Onboard RAIN pan-European windstorm dataset into physrisk

[joemoorhouse]

The feature is to on-board the dataset:
https://data.4tu.nl/articles/dataset/Pan-European_data_sets_of_windstorm_probability_of_occurrence_in_present_and_future_climate/12719387
into physrisk,

The on-boarding script itself would be added to the hazard repo:
https://github.com/os-climate/hazard
following for example the pattern of:
https://github.com/os-climate/hazard/blob/main/src/hazard/onboard/tudelft_flood.py

The aim is also to apply damage functions from:
Feuerstein, B. et al. Towards an improved wind speed scale and damage description adapted for Central Europe. Atmos. Res. 100,
547–564 (2011).
https://trustedspotter.eu/content/images/1-s2.0-S0169809511000032-main.pdf

I would suggest we make use of the configuration-based functions of #312 - to use this as a test case.

[joemoorhouse]

Hi @jmcano-arfima, adding this issue in physrisk even if some of the work in hazard.

[jmcano-arfima]

Hi @joemoorhouse, thanks for opening the issue! Let me add two things:

The on-boarding of the RAIN data should be relatively straightforward, but it is provided as a historical return period/OEP map + relative changes in OEPs for the different scenarios. From what I understand other datasets (such as TUDelft riverine) were on-boarded from several return period maps. In any case the math is of course trivial, but it might require a bit more legwork on the dev side. If this is a common occurrence we could provision some standard methods to do the conversion (if not already done).

On the other hand, regarding Feuerstein, B. et al. damage functions, I fully agree that it would make an excellent test case. Let me share a digitization of the curves that was provided by the ECB:
Wind_Risk_and_Loss_Assessment_InputsECBProvided.xlsx
(let me know if we have a better solution to share files)

Here we could provide support for building typing à la Feuerstein but equally important (or even more so) would be to define the default behavior when this data is absent. From what I understand the ECB is building vulnerability curves out of weighted combinations of the original six at country level using the data in the 'Building Type per country_v2' sheet (that best fits the XWS catalogue). I think this would also be the best solution for us as it is quite economical in the sense that it only uses asset location, which is the bare minimum we are already asking for.

Edit: The above is what I understand from the provided ECB excel, but I will probably try to contact them again to confirm it myself.