EGU2020-20446
https://doi.org/10.5194/egusphere-egu2020-20446
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Probabilistic flood loss estimation for residential buildings in Europe

Max Steinhausen, Kai Schröter, Stefan Lüdtke, and Heidi Kreibich
Max Steinhausen et al.
  • GFZ German Research Centre for Geosciences, Hydrology, Potsdam, Germany (max.steinhausen@gfz-potsdam.de)

Floods are the most costly natural disasters for European economies and expected to increase in frequency and magnitude within a changing climate. Governmental agencies, as well as the (re-)insurance sector, rely on accurate flood loss estimations on the European scale to support climate change adaptation policies, prepare for economic impacts, for instance, via the EU solidarity fund and calculate premiums.

Flood loss estimation on the European scale is currently based on deterministic depth-damage functions different for each country. This leads to a fragmented approach in flood loss estimation, greatly simplifying the representation of damage processes without information about associated uncertainties. To overcome these shortcomings we developed the Bayesian Network Flood Loss Estimation MOdel for the private sector (BN-FLEMOps). BN-FLEMOps estimates relative loss to residential buildings depending on flood experience of the population, precautionary measures, building area, building type, return period, duration and water depth (Wagenaar et al. 2018). The structure of this probabilistic multi-variable model is based on empirical data from post-flood surveys and uses consistent continent-wide proxy data for European scale application. BN-FLEMOps was successfully validated in three case studies in Italy, Austria and Germany. The officially reported loss figures of the past flood events were within the 95% quantile range of the probabilistic loss estimation (Lüdtke et al. 2019).

The probabilistic approach enables the quantification of uncertainties of the loss estimates. Model outputs are generated as loss distributions in high spatial resolution, offering Europe-wide information about risk and uncertainty. Thus, providing support for decision-making processes in flood risk management.

Easy applicability to the BN-FLEMOps model is ensured by its implementation in the standardized OASIS loss modeling framework (lmf). The OASIS lmf enables a plug and play combination with various input data sets and other models.

A first application of BN-FLEMOps for a Europe-wide 100 years flood hazard scenario provided by the Joint Research Center resulted in accumulated loss for residential buildings in Europe of 79.0 billion euro (Q20 = 32.3; Q80 = 213.8).

 

References

Lüdtke, S., Schröter, K., Steinhausen, M., Weise, L., Figueiredo, R., Kreibich, H. (2019 online first): A consistent approach for probabilistic residential flood loss modeling in Europe. - Water Resources Research. DOI: http://doi.org/10.1029/2019WR026213

Wagenaar, D., Lüdtke, S., Schröter, K., Bouwer, L. M., Kreibich, H. (2018): Regional and Temporal Transferability of Multivariable Flood Damage Models. - Water Resources Research, 54, 5, pp. 3688-3703. DOI: http://doi.org/10.1029/2017WR022233

How to cite: Steinhausen, M., Schröter, K., Lüdtke, S., and Kreibich, H.: Probabilistic flood loss estimation for residential buildings in Europe, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20446, https://doi.org/10.5194/egusphere-egu2020-20446, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 2 – uploaded on 03 May 2020
add references and online resources, add cc by logo
  • CC1: Comment on EGU2020-20446, Chiara Arrighi, 04 May 2020

    @Max, thank you. I see different scales in the building footprint aggregation (slide 5, i.e. italy vs the Balcans), why? and what effect you might expect on the results?

    • AC1: Reply to CC1, Max Steinhausen, 04 May 2020

      @Chiara Arrighi: In EU member states we used the distribution of building footprint areas in NUTS-3 regions. For some countries this regional differentiation based on NUTS-3 was not possible. Therefore, we aggregated on a national level. This makes the input data less accurate and introduces higher uncertainties into the model. We would expect less precise loss estimations in these countries.

Presentation version 1 – uploaded on 03 May 2020 , no comments