The project CARLOFFF studies heavy rainfall events that occur under convective atmospheric conditions, and derives future projections from various climate models and emission scenarios. Specifically, sub-daily convective environments from ERA5 are classified according to given, impact-relevant extremeness indices (such as the weather extremity index). The classifiers are chosen from a range of classical and novel methods, including deep learning, and are applied to an ensemble of CMIP6 simulations. Phase 2 of CARLOFFF puts an additional focus on regionalized projections for Germany along with corresponding uncertainty assessments. With regard to the latter, the full uncertainty of projected extremeness indices is disentangled by means of an ANOVA, to isolate structural, natural, and emission factors. .

Website: CARLOFF
Institution: University of Potsdam
Contact: Maik Heistermann, Gerd Bürger