A Subseasonal Forecast Index Focused on the Risk of Extremes

Subseasonal to seasonal (S2S) forecasting of extreme weather events is a topic of great interest and strong demand in many weather-sensitive sectors of business and industry, because these “infrequent events can greatly disrupt society’s functioning and are therefore of significant concern to many users” (National Academy of Sciences 2016). Many of Prescient Weather’s World Climate Service customers face their most critical challenges when weather extremes threaten, but these occasions also provide the most profound opportunities for risk mitigation and, therefore, the greatest potential for competitive advantage.

In recent years the short‐ to medium‐range forecasting community has become familiar with the Extreme Forecast Index (EFI) and the Shift of Tails (SOT) index that were developed by ECMWF to help assess the risk of extreme events (Lalaurette 2003, Zsoter 2006). The key advantage of these tools is that they compare the realtime forecast ensemble distribution to the climatological distribution of the model history, with an emphasis on the tails of the distributions. This approach allows for a standardized index that reveals the significance of the ensemble shift relative to the local climate.

However, the EFI and SOT metrics are best suited to short- to medium-range forecasts and are less useful for S2S time scales. For example, the EFI and SOT values are dependent on lead time and rarely show large signals in the mid-latitudes at lead times beyond two weeks; the SOT index is nearly always negative for the subseasonal time frame. Moreover, the index definitions do not include verification data and do not take account of the skill of the forecasts; this is not a major shortcoming for short-range forecasts that are usually very skillful, but it limits the usefulness of the approach for long-lead forecasts that often have marginal skill.

To address the need for improved long-lead guidance for extremes, we have developed a new Extreme Potential Index (EPI) for subseasonal forecasts. The EPI definition is similar to the SOT index, but it also depends on the skill of the model in predicting major extremes in the model forecast history. As a consequence, the index takes its largest values where both the realtime forecast is suggesting an extreme and the model has been successful in anticipating observed extremes at the same lead time in the past.

Calculating the EPI is a major data processing task that involves both the reforecast history and the corresponding verification, but the result is a powerful tool that identifies meaningful extreme-focused signals in the long-lead model ensemble forecasts. We discuss the EPI definition, its advantages for S2S time scales, and examples of its utility in realtime forecasting.

Acknowledgments

Prescient Weather Ltd research reported here was supported by the National Oceanic and Atmospheric Agency with Contract WC-133R-16-CN-0103.

References

Lalaurette, F. (2003), Early detection of abnormal weather conditions using a probabilistic extreme forecast index. Q.J.R. Meteorol. Soc., 129: 3037-3057. https://doi.org/10.1256/qj.02.152

National Academies of Sciences, Engineering, and Medicine. 2016. Next Generation Earth System Prediction: Strategies for Subseasonal to Seasonal Forecasts. Washington, DC: The National Academies Press. https://doi.org/10.17226/21873

Zsoter, E., 2006: Recent developments in extreme weather forecasting. ECMWF Newsletter, No. 107, ECMWF, Reading, United Kingdom, 8–17.