Evaluating Performance and Improvements of the ECMWF Integrated Forecasting System

The European Centre for Medium-Range Weather Forecasts (ECMWF) is a world leader in the production of global weather forecasts. ECMWF’s strategic goal is to provide operational ensemble-based analyses and predictions, using an integrated Earth system model, as consistent as possible between its various components and across time ranges. Currently, skill in medium-range weather prediction extends, on average, to about one week ahead. By 2025 ECMWF’s goal is to predict high-impact weather up to two weeks ahead and to predict large-scale patterns and regime transitions up to four weeks ahead.

ECMWF has developed a set of headline scores to monitor progress towards these strategic goals. These have been carefully chosen to monitor different time and space scales to address the multi-scale nature of the forecasting system. The scores measure both the probabilistic performance of the ensemble forecasts and the deterministic skill of the underlying model. Two recent additions to the set of headline scores provide an additional focus on the frequency of large forecast errors in the medium range, and on the skill in predicting large-scale temperature anomalies three weeks ahead.

The presentation will provide an overview of the performance of the ECMWF Integrated Forecasting System (IFS), demonstrating how the forecast skill has changed over recent years. We will show how the different scores highlight different aspects of forecast skill, and how together these are used to evaluate scientific developments including upgrades to data assimilation, model physics, resolution increases and ensemble configuration. The improvements from recent upgrades to the ECMWF IFS will be presented.