Global physical risk modelling for climate services and research

14:00 Wednesday 29 May

SS024 • OC140

Room S7

 

Samuel Eberenz (Switzerland) 1,2; David N. Bresch (Switzerland) 1,2

1 - ETH Zürich; 2 - MeteoSwiss

Objectives

Novel climate services assess the physical risk globalized companies and investors face under current and future climatic conditions. This requires globally consistent quantification of weather and climate risks. Lacking and inhomogeneous data have been limiting risk assessments to local and regional cases. To address this gap, a fully probabilistic, spatially explicit, and globally consistent tropical cyclone (TC) risk model was developed and calibrated against historical loss data. The model was published in open-source Economics of Climate Adaptation tool CLIMADA and implemented together with the research firm Carbon Delta AG.

Methods

The TC hazard set is based on historical cyclone tracks. To estimate global exposure, gridded nightlight data, population density and non-financial wealth were combined. For calibration, the damage function was fitted regionally to historical yearly losses. For uncertainty analysis, probabilistic TC events were created with a random-walk algorithm.

The calibrated model can be used compute the expected economic damage per exposed asset as a measure of risk today, the incremental increase from economic growth and the further incremental increase due to climate change.

Together with Carbon Delta, physical risk on a company level was computed by applying the calibrated model to the company’s data base of production locations and fixed asset values for more than 20,000 publicly traded companies.

Results

The calibrated model is capable of reproducing the magnitude and distribution of annual losses for historical events. The fitted damage functions show a large regional variability. Results are sensitive to several factors, total asset value, hazard intensity, years considered for calibration, and the magnitude of losses reported.

We present the operational model set-up implemented with Carbon Delta, as well as current and future TC risk figures on a company, investment portfolio and country level.

Conclusions

Global consistency and ground truth are important criteria for many climate risk products. Both can be provided by a regionally calibrated probabilistic impact model. The modular architecture of CLIMADA allows to assess TC risk for asset portfolios over different regions, time scales and scenarios (e.g. climate change), and to analyze them on a variety of aggregation levels.

The risk figures come with considerable uncertainties attached. Due to the high sensitivity to the different components of the model (hazard set, exposure, damage function), the calibration is specific to the particular setup. However, modularity makes the calibration process easily reproducible for different setups and transferable to other perils like flood or drought.