New Publication | Natural Hazards and Earth System Sciences
What would it take for compound event science to actually be used?
Climate extremes rarely arrive alone. A drought that primes a landscape for wildfire, a heatwave that coincides with a windless spell across an entire power grid, a second flood that hits a region before it has recovered from the first: such compound events can cause damage far beyond the sum of their parts. Most risk assessments, however, still treat hazards as if they occurred independently and therefore systematically underestimate them. In an invited perspective in Natural Hazards and Earth System Sciences, 30 researchers led by Kai Kornhuber (IIASA and Columbia University), among them Benjamin Poschlod (TUHH, IGWS), turn the usual question around: they ask not what is still unknown about compound events, but what would make the knowledge already at hand usable for the people who have to act on it.
The paper works through four pillars of the field and names the bottleneck in each. Impact data, the empirical link between events and damages, are frequently incomplete, commercially restricted, or filed under a single hazard, so that the wind, rainfall and storm surge damages of one cyclone are never connected; only about 19 % of the disasters recorded in the widely used EM-DAT database are currently recognisable as multi-hazard events. Early warning systems still largely warn hazard by hazard, which in one 2021 case in the southeastern United States meant advising people to seek shelter in basements against a tornado while flash flooding was forecast at the same time. Climate models face a trade-off between the fine resolution needed to resolve local processes and the large ensembles needed to sample rare hazard combinations, a section co-authored by Poschlod that points to high-resolution large ensembles and storyline approaches as pragmatic ways forward. And some uncertainty cannot be reduced at all: because of internal climate variability, the frequency of three consecutive dry years in central North America differs by more than a factor of ten between equally plausible climate realizations.
From this, the authors distil four guiding principles for the field: climate risk information should be provided free of charge and presented intuitively, be detailed enough in space and time for the application at hand, be delivered as early as possible even when imperfect, and be transparent about its assumptions, limitations and underlying data. Their central message is that usability is not a communication exercise added at the end of a project, but a design decision that belongs at its beginning.
Read the full open-access article: Kornhuber, K., Bevacqua, E., Madruga de Brito, M., Jäger, W. S., Rivoire, P., Rogers, C. D. W., Banfi, F., Batibeniz, F., Carruthers, J., De Michele, C., De Angeli, S., Deidda, C., de Ruiter, M. C., Fink, A. H., Goulart, H. M. D., Küpfer, K., Ludwig, P., Maraun, D., Messori, G., Nath, S., O’Loughlin, F., Pinto, J. G., Poschlod, B., Ramos, A. M., Raymond, C., Ribeiro, A. F. S., Singh, D., Suarez Gutierrez, L., Ward, P. J., and White, C. J. (2026): Invited perspectives: Towards usable compound event research. Natural Hazards and Earth System Sciences 26, 4309–4334. https://doi.org/10.5194/nhess-26-4309-2026
