Datasets used in the analysis include: 1) Urban Morphological Zone (UMZ) from Urban Atlas 2012. UMZ is the reference unit for the city morphology. They are regarded as the best approximation of the “real” city formand defined as a set of urban areas laying less than 200 m apart, within the core city administrative boundaries). 2) LISFLOOD model outputs from JRC. The discharge return levels were derived for every river pixel for return periods of 100 years. For time window of 30 years (2071–2100), a Gumbel distribution was fitted to the annual maximum discharges simulated by LISFLOOD in every grid cell of the modelled domain based on 12 models and the A1B scenario (Rojas et al.,2012; Rojas et al.,2013). The resultant modelled flood area was intersected with the Urban Morphological Zone extent, and the proportion of potentially flooded UMZ area was calculated for each city by dividing the potentially flooded area by the total UMZ area.

Importantly, the indicator is based on elevation and does not includecurrently existing or plannedflood protection measures like dams, dikes, etc., as data for these are not yet available. Areas shown here as potentially at risk of flood might in reality be protected by flood defences. However, since flood protection measures can fail in certain circumstances, the flood risk remains.