Seasonal Predictability of Precipitation Associated with Atmospheric Rivers in Western North America

This work aims to evaluate the seasonal predictability of precipitation, particularly that due to atmospheric rivers (ARs), by utilizing the empirical linkages between Pacific sea surface temperature (SST) and seasonal precipitation (P) in western North America. We identify temporally coupled and physically meaningful spatial modes in SST and P fields, at appropriate lead times, via Canonical Correlation Analysis (CCA) and, after considering their physical significance, use these coupled modes to construct a statistical forecasting model. We apply the model for forecasting seasonal precipitation accumulations associated with atmospheric river activity during early (October - December) and late (January - March) AR seasons. The best model skill is obtained for forecasting the late cool season precipitation accumulations particularly over southern California. Significant prediction skills can be achieved using antecedent SST fields at lead-times up to two months, i.e. using November SST to predict JFM precipitation. Much of this predictability is due to slowly evolving and/or persistent SST modes in the tropical and north Pacific Ocean (ENSO and PDO) as well as large regional SST anomalies in the north-eastern Pacific, a particularly active region in recent years. In the context of seasonal predictability, we will assess and discuss the spatial-orographic patterns of precipitation determined by the variable magnitude and direction of vapor transport in land-falling atmospheric rivers.