Climate risks in Hungary over the coming decade

Models and their limits. Climate models are the best tools available, and the study relies on them throughout: the two domestic projection sets (HungaroMet’s four regional runs and the 28 bias-corrected regional runs of the FORESEE-HUN database) and decadal predictions downscaled to Hungary. That models capture as much as they do is a major achievement. Reality, however, is more complex than any model can represent: there are known technical limits, and there are factors on which we currently have no reliable knowledge. Where the observations already sit above the whole range of model runs, the study therefore does not plan from the model average.

The planning-minimum rule. Where domestic climate models fall short of measured data, the measured value serves as the starting point, while the lower limit of the model range represents the smallest conceivable value – not the best-case scenario. The rule was verified separately for each indicator and is applied only where it can be justified. For the duration of dry spells, the level of intense rainfall and the number of extremely hot days reaching 35 °C, the models do not fall short of the measured values, and the rule does not apply.

Uncertainty without probabilities. The most serious domestic outcomes are compound events – groundwater failing to recover after a dry winter, accompanied by summer heat and low river flows. Their joint probability cannot be reliably estimated from the available data, and because the hazards share a common cause, assuming independence would underestimate their joint occurrence. The findings therefore carry a confidence level, not a probability class.

Two ways to be wrong. One can prepare for something that does not happen, or fail to prepare for something that does. In a planning context, failing to prepare is the more costly mistake: the consequences of a missed investment span decades and are partly irreversible, while the cost of unnecessary preparation is one-off and visible. The authors state openly that weighing these two errors is a value judgement, not a scientific finding.

Confidence levels.

Data. The analysis uses data from HungaroMet’s open data portal, the National Water Authority’s data service, the National Adaptation Geoinformation System (NATéR) and the Copernicus Climate Change Service. Model-based estimates draw on the bias-corrected regional runs of the FORESEE-HUN database (Department of Meteorology, ELTE), on decadal predictions downscaled to Hungary by the National Observatory of Athens within the Augures project, and on the regional database of the IPCC Sixth Assessment Report. The calculations were made from the source data with AI support and were subjected to independent re-calculation, which does not necessarily reveal blind spots arising from a common methodological approach.

Not covered: estimates of drought damage to the national economy, return periods, the total water balance, the climatic attribution of individual events and an aggregate “cost of inaction”. The limits of the analysis are set out in Annex E of the full report.

English version. Courtesy translation of the Hungarian original, produced by machine translation (DeepL) and checked against the Hungarian text; not professionally edited. The Hungarian text is authoritative. Figures and maps carry Hungarian labels; their captions are in English. Hungarian originals: full report · short assessment.