Optimizing Analog Ensembles for Sub-Daily Precipitation Forecasts

This study systematically explores existing and new optimization techniques for analog ensemble (AnEn) post-processing of hourly to daily precipitation forecasts over the complex terrain southwest British Columbia, Canada. An AnEn bias-corrects a target model forecast by searching past dates with similar (i.e., analogs), using verifying observations as members. The weather variables predictors) that select best analogs vary among stations seasons. First, different predictor selection are evaluated we propose an adjustment in forward procedure considerably improves computational efficiency while preserving skill. Second, temporal trends predictors used further enhance predictive skill, especially at shorter accumulation windows longer horizons. Finally, this introduces modification search allows within time window surrounding lead time. These supplemental times effectively expand training sample size, which significantly all performance metrics—even more than weighting temporal-trend steps combined. optimizes AnEns moderate intensities but also shows good median heavier rates. Precipitation is most challenging predict finer resolutions times, yet those see largest enhancement skill from post-processing. post-processing, including developed herein, can improve performance.