Maximizing the forecasting skill of an ensemble model

SUMMARY An ensemble model integrates forecasts of different models (or parametrizations the same model) into one single forecast. This procedure has names in literature and is approached through philosophies theory practice. Previous approaches often weighted equally or according to their individual skill. Here we present a more meaningful strategy by obtaining weights that maximize skill ensemble. The based on multivariate logistic regression exposes some level flexibility emphasize aspects seismicity address end users. We apply operational earthquake forecasting system Italy demonstrate its superior over best forecast with statistical significance. In particular, highlight improves when exploiting fitting ensemble, for example using only recent not entire historical data.