The EUMETNET EUPP Benchmark

This website is under construction, more will come, be patient !

This website is the home of the EUPP Benchmark. The Benchmark is an important deliverable of the EUMETNET PP module, aiming to provide high-quality benchmark datasets and infrastructures to compare the various methods used to post-process the forecasts in National Meteorological Services. Indeed, the need to postprocess different types of forecasts is a challenges that all Weather Centers are currently facing, in order to provide consistent and reliable forecasts in a seamless fashion.

The format of the benchmark was discussed and drafted at the December 2019 PP module workshop. The main idea and goal emerging at this workshop was to try to automate it progressively as much as possible, with multiple different aspects of postprocessing being treated, in a modular fashion.

The first phase of the benchmark (2019-2023) focused on station-based postprocessing for station forecasts, using station observations and reforecasts at the nearest point.

The second phase (2024-2028) just started and will focus on gridded postprocessing for which gridded observations, forecasts and reforecasts are needed.

During both phases, datasets were/are created according to opendata standards to make the comparison of postprocessing methods possible. These datasets are shared freely with the community.

The activities of the benchmark are developed on GitHub: https://github.com/orgs/EUPP-benchmark/