Improving S2S Predictibility of Hydrological Quantities Over California with Novel Data-Based Modelling

Improving Subseasonal to Seasonal (S2S) predictability of hydrological quantities is important for a variety of applications, such as agriculture, water resources, and emergency management. Despite its importance, prediction at S2S time scales is much less skillful and under-developed than that at weather timescales. For example, in 2023, S2S forecasting failed to predict the historically snowy winter in California. Here, we provide a comprehensive analysis of the predictability of key hydrological variables—snow water equivalent, evapotranspiration, soil moisture, runoff, temperature, and precipitation for river basins in the southwestern United States, focusing on the California region. The analysis utilizes a variety of predictor variables, including basin-average and pixel-based hydrological data, global (ocean + land) surface temperature, and teleconnection indices, to develop seasonal forecasts with lead times ranging from 0 to 5 months.

Results indicate that hydrological quantities, such as soil moisture, runoff, and SWE are more predictable than temperature and precipitation, especially during late spring and summer. These quantities show a high degree of persistence. However, a model that combines all the predictors outperforms persistence models, especially at longer lead times, highlighting the added value of incorporating diverse data sources. Furthermore, S2S predictions using these models can be better than those produced with global seasonal forecast systems, such as CFSv2 and SEAS5. While the models were evaluated with cross validation, hindcast verification, conducted for the period 2000-2022, corroborated the robustness of our models, with only marginal declines in skill compared to cross-validated forecasts. Overall, the findings underscore the value of incorporating a range of local and global predictors in seasonal hydrological forecasting in places like California. The developed models offer potential for enhancing water resource management and planning in these and similar regions.