Reforecast pooling for flood hazard estimation: strengths and weaknesses
Abstract
Estimating the magnitude of extreme flood events occurring less than twice a century is challenging due to the shortness of observed time series. Here, we develop a new approach to increase the sample size available to estimate the frequency of extreme local and regional flood events - pooling reforecast ensemble members from the European Flood Awareness System (EFAS). We assess the added value of such pooling and determine where in Europe one might expect the most extreme events. We work with a set of 234 catchments from the Global Runoff Data Center for which EFAS model performance is satisfactory when comparing simulated to observed flood events. We pool EFAS-simulated flood events for 10 perturbed ensemble members and lead times from 22 to 46 days, where flood events are only weakly dependent. The resulting large ensemble consists of 130 time series and enables analyses of extreme events. We demonstrate that such ensemble pooling produces more robust estimates with considerably reduced uncertainty bounds than observation-based estimates but may equally introduce biases arising from the simulated meteorology and hydrological model. Our results show that specific flood return levels are highest in steep and wet regions and are comparably low in regions with strong flow regulation through dams. Furthermore, our pooled flood estimates indicate that the probability of regional flooding is higher in Central Europe and Great Britain than in Scandinavia. We conclude that reforecast ensemble pooling is an efficient approach to derive robust local and regional flood estimates in regions with sufficient hydrological model performance thanks to a substantial increase in sample size.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H43A..03B