysis of historical and reconstructed rainfall data suggests that droughts as severe as the 1994--2002 Yaqui drought, usually associated with wintertime rainfall deficits, are not atypical. Using the wintertime ENSO-rainfall relationship, two simple empirical forecast models for possible use by water system managers are demonstrated. Next, a methodology for statistical downscaling of monthly mean precipitation from large scale reanalysis circulation fields is described and applied to five widely-separated locations. In each case, the EOF-based empirical downscaling models (EDMs) successfully reproduce the observed annual cycle while eliminating biases and spurious trends seen in reanalysis precipitation. In many cases, they also reproduce monthly precipitation anomalies with greater fidelity than either reanalysis or regional climate model (RCM) simulations. These EDMs are then applied to circulation fields from an ensemble of global climate models (GCMs) to obtain alternate projections of precipitation changes between the late 20th and late 21st centuries.
展开▼
