By Dominik Paprotny

Summary: Growth in flood losses in Europe can be explained by growing exposure, while vulnerability is shown to decline with economic development.

Lost in perception

The common perception of flood losses is that they are growing and this perceived increase is driven by a changing global climate. The absolute value of losses has indeed grown worldwide (Munich Re, 2018), however the actual source of the growth in flood risk still eludes the public consciousness. Flood risk is the interplay of three components: hazard (flood intensity and its likelihood), exposure (an inventory of things that are in harm’s way) and vulnerability (how badly things get damaged if a flood strikes them). A study published last year on flood losses in Europe that have occurred since 1870 (Paprotny et al., 2018a) shows the most prominent role of exposure in driving the changes in flood risk.

The new normal

The analysis presented in Paprotny et al. (2018a) was based on two sources: a consistent and spatially-defined catalogue of damaging floods from 1870 to 2016, and detailed information on the land use, population, assets and economic production in each affected area, both at the time of the event and at present (Paprotny et al., 2018b). This enables adjustment, or ‘normalization’, of reported flood losses by removing the effect of land-use change and economic growth. As a consequence, the upwards trend in the number of affected people and economic losses turned out much lower for 1870–2016 than without such adjustment. For 1950–2016, it shows that economic losses were actually declining. ‘Normalized’ fatalities indicated a steep decrease.

It was further discovered that the records of events and losses are biased towards modern times: information on smaller floods that have occurred deep in the past is largely missing, due to lower accessibility of primary sources and fewer mentions in historical books. If we rank the floods by the severity of (normalized) losses they have caused, the top 20% of events are evenly distributed over time, while the smaller the flood, the higher upwards trend is recorded. Considering those unknown losses based on the expected distribution of events (as for each big flood there have to be several small ones), a decline in flood losses is evident, especially since 1950.

Safer in the city

What about the remaining two factors, then? Previous research has indicated that extreme discharges in most rivers of Europe are not occurring any more frequently today than in the past, though climate data indicate some increase in heavy rainfall that leads to flash floods, while sea level rise is contributing to higher storm surges (Mudelsee et al., 2003, Kundzewicz et al., 2005, Glaser et al., 2010, Hall et al., 2014). Notwithstanding, any increase in the hazard must be matched by an even greater decline in the vulnerability in order to explain the observed trends in flood losses (stable or declining depending on the measure).

Data on land use, population, economic production and wealth in the floodplains at the time when the top 20% floods in the catalogue have taken place are revealing. The amount of losses (fatalities, persons affected, monetary value) relative to the number exposed in the hazard zone is correlated in particular with the urbanization and economic development. Lower vulnerability was particularly related to a higher percentage of the floodplain under an urban fabric; population density; GDP per capita; and percentage of assets in infrastructure. Conversely, higher vulnerability was mainly correlated with the percentage of assets belong to the agricultural sector, and to a lesser extent, to industrial share of assets or agricultural land use (Paprotny et al., 2018a).

The lower vulnerability of urban areas compared to rural areas could be to some extent explained by better structural flood protection of the former, but the percentage of the flood hazard zones actually flooded during the most extreme events is weakly related to land use or population density (Paprotny et al., 2018a). Hence, lower relative losses in Europe must be explained by lower vulnerability of residential, services and infrastructure assets compared to industrial and particularly agricultural assets. Part of the explanation could be technological development (better construction methods, early warning, evacuation), but also by evolving institutional sector (the rise of professional emergency services, flood prevention/response agencies and regulations).

Getting out of the way

The findings from Paprotny et al. (2018a) provide an insight on how flood vulnerability changes with the socio-economic development of countries. This is relevant given that the demographic, economic and settlement structures of ‘developing’ countries resemble far more Western Europe in the early 20th century rather than the early 21st. Countries with more rustic and poorer populations, as well as largely agriculture-based economies, are therefore more vulnerable to floods, but the example of Europe shows that they can adapt effectively so that the losses won’t grow as fast as their exposure, which – at current trends – will rise due to urbanization and development.

Yet the reduction of exposure is also a possibility. Preventing urban sprawl, for instance, would help; the recent rise of urban floods from extreme cloud outburst in European cities can be largely attributed to growth in soil-sealing artificial surfaces (eight-fold since 1870), which is far faster than the increase in urban populations. Floodplains should be less intensely used as well. Significant restrictions on construction in hazard zones could help particularly countries which have their most intense development phase still in front of them.


Glaser, R., Riemann, D., Schönbein, J., Barriendos, M., Brázdil, R., …, and Himmelsbach, I. (2010) ‘The variability of European floods since AD 1500’, Climatic Change, 101, pp. 235–256.

Hall, J., Arheimer, B., Borga, M., Brázdil, R., Claps, P., …, and Blöschl, G. (2014) ‘Understanding flood regime changes in Europe: a state-of-the-art assessment’, Hydrology and Earth System Sciences, 18, pp. 2735–2772.

Munich Re (2018) Weather-related loss events worldwide 1980 – 2017. Available at (Downloaded: 22 February 2019).

Paprotny, D., Sebastian, A., Morales Nápoles, O., and Jonkman, S.N. (2018a) ‘Trends in flood losses in Europe over the past 150 years’, Nature Communications, 9, 1985.

Paprotny, D., Morales Nápoles, O., and Jonkman, S.N. (2018b) ‘HANZE: a pan-European database of exposure to natural hazards and damaging historical floods since 1870’, Earth System Science Data, 10, pp. 565–581.

Kundzewicz, Z. W., Graczyk, D., Maurer, T., …, and Szwed, M. (2005) ‘Trend detection in river flow series: 1. annual maximum flow’, Hydrological Sciences Journal, 50, 797–810.

Mudelsee, M., Börngen, M., Tetzlaff, G., Grünewald, U. (2003) ‘No upward trends in the occurrence of extreme floods in central Europe’, Nature, 425, 166–169.

BioDominik Paprotny is a postdoctoral researcher at the GFZ German Research Center for Geosciences. A geographer by training, he carried out highly interdisciplinary research on flood risk in Europe at Delft University of Technology, before moving to GFZ to work on modelling flood damages to residential and commercial buildings.

Flood losses aggregated to 30-year periods, reported numbers and with three types of adjustments: normalization, gap-filling of missing (normalized) loss data and estimation of underreporting of small flood events and normalized damages they caused, for (a) number of events; (b) area inundated; (c) fatalities; (d) persons affected; e) financial value of losses with normalization by GDP; and (f) financial value of losses with normalization by wealth.