Seasonal Ensemble Forecast Post-processing

In many parts of the world, water resources systems manage sub-seasonal to seasonal (S2S) variability in climate and runoff in part through the use of operational streamflow forecasts, supplemented by predictions of climate and other hydrologic variables. S2S hydrologic forecasts are produced through both statistical and dynamical (model-based) approaches, and separate S2S forecasts may be combined in multi-model frameworks to increase their skill. Statistical post-processing can be used to enhance the skill and reliability of model-based S2S predictions, and to reduce bias, as well as to merge forecasts from multiple approaches. This chapter describes seasonal hydrologic forecast approaches and products, and presents common techniques used in both the post-processing of single ensemble forecast series as well as the combination of multiple forecasts. Also discussed are the sources of S2S hydrological predictability and particular challenges and opportunities related to post-processing seasonal hydrologic predictions, for which the sample sizes of past simulations, observations and predictions are relatively more limited than in the context of short to medium range prediction.