Improving seasonal forecasts through the state of multiple large-scale climate signals to inform water management

Increasingly variable hydrologic regimes combined with more frequent and intense extreme events are challenging water management, emphasizing the need for accurate medium- to long-term predictions to timely prompt anticipatory operations. Modern forecasts are becoming increasingly skillful over short lead times, but predictability generally decreases at longer lead times. Global climate teleconnections, such as El Niño Southern Oscillation (ENSO) and North Atlantic Oscillation (NAO), may contribute to extending forecast lead times. However, the contribution of ENSO states to local predictability depends on the degree to which local conditions are affected by this climate state. This teleconnection is well defined in some locations, such as Australia or Chile, while there is no consensus on how it can be detected and used in other regions, like Europe or Africa. In this work, we contribute a general framework that builds on the Nino Index Phase Analysis to capture the state of two large scale climate signals, namely ENSO and NAO. We use these teleconnections to forecast local hydroclimatic variables on a seasonal time scale. For each phase of the considered climate signals, our approach identifies relevant anomalies in pre-season Sea Surface Temperature that influence the local hydrologic conditions, which are first aggregated via Principal Component Analysis, and then used as inputs in a multivariate nonlinear forecast model of seasonal precipitation. The resulting seasonal meteorological forecasts are then transformed into streamflow predictions. Finally, these hydrological forecasts are used to inform water system operations. The framework is demonstrated through an application to the Lake Como basin, Italy, a regulated lake mainly operated for flood control and irrigation supply. Results show the existence of a never-proven-before high correlation between seasonal SST values and one season-ahead precipitation. Precipitation and streamflow forecast build on this correlation are then used for informing lake operation, proving to be an asset to anticipate extreme weather events, including the accurately predicted 2003 and 2005 droughts. The lake regulation benefits from such information by timely activating hedging strategies which ultimately result in more reliable water supply operations.