Flood Summary Products Overview

Probability of forecasts exceeding the EFAS 5-year return period threshold. Forecasting range 0-48 hours. 

Probability of forecasts exceeding the EFAS 5-year return period threshold. Forecasting range 2 -10 days.

Persistence probability (average of last 2 runs) of forecasts exceeding the EFAS 2-, 5-, and 20-year return period thresholds.

Map

Estimated flood extent at 100m resolution based on forecasted flood magnitude and not considering the role of flood defences. Note: If EFAS doesn’t forecast significant flood impact the layer will be empty. (More information).

Potential impact of floods on population, land use (agriculture, urban) and infrastructures (major roads). Impacts are based on the rapid flood mapping layer (above) and aggregated over NUTS regions. The level of flood risk is derived from the combination of expected impacts and likelihood. Note: If EFAS doesn't forecast significant flood impact the layer will be empty. (More information).

Reporting Points where forecast outputs are available. Yellow, Red and purple squares highlight stations where the flow is forecasted to exceed respectively the EFAS 2, 5 and 20-year return period thresholds. Dark outlines indicate a threshold exceedance within the next two days. The labels show the ECMWF-ENS (left) and COSMO-LEPS (right) forecast probabilities and the forecast probability tendency over the last 3 forecasts (arrows). Grey (blue) squares are stations where (real-time post-processed) forecast hydrographs are available. For these points, yellow, red or purple outlines signify that the flow is forecasted to exceed the EFAS return period thresholds, but they do not meet the criteria to be flood reporting points. (More information).

The layer aggregates tweets about floods in a specific area. Data collection is triggered on the basis of weather forecasts. Each tweet's text likelihood of being about a flood is classified using ML models (scale 0 to 1). Each tweet is geotagged using either the location mentioned in its text or the coordinates of the tweet itself. If the location extracted is out of the administrative areas for which the collection is triggered, the tweet will be discarded from further analysis. Geotagged tweets are grouped into 3 categories, low (flood relevance prob < 0.2) - mid (0.2-0.8) - high (>0.8) for each area. According to the ratio between number of tweets in 'mid' bucket and 'high' bucket (mandatory more than 10 in any case) the administrative area is 'colored' in grey (low>1), orange (high>5 X mid) or red (high>9 X mid). Finally, for regions in red or orange, a script extracts the 5 most representative tweets. The layer is updated every 30 minutes

Reporting points where maximum discharge is expected today (within the next 1-2 days; 3-5 days; >5 days). Roman numbers denote when exactly maximum discharge is expected. Colour denotes the highest alert level expected by a certain forecast.