Impact attribution of European floods, 1950-2020

This collection contains impact attribution of 1729 riverine, flash, coastal and compound floods that affected Europe between 1950 and 2020. Each event was reconstructed with a modelling chain combining hazard, exposure, and vulnerability. A total of six drivers are included in two versions: factual (historical) being the best estimate of the drivers per each event, and counterfactual, being an estimate of the driver under static, 1950 conditions. All combinations of possible factual and counterfactual drivers results in 64 scenarios. A list of all scenarios is contained in 'Attribution_scenarios_list.csv', where the columns are as follows:

• Scenario - scenario number used in attribution results 
• Climate_change - effect of changes in river discharges, storm surge heights, wave run-up, and long-term sea level rise
• Catchment_alteration - effect of changes in land use, reservoir capacity and water demand on the hydrological cycle
• Exposure_growth - effect of change in population stock and gross domestic product (GDP) at subnational region level (NUTS3)
• Exposure_local - effect of local-scale exposure changes: land use structure, GDP composition by sector, asset-to-GDP ratio, building of new infrastructure
• Flood_protection - effect of change in probability that a hydrological flood will cause significant socioeconomic impacts, primarily through the breach of structural defenses
• Vulnerability - effect of changing relative loss due to local factors, e.g. private precaution, building material, early warning, emergency measures, etc.

'Event_data.zip' contains six files per each flood event identified by their HANZE ID number (https://zenodo.org/records/12635205):

• 'Attribution_mean_ID.csv' - mean estimate of impact per scenario (fatalities, persons affected, economic loss in 2020 euros)
• 'Attribution_params_ID.csv' - additional impact data per scenario: chance of flood protection failure (fraction), change of fatalities (fraction), relative fatalities (%), relative population affected (%), relative economic loss (%), number of impacted NUTS3 regions, potential absolute flood exposure (fatalities, population affected, economic loss)
• 'Attribution_eco_ID.csv' - mean estimate of economic loss in 2020 euros per scenario per NUTS3 region
• 'Attribution_fat_ID.csv' - mean estimate of fatalities in 2020 euros per scenario per NUTS3 region
• 'Attribution_pop_ID.csv' - mean estimate of population affected in 2020 euros per scenario per NUTS3 region
• 'Attribution_unc_ID.npy' - impact per scenario (fatalities, persons affected, economic loss in 2020 euros), 1000 samples of the uncertainty distribution of impact.

'Aggregated_data.zip' contains six files of aggregated impacts (fatalities, persons affected, economic loss in 2020 euros) per scenario:

• 'Attribution_aggregate_event_IMPACT.csv' - all events aggregated in one file per impact type, displaying:
  • HANZE_ID - unique ID of the event in HANZE database
  • Year - year when the event started
  • Code - country two-letter code
  • Type - type of flood
  • scen_X - scenario number X
• 'Attribution_aggregate_event_IMPACT.csv' - estimated NUTS3-level impacts of all events that affected a given region between 1950 and 2020 (see NUTS3 region map in https://zenodo.org/records/12635205).

Graphs of attribution per event, and maps of aggregated attribution can be viewed online: https://naturalhazards.eu

Results can be reproduced using code (HANZE model v2.4) and input data also available from Zenodo.