The Flood probability layers show where on the model river network EFAS forecasts a flood within/ later than 48 hours.
Two flood probability layers show the probability of exceeding the 5-year return period in the EFAS forecasts at different lead time ranges:
- 5-year exceedance < 48 h indicates the probability of flooding in the following 48 h, a lead time range for which formal notifications are not issued.
- 5-year exceedance > 48 h indicates the probability of flooding from lead time 2 to 7 days, when formal notifications are issued.
Figure 1. Example of the layer Flood Probability > 48 h on the forecast of March 23 2024 at 00 UTC.
Total probability
The flood probability layers are based on total probability, which means that the hydrological forecasts based on the four different Numerical Weather Prediction (NWP) models (see EFAS Meteorological forecasts) (ECWMF-HRES, ECMWF-ENS, COSMO-LEPS, DWD-ICON) need to be combined. This combination is based on the NWP probabilistic skill measured in terms of the Brier score ($BS$):
$$\text{BS} = \frac{1}{T}\sum_{t=1}^{T} \left( P_{obs,t} - P_{pred,t} \right)^2$$
where $T$ is the number of time steps, $P_{obs,t}$ is the observed probability of exceedance, and $P_{pred,t}$ is the predicted probability of exceedance at a specific time step $t$. Brier scores were computed for every NWP and lead time using the historical archive of EFAS v4 forecasts and reanalysis. The resulting Brier scores were converted into weights by inverse distance weighting:
$$w_{nwp,lt} = \frac{BS_{nwp,lt}^{-p}}{\sum_{i=1}^{4} BS_{i,lt}^{-7}}$$
where $w_{nwp,lt}$ is the weight assigned to a specific NWP and lead time, and $BS_{nwp,lt}$ is the Brier score of that NWP at that lead time.
Figure 2. Distribution of weights over NWP models and lead time. DWD stands for DWD-ICON, HRES for ECMWF-HRES, COS for COSMO-LEPS and ENS for ECMWF-ENS. Orange colours represent deterministic models and blue colours probabilistic models.
As shown in the figure above, the assignment of weights gives priority to probabilistic NWPs over deterministic counterparts, particularly on the ECMWF-ENS model, which proved higher skill. The predominance of ECMWF-ENS increases with lead time.