Double Ensemble Estimates of Precipitation in the Southeastern United States for Extreme ENSO Events

Winter precipitation in the southeastern United States associated with extreme warm and cold ENSO events is examined through the use of synthetic Pacific Ocean sea surface temperature anomalies (SSTAs) and a global spectral model. A new double-ensemble technique is used to generate precipitation estimates that account for uncertainty in both the atmospheric initial conditions and sea surface temperature anomalies. The double-ensemble precipitation estimates are generated by running the FSU Global Spectral Model with an initial condition ensemble and a sea surface temperature boundary condition ensemble. The initial condition ensemble consists of ECMWF analyses from 10 days. The SSTA ensemble consists of Nov. through Feb. Pacific Ocean SSTs for four warm and four cold events. This results in a double ensemble consisting of 40 precipitation fields for each ENSO extreme. The resulting double-ensemble precipitation estimates are compared to both in- situ and satellite data. These precipitation estimates agree well with observations in both spatial pattern and magnitude. They indicate, in agreement with observations, that on average the southeastern United States receives more precipitation during ENSO warm event winters than during ENSO cold event winters. The double ensemble area average winter precipitation totals for the southeastern United States from the double-ensemble winter (December, January, and February) precipitation totals range from a maximum of 359 mm during an extreme warm event and a minimum of 239 mm during an extreme cold event, indicating a 50% increase in precipitation in the southeastern U.S. from an extreme cold event to an extreme warm event. There is considerable case to case variability within the ensemble, which supports the need for an ensemble approach. The spread in December-January-February total precipitation estimates and the uncertainty in the mean difference from warm to cold events is calculated and found to be sufficiently small over land, indicating excellent confidence in the double-ensemble precipitation estimates.