Abstract:ECMWF model exhibited the best performance across all lead times. Multimodel forecasts exhibited better performance during some seasons. No systematic improvements are observed between ENSO‐neutral/active conditions but becomes more important on monthly scale. 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.

Real-Time Evaluation of the Uncertainty in Weather Forecasts Through Machine Learning-Based Models

A Learning Based Approach for Uncertainty Analysis in Numerical Weather Prediction Models

Predictive modeling of severe weather impact on individuals and populations using Machine Learning

Monthly Precipitation Forecasts Using Wavelet Neural Networks Models in a Semiarid Environment

Probabilistic radar rainfall nowcasts using empirical and theoretical uncertainty models

Predicting weather forecast uncertainty with machine learning

Real-time error correction for flood forecasting based on machine learning ensemble method and its uncertainty assessment

Measuring the Confidence of Traffic Forecasting Models: Techniques, Experimental Comparison and Guidelines towards Their Actionability

Measuring the Confidence of Single-Point Traffic Forecasting Models: Techniques, Experimental Comparison, and Guidelines Toward Their Actionability

From data to interpretable models: machine learning for soil moisture forecasting

Precipitation forecasting using random forest over an ecuadorian andes basin

Quantifying the uncertainty of precipitation forecasting

Quantification of predictive uncertainty in hydrological modelling by harnessing the wisdom of the crowd: A large-sample experiment at monthly timescale

Towards replacing precipitation ensemble predictions systems using machine learning

Improve short-term precipitation forecasts using numerical weather prediction model output and machine learning

Evaluation of subseasonal precipitation forecasts in the Uruguay River basin

Impact of Employing Weather Forecast Data as Input to the Estimation of Evapotranspiration by Deep Neural Network Models

Improving forecasts of precipitation extremes over Northern and Central Italy using machine learning

Online machine-learning forecast uncertainty estimation for sequential data assimilation

Probabilistic predictions for meteorological droughts based on multi-initial conditions

Flash floods in Mediterranean catchments: a meta-model decision support system based on Bayesian networks