Scripps Institution of Oceanography

Forecasting Extreme Events

Statistical Forecast of the Extreme Precipitation and Temperature Event Frequencies for December-January-February 1997/98.

The categorical ENSO-phase-based statistical forecasting scheme uses the hybrid coupled model prediction of ENSO phase and intensity defined by NINO 3.4 SST and also takes into account the current phase of the north Pacific decadal oscillation. Results are based on 60 winters (1934-1993) of daily precipitation and temperature data recorded at 168 US stations. Forecasts are issued in three categories (low, average and high) defined by the terciles of the 60-winter climatology of heavy rainfall frequency (HRF), extreme warm (EWF) and extreme cold (ECF) outbreak frequencies. A heavy rainfall event is defined by daily precipitation totaling more than the 75th percentile of the local DJF daily rainfall climatology. Extreme warm (cold) events are defined by daily average temperature above (below) the 90th (10th) percentile of the DJF daily rainfall climatology at each station. The forecasted category is the one containing the mean of a variable (HRF, EWF or ECF) for past winters characterized by similar climatic conditions in the Pacific to those predicted by coupled models and persistence for the coming winter. Skill is assessed via cross-validation techniques and statistical significance is estimated via bootstrap resampling. Only those predictions in the low and high categories with skill levels significant at the 0.05 level of significance are displayed.

Last modified: 24 July 1997