An ensemble of data-driven weather prediction models for operational sub-seasonal forecasting

TL;DR

This paper introduces an operational multi-model ensemble system combining data-driven models with ECMWF's ocean model, achieving near-state-of-the-art sub-seasonal weather forecasts at 1-degree resolution for up to 4 weeks.

Contribution

It presents a novel ensemble system that integrates hybrid data-driven models with traditional models for improved sub-seasonal weather prediction.

Findings

Ensemble outperforms ECMWF raw ensemble by 4-17% for 2-meter temperature.

Bias correction reduces ECMWF ensemble advantage to 3%.

Achieves near-state-of-the-art forecasts with multi-model ensembling.

Abstract

We present an operations-ready multi-model ensemble weather forecasting system which uses hybrid data-driven weather prediction models coupled with the European Centre for Medium-range Weather Forecasts (ECMWF) ocean model to predict global weather at 1-degree resolution for 4 weeks of lead time. For predictions of 2-meter temperature, our ensemble on average outperforms the raw ECMWF extended-range ensemble by 4-17%, depending on the lead time. However, after applying statistical bias corrections, the ECMWF ensemble is about 3% better at 4 weeks. For other surface parameters, our ensemble is also within a few percentage points of ECMWF's ensemble. We demonstrate that it is possible to achieve near-state-of-the-art subseasonal-to-seasonal forecasts using a multi-model ensembling approach with data-driven weather prediction models.