Global calibration of regional tropical cyclone impact functions

Spatially explicit weather and climate risk assessments utilize impact functions (IFs) to translate hazard intensity into economic impact. However, global scale risk assessments are often lacking regionally calibrated IFs. In the case of tropical cyclones (TCs), IFs calibrated for the US have thus so far been applied for assessments of direct economic risk in other parts of the world. For example, in industrialized East Asian countries, where many TCs make landfall, these IFs lead to modelled damages orders of magnitude larger than reported. To improve the global representation of TC vulnerability, we calibrate regional IFs in a risk modelling framework with TC hazard intensity represented by high resolution maximum sustained wind speed. The calibration is based on track data and reported economic damages for 424 TCs making landfall on 62 countries worldwide, accounting for 76% of normalized reported damages from 1980 to 2017.

With our calibration we identify idealized IFs for eight world regions. These eight idealized IFs are based on two complementary optimization functions, one focusing on the deviation per event and the other on total damage per region. The IFs from the two approaches agree well for North America but deviate for other world regions. This reflects large uncertainties in model setup and input data. Sources of these uncertainties can be (i) ocean basin specific TC characteristics, (ii) the simplified representation of TC hazard intensity by wind speed alone, (iii) the assumed inventory of asset exposure per country, (iv) adaptation and development of the built environment over time, and (v) large uncertainties and biases in the reported damage data. In addition to the optimization, we computed best fitting IF steepness for each single TC event. Best fitting IF steepness shows a wide spread within each region, illustrating the model setup’s limitations when it comes to simulating the precise impact of single events. The spread in IF steepness can thus be used to inform probabilistic TC impact modelling beyond the use of a single deterministic IF, implicitly accounting for the uncertainties in model and input data.

Despite their limitations, the regionally calibrated IFs represent an improvement compared to globally applying IFs calibrated for the US, allowing for global scale TC risk assessments. The entire model setup is based on open data and made publicly available as part of TC module of the free, open-source natural catastrophe impact modelling project CLIMADA (https://github.com/CLIMADA-project/climada_python).

Future research should focus on a more adequate representation of TC hazard as a combination of wind, storm surge and torrential rain. In order to render this successful, it would be greatly beneficial for future reporting to contain information about this split, too. In addition, research on the reasons behind the large inter-regional differences in calibrated IF steepness would further improve our understanding on global TC vulnerabilities.