Search

The development of subseasonal forecasts has seen significant advancements, transforming our ability to predict weather patterns and climate variability on intermediate timescales ranging from 2 weeks to 2 months. Motivated by the need to enhance our understanding of subseasonal precipitation forecasts and their applicability to the hydrology forecast, this study retrospectively analysed precipitation ensemble forecasts from subseasonal prediction models in the Uruguay River basin nearby Salto Grande dam. Three models were considered: two from the S2S project (ECMWF and CNRM) and one from the SubX project (GEFS). Model forecasts were analysed on a weekly time scale using both deterministic and probabilistic approaches. Multimodel probabilistic forecasts combining the three different models were built to increase forecast skill. Individual models have a skill larger than or equal to the climatological forecast until 2 weeks in advance. Particularly, ECMWF shows better skill in both ensemble mean and probabilistic forecast. Multimodel probabilistic forecast improves the skill of the forecast throughout the year, with the skill even surpassing the climatological forecast by up to 4 weeks in advance during the summer. In addition, model skill was analysed considering the state of the El Niño–Southern Oscillation (ENSO) on a weekly and monthly basis. On weekly time scales the ENSO state modifies model skill differently depending on the sub-basin and season considered. However, the influence of ENSO on forecast skill is more clearly observed on monthly time scales, with largest improvement in the lower basin during springtime. The results of this work suggest that subseasonal models are a promising tool to bridge the gap between weather and climate forecast in the Uruguay River basin and have the potential to be utilized for hydrological forecasting in the study region., Peer Reviewed, Postprint (author's final draft)