ECMWF opens 9-kilometre forecast data for localized applications

The European Centre for Medium-Range Weather Forecasts (ECMWF) has made a subset of its highest-resolution 9-kilometre forecast data free and open, expanding access to numerical weather predictions with finer spatial detail than ECMWF’s 25-kilometre products. The October 2026 announcement is a data-access measure rather than a new evaluation study: it does not report a quantified improvement in forecast skill or establish outcomes for individual forecast applications. The open subset is drawn from ECMWF’s full Real-time Catalogue and is intended to enable users to experiment, develop products and build services.

ECMWF says the finer grid can distinguish localized terrain features that are combined at coarser spacing. For example, a 25-kilometre forecast may indicate rain across a broad area without establishing which valley will receive it, whereas 9-kilometre data can represent individual slopes and river basins. In flood forecasting, ECMWF says this additional detail can help hydrological models identify where water may accumulate, including particular locations relevant to warnings for individual towns.

The announcement describes this as a potential benefit of higher resolution and does not provide a controlled comparison showing that every resulting flood forecast is more accurate. The same terrain-level distinction may also support localized assessments of malaria and dengue risk. According to ECMWF, transmission switches on above a temperature threshold, and nearby highland and lowland areas can fall on opposite sides of that threshold.

At 25-kilometre resolution, a cool highland and warmer valleys below it can be collapsed into a single average. The 9-kilometre grid separates cool ridges from warm lowlands more clearly, revealing where modeled temperatures enter the range associated with disease transmission. Health ministries and their partners can then connect forecasts with conditions in particular places.

These examples illustrate possible uses of the data rather than measured changes in disease surveillance or public-health outcomes. Operational constraints remain. ECMWF produces about 100 terabytes of data per day, comprising more than 450 parameters, and describes these large, complex and continuously updated datasets as a major delivery challenge.

The free 9-kilometre subset is therefore released with a two-hour delay rather than in real time. ECMWF says this interval supports reliable and timely delivery across the network while allowing it to maintain commitments to its 35 Member and Co-operating States. Organisations that require the 9-kilometre data in real time can obtain them through a service agreement, while the 25-kilometre data remain freely available in real time to all users.

Users must attribute the data to ECMWF. The release extends an existing open-data step announced a year earlier, when ECMWF made its 25-kilometre resolution data free. That product has already been used by organizations including the World Food Programme, which reported using the data to support timely assistance ahead of climate hazards, including severe droughts in Mozambique in 2024.

The new announcement positions the 9-kilometre subset as a further step toward more detailed forecasts for flood warnings, health applications and early-warning systems. ECMWF Director-General Florian Pappenberger said broader access to the data could help national meteorological and hydrological services enhance forecasting capabilities and support early-warning systems.

Subject of Research: Free access to ECMWF’s 9-kilometre numerical weather prediction data for localized forecasting applications

Article Title: 9 km free and open data offers a step change in localized forecasts

Article References: 9 km free and open data offers a step change in localized forecasts. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: ECMWF, 9-kilometre forecasts, open weather data, numerical weather prediction, localized forecasting, flood forecasting, hydrological modelling, malaria risk, dengue risk, early warning systems, free, open