{"id":"arxiv/2608.20017","url":"https://arxiv.org/abs/2608.20017","tags":["nlin.CD","math.DS"],"score":0,"title":"Understanding the superiority of multi-model ensemble forecasts through reservoir computing","domain":"nlin.CD","source":"arxiv","authors":["Daniel Estevez Moya","Francesco Martinuzzi","Edmilson Roque dos Santos","Erick Alejandro Madrigal Solis","Ernesto Estevez Rams","Holger Kantz"],"reviews":0,"subject":"nlin.CD","summary":"Weather forecasting and climate projection frequently use multi-model ensembles (MMEs) to improve short-term forecasts by averaging across models. However, this practice is often not well justified or validated. Using reservoir computing (RC) as a computationally efficient alternative to large-scale physical models, we assess the validity of the MME approach for chaotic time series. By training multiple randomly constructed RCs on the same dataset, we create a multi-model ensemble in which each model has its own unique error. These model errors lead to very different forecasting…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-08-20","reactions":{}}