A Bayes‐Based Approach Against Sample Imbalance to Improving the Potential Forecasts of Gale.

Sample imbalance prevents the Bayesian model from making effective potential forecasts of gale. An approach is thus proposed to modify the Bayesian model to deal with sample imbalance and verified using years (2015–2019) of reanalysis and radar data. The approach reduces sample imbalance by the resampling based on the value ranges of environmental parameters and the application of multi‐layer conditional probabilities. The samples that are insensitive to gale forecasts are excluded according to the value ranges of environmental parameters for gale occurrence. This resampling greatly improves gale occurrence hits of and suppresses its false alarms, leading to significant improvement in gale forecasting skill. On the basis of the resampling, the application of the multi‐layer conditional probabilities that forecast gale occurrence balances the samples to an equivalent magnitude. Consequently, the false alarms are further suppressed although some hits are reduced, resulting in a higher forecasting skill of gale. Plain Language Summary: Due to large sample imbalance, the Bayesian model with the basic probability is difficult to use effectively in the potential forecasts of gale. This study proposes an approach to modify the Bayesian model to handle sample imbalance, and verify the application of the modified Bayesian model in the potential forecasts of gale using years (2015–2019) of reanalysis and radar data. The approach deals with the sample imbalance problem through the resampling based on environmental parameters and the application of multiple layers of conditional probabilities. Excluding the samples that are insensitive to gale forecasts based on the value ranges of environmental parameters for gale occurrence significantly reduces sample imbalance. The resampling in this way significantly improves gale occurrence hits and suppresses its false alarms, leading to a great improvement in gale forecasting skill. On this basis, applying multiple layers of conditional probabilities that forecast gale occurrence further balances the samples to an equivalent magnitude. As a result, although the hits of gale occurrence are somewhat lowered, its false alarms are further reduced, leading to a higher forecasting skill of gale. Key Points: The Bayesian model with the basic probability tends to forecast gale non‐occurrence, resulting in few hits of gale occurrenceBased on the Bayesian model with no basic probability, resampling greatly promotes gale occurrence hits while suppressing its false alarmsAdditional application of multi‐layer conditional probabilities further suppresses gale occurrence false alarms while promoting its hits [ABSTRACT FROM AUTHOR]