Comparison of Correction Methods for Seasonal Forecasts of Temperature over Central Europe

The direct applicability of seasonal forecasts is limited by their coarse spatial resolution, an issue that is particularly visible in mountainous regions. Therefore, post-processing procedures are required to improve forecast quality and obtain results suitable for regional-scale applications.

In this study, we compare two correction methods for improving seasonal forecasts of 2-meter air temperature (T2m): quantile mapping and vertical temperature correction using a lapse-rate approach. We use seasonal forecast outputs from the ECMWF model provided by the Copernicus Climate Change Service (C3S), with the domain restricted to Central Europe and centred over Poland (13–26°E, 47.5–55°N).

ERA5 reanalysis data were used for a 10-year training period in the quantile mapping procedure, which is based on non-parametric, robust empirical quantiles and applied independently at each grid point. In parallel, a simple physically based correction incorporating vertical temperature lapse rates was evaluated.

Forecast performance was assessed for selected months. Preliminary results indicate that the lapse-rate-based correction outperforms quantile mapping in reproducing local temperature patterns over the study area.