River flood risk: European scale
Vulnerabilities - European assessment river hazard
Some 0.03% of the European population, on average, are thought to have been affected by river flooding annually between 1870 and 2016, at a yearly average cost of 0.08 - 0.09% of gross domestic product (68).
Severe river floods often result in huge economic losses and fatalities. During the period 1950-2015, these events affected around 18 million people and caused economic losses of approximately 133 billion USD in Europe (54).
Five large-scale homogeneous regions in Europe have been identified in terms of flood regimes, based on the longest available flow series from across Europe (hydrological data from gauging stations in 25 European countries) (13). These regions are:
- An Atlantic region, from the Iberian Peninsula to Denmark and central Germany in the east and Iceland in the north. In this region floods are mainly driven by persistent and intense rainfall associated with frontal storms caused by extra-tropical depressions moving into Europe by westerly atmospheric circulation. For this region, the available flow series show a flood-rich season from December to March, and a flood-poor season from May to September.
- A Continental region in Central Europe, from eastern Germany to the Baltic states and from Slovakia and northern Austria to southern Sweden. In this region, snowmelt, synoptic depressions and atmospheric blocking in winter and spring are occasionally complemented by Vb systems in summer, which are Central European cyclonic weather patterns embedded in mid-latitude synoptic depressions, linked to the large-scale atmospheric flow coming from the Atlantic and encircling the northern hemisphere, and further fed by local depressions of Mediterranean nature. For this region, the available flow series show two flood-rich months in March and April, and two flood-poor months in September and October.
- A Scandinavian region that includes the Nordic countries, except Denmark and southern Sweden, where floods are often driven by snowmelt potentially in combination with rainfall mostly generated by synoptic systems. For this region, the available flow series show two flood-rich months identified in May and June, and a flood-poor season between November and April.
- An Alpine region that includes rivers with headwaters in the Alps, where most of the largest floods are summer floods resulting from blocking situations and more minor ones that may occur in winter associated with snowmelt and rain on snow. In the Alpine region, a flood-rich season extends from May to July, and a flood-poor season is detected from September to March.
- A Mediterranean region, from eastern Spain to Romania, where floods are driven by multiple mechanisms of both continental and maritime nature. In the Mediterranean region, a mixture in flood generating mechanisms leads to a similar monthly frequency of floods throughout the hydrological year.
Changes return periods 20-year, 50-year, and 100-year river floods
Return periods are changing
Changes have been examined in return periods of 20-year, 50-year, and 100-year river floods, globally, with 1970 as a reference point (69). The analysis is based on more than 10,000 river gauge records. At each site, the authors estimated the magnitude of the 20-, 50-, and 100-year flood in the 1970s, and the changing return period of these 1970s floods in the current situation.
Western Europe: reduction return period
In temperate climate zones including Western Europe, return periods have decreased for the 20- and 50-year floods. On average, in temperate climates globally, floods that had a return period of 20 and 50 years in 1970, now have return periods of 8 and 21 years, respectively. According to the analysis, the return period of a 100-year flood has increased. One should be aware, however, that the number of sites used to evaluate these trends is much smaller for the 100-year floods than for the 20-year floods. After all, the numbers of sites with a sufficiently long record to carry out these analyses are much smaller for increasingly rare extremes. The conclusions on the 100-year-floods are, therefore, of little value (69).
The global picture: increase return period
Remarkably, the overall global impression of a changing return period is not a reduction, but an increase of these flood return periods: The 20-, 50-, and 100-year floods of the 1970s have become, on average, 41-, 152-, and 358-year floods under present-day conditions. With the exception of the temperate zones, this increase of return period was found for all other climate zones (cold, arid, tropical, polar) and includes eastern Europe. This indicates that, on a global scale, the frequency of occurrence of these flood levels has decreased. According to the authors, this may be due to a combination of climatic changes, land cover changes such as urbanization, agriculture, and other anthropogenic influences on river regimes (69).
An indication, no precise estimates
The authors stress that these ‘new’ return periods should be seen as indicative of the general direction of change rather than precise estimates. After all, the results depend on the selected sites with sufficiently long historic flood records (69).
No evidence (yet) of climate change impacts
Compelling evidence that the number of major river floods is increasing at a global scale is lacking. Conclusions in previous scientific studies that the occurrence of river flooding is changing due to climate change are based upon small samples of catchments or short periods of record. Generalizations about climate-driven changes in floods based upon these studies are ungrounded. This firm statement was concluded by a number of scientists that studied the occurrence of major floods across North America and Europe. The results of their study, published in the Journal of Hydrology, support the conclusion of the IPCC (44) that globally there is no clear and widespread evidence of changes in flood magnitude or frequency in observed flood records (43).
The impact of climate change on river floods can only be studied when the effects of human catchment alterations can be separated from those of climate. These human alterations include urbanisation, land-use change, diversions, abstractions and reservoir regulation. In many previous studies these alterations may have impacted the change in the number of floods (43).
The study in the Journal of Hydrology presents the first intercontinental assessment of climate-driven changes over time in the occurrence of major floods: floods with a return period of 25, 50 and 100 years, that are likely to have the greatest societal effects. For 1204 diverse but minimally altered catchments, long-term trends in major flood occurrence have been analysed. These sites include catchments in Canada, the United States, France, Ireland, Norway, United Kingdom, Denmark, Finland, Germany, Iceland, Spain, Sweden and Switzerland. All catchments had <10% current urban area, no substantial flow alteration or known substantial land cover change, good quality peak-flow data and <10 years missing data. Floods were analysed for the periods 1961 - 2010 and 1931 - 2010 (43).
For this period of 80 years there was no compelling evidence for consistent changes over time in major-flood occurrence. These results differ from the most comparable previous study (45), which found significant increases in the occurrence of 100-year floods during the 20th century for 16 very large catchments in the Northern Hemisphere. Flood trends at these catchments may be impacted by human alterations (43).
Vulnerabilities - Areas in which surrounding rivers flood at the same time are increasing
When rivers flood, nearby rivers often flood at the same time. The distance over which multiple rivers flood near synchronously far exceeds the size of individual drainage basins. This was shown from an analysis of annual flood data from several thousand European rivers over the period 1960-2010 (58,67).
A new concept: the flood synchrony scale
The researchers defined a new concept to characterize the spatial extent of floods: the flood synchrony scale (58). This scale is defined as the maximum radius around an individual river gauge within which at least half of the other river gauges also record flooding almost simultaneously. ‘Almost simultaneously’ refers to a time interval of 7 days to capture responses to single forcing events (e.g., a rainstorm), allowing for somewhat lagged responses of rivers to the same weather systems. Data were taken from the European Flood Database (59), consisting of dates of annual maximum stream flows or water levels in European river catchments for each calendar year from 1960 to 2010.
The flood synchrony scale varies regionally. In a band stretching from northern Spain toward the Alps, into central Europe and the Carpathians, flood synchrony scales are generally less than 100 km. This indicates that annual maximum floods are relatively localized around these mountainous regions. Floods are usually correlated over larger distances in the rest of Western Europe, and flood synchrony scales frequently exceed 250 km across large swaths of northeastern Europe. The average flood synchrony scale for European floods represents a surface area of almost 70,000 km2, which is over an order of magnitude larger than the typical size of drainage basins in the database (58).
No simple correlation with precipitation
One might expect this synchrony scale to correlate with annual maximum daily precipitation. This appears not to be the case, however. Many annual floods do not result from maximum annual precipitation but instead (for example) from snowmelt or subextreme precipitation during times of high soil moisture (60). Also, topography appears to substantially influence the flood synchrony scale: flooding tends to be more spatially coherent at lower altitudes and in flatter landscapes (61).
Remarkably, over the period 1960-2010, flood synchrony scales have grown by about 50%: averaged across Europe, the flood synchrony scale has grown by roughly 1.1 % per year (±0.55, 95% confidence interval). This means that flooding has become more spatially synchronized. Flood synchrony scales have grown across most of Europe, most notably in parts of the British‐Irish Isles and large parts of Germany, Belgium, the Netherlands, Austria, Italy, Sweden and the Balkans, but have declined in eastern Poland, Romania, and parts of Russia. No explanation is given for this trend over the last decades (58). Other studies have shown that the characteristics of river floods are changing under climate change. Possibly this affects flood synchrony as well.
Increase in central Europe and the British Isles
In the period 1960-2010, the extent of rivers with simultaneous high discharges increased in central Europe and the British Isles. In these parts of Europe, the relevance of soil moisture excess as a flood generation process has increased because precipitation and soil moisture have increased. Soil moisture conditions are linked to large-scale weather patterns and change in a similar way over large areas. Thus, when soil moisture becomes a more dominant driver of high discharges, more rivers in neighbouring catchments may flood at the same time (67).
Decrease in Eastern Europe
In Eastern Europe, the opposite is the case. In the period 1960-2010, the extent of rivers with simultaneous high discharges decreased in Eastern Europe. This decrease agrees with a decrease of the relevance of snowmelt in Eastern Europe, where it is the most important process. Snowmelt affects a large area, and many river catchments, in the same way. If this flood generation process becomes less important in Eastern Europe, the area over which flooding occurs simultaneously gets smaller (67).
Changes in spatial extent and magnitude are related
The analysis also shows that changes in the spatial extent of simultaneous peak discharges go hand in hand with changes in the magnitude of these peak discharges. This means that flood risk in central Europe and the British Isles increases both due to the increase of this spatial extent of peak discharges and due to the increase of the magnitude of these discharges. The change in flood generation processes affects both the spatial extent of the conditions of river flooding and the extremity of these conditions (67).
If these trends persist into the future, the combination of stronger floods and larger extents is likely to increase the flood risk substantially. This highlights the increasing importance of transnational flood risk management (67).
Northwestern Europe: trends in compound flooding
Compound floods in delta areas are the co-occurrence of high coastal water levels and high river discharge. Correlations between high water levels at the coast and in rivers have been reported in the scientific literature both globally (64), and regionally across Europe (65). There are two mechanisms that may cause such events. First, high coastal water levels may affect river flows and water levels by backwater effects or by reversing the seaward flow of rivers, particularly in low-lying areas. Second, high coastal water levels and high river discharge may stem from a common meteorological driver. Severe storm periods may be associated with high winds leading to storm surges, and at the same time with high precipitation followed by inland flooding (63).
Compound flooding may lead to significant impacts and much more disastrous consequences than each of these extremes individually. The occurrence of compound floods from extreme coastal water levels and peak river discharge was studied for northwestern Europe, along the coastlines from France to Sweden. Datasets were used for locations at the North Sea and the Baltic Sea covering most of the previous century (63).
In the definition of compound floods, water level extremes at the coast and in rivers do not have to occur at exactly the same spot and the same time. The impact of a storm may be noticed in river discharge a couple of days later than in storm surge levels at the coast because it will take some time for rainfall to reach the river. Also, the river outlet may be far from the area where the storm was felt. In this study a flood event was called a compound event when high water levels at the coast and in the river were no more than 500 km apart and occurred no more than 7 days apart (63).
There appears to be a distinct difference between mid-latitude and high latitudes regions in northwestern Europe: compound flooding occurs more often in mid-latitude regions, along the west facing coasts of the United Kingdom, Germany, the Netherlands, and Sweden. No clear explanation was given for this. There may be a link with weather oscillations that have a stronger impact on certain parts of northwestern Europe. This may explain the presence of decadal and multidecadal variability in the magnitude and frequency of compound floods. There have been compound flood-rich and flood-poor periods, like the 1960s and the 1990s, respectively (63).
This variability makes it difficult to find long-term trends related to climate change or other human interventions. Although no attempt was made to relate trends in compound floods to climate change, the authors stress that global warming could increase the severity and frequency of compound extremes (66) by changes in wind speed and in rainfall intensity of severe storms.
Vulnerabilities - Europe's socioeconomic developments recent decades
Direct economic river flood damages in Europe have increased over recent decades but this increase is due to development in flood zones and not due to observed climate change (4). In general, few changes in flood trends can be attributed to climate change, partly due to the lack of sufficiently long records (5).
According to the NatCatSERVICE database, almost 1,500 floods and wet mass movement events happened in EEA member countries in the period 1980-2013, with more than half of them since 2000. These floods have resulted in over 4,700 fatalities and caused direct economic losses of more than EUR 150 billion (based on 2013 values), which is almost one-third of the damage caused by all natural hazards (41).
Vulnerabilities - European assessment changing peak flows so far
Globally, river runoff in snow-dominated or glacier-fed high mountain basins is projected to change regardless of emissions scenario, with increases in average winter runoffand earlier spring peaks. In regions with little glacier cover, such as the European Alps, most glaciers have already passed the peak of summer runoff due to increased glacier melting (62).
The seasonal timing of river floods across Europe has been changing since 1960. This was concluded from a consistent pan-European assessment of observed flood seasonality trends between 1960 and 2010, the first evaluation of how climatic changes are influencing flood regimes at the scale of the entire European continent. The assessment was based on a database that pools flood records over the period 1960-2010 from more than 4000 riverine gauging stations across 38 European countries. The results highlight the existence of a clear climate signal in flood observations at the continental scale (40).
Earlier floods northeastern and western Europe
River floods occur earlier throughout northeastern and western Europe. Warmer temperatures have already led to earlier spring snowmelt floods throughout northeastern Europe. Sustained winter rainfall led to earlier soil moisture maxima (saturated soils) and thus to earlier winter floods in western Europe. The time shifts are quite large. Compared with the 1960s floods now occur more than 8 days earlier in areas of northeastern Europe (southern Sweden, the Baltics), at 50% of the gauging stations in this region. In western Europe, along the North Atlantic coast from Portugal to England, 50% of the gauging stations show a shift toward earlier floods by at least 15 days since the 1960s. In fact, 25% of the gauging stations in western Europe show earlier floods by even more than 36 days since the 1960s (40).
Later floods around the North Sea and parts of the Mediterranean coast
River floods occur later around the North Sea and some sectors of the Mediterranean coast. According to the authors of this study this is due to delayed winter storms associated with polar warming. Around the North Sea (southwestern Norway, the Netherlands, Denmark, and Scotland), 50% of the gauging stations show a shift toward later floods by more than 8 days since the 1960s. In some parts of the Mediterranean coast (northeastern Adriatic coast, northeastern Spain), 50% of the gauging stations show a shift toward later floods by more than 5 days since the 1960s (40).
Vulnerabilities - European assessment changing peak flows 21st century
The timing of peak river flows varies from one catchment to another because flows are not generated the same way for different catchments. In some catchments precipitation is dominated by rainfall in winter and extreme flows typically occur in winter. In other catchments snow accumulates in winter months and extreme flows are dominated by snowmelt and occur in spring. A third category represents catchments where both rainfall and snowmelt are important constituents of extreme river flow. Along with the timing of extreme flows, also the peaks of these flows and the total volume of the flood waves over a number of days vary from one catchment to another. It is to be expected that extreme flows also change differently under climate change, depending on how the ratio between rainfall and snow will change (32).
By the end of the 21st century, climate change is projected to alter European floods in complex ways. Decreases in flood peaks are reported for Northern Europe (48) and the Mediterranean (49). For Northern Europe this is due to increased temperatures that reduce snow accumulation in winter leading to less melting water in spring (50). For the Mediterranean this is due to strong decreases in annual precipitation (51). Different signs of change are in general reported for Central Europe and the British Isles (52).
Projections of changes in peak river flows and flood wave volumes were made for catchments in different parts of Europe, varying from snow-dominated catchments in Norway, rain-dominated catchments in Belgium and Denmark, and German and Polish catchments with mixed flood regimes driven by rainfall in winter and snowmelt in spring. In total 11 catchments in 9 countries were studied, covering Europe from Norway to Cyprus. Projections were made for the period 2071 – 2100, compared with 1961 – 1990, under a moderate scenario of climate change (the so-called SRES A1B emission scenario) and based on a large number of climate and hydrological models, and statistical procedures. In this study peak flows were defined as floods that occur on average once every year to once every 5 years (32). From this study the following can be concluded.
- Floods driven by rainfall increase. The results indicate that extreme flows generally increase by the end of the 21st century in catchments with rainfall-dominated flood regimes, consistent with projected changes of more extreme precipitation in winter. In Mediterranean catchments such as the studied one in Cyprus, however, extreme winter precipitation and extreme flows may decrease (32).
- Floods driven by snowmelt decrease. In catchments with spring floods caused by snowmelt, however, a decrease of extreme flows is projected because less snow accumulates in winter and snowmelt starts earlier. An exception are high northerly mountain catchments in which snowmelt is expected to continue to be important in the future, whilst increases in extreme precipitation during and following the snowmelt season will contribute to an increase in the flood hazard (32).
- Floods driven by rainfall and snowmelt (mixed) increase. In catchments with mixed flood regimes, the extreme flows are expected to increase, suggesting that in those catchments the increase in extreme precipitation dominates over the reduction of snowmelt and that the flood regime shifts towards a rainfall-dominated flood regime (32).
European floods under 1.5, 2, and 3 degrees global warming
The Paris Agreement sets out actions to limit global warming well below 2°C, and preferably below 1.5°C. The impact of climate change on European floods was studied for these levels of global warming, and for 3°C global warming, based on several climate models (GCMs) and hydrologic models, and three scenarios of climate change (the RCPs 2.6, 6.0, and 8.5). The period 1971-2000 was selected to represent present-day conditions. Two flood indicators were studied: high flows (the height of the river flow that is exceeded 10% of the time), and floods (the annual maximum flow). Only river basins with a contributing area larger than 1000 km2 were considered (47).
The study indicates that both high flows and floods will decrease in the Mediterranean, up to -31% and -17% respectively (at 3°C global warming), as a result of a decrease in total annual precipitation. Particular hotspots of projected changes in the Mediterranean region are the Iberian Peninsula and the Balkans. In Northern regions, high flows are projected to increase due to increasing precipitation (up to 12% under 3°C global warming), but floods are projected to decrease due to less snowmelt (up to -5% under 3°C global warming). Atlantic and Continental Europe can be considered as a transition zone between decreases in Southern and increases in Northern Europe. Projected changes in these regions are generally less than 10% in magnitude. These results are in line with previous studies (53). No clear differences were found for 1.5°C and 2°C global warming, respectively. According to the authors, adaptation measures for limiting the impacts of global warming could be similar under 1.5°C and 2°C global warming, but have to account for significantly higher changes under 3°C global warming.
Vulnerabilities - European multi-hazard assessment 21st century
The most relevant hazards for Europe in terms of average annual losses and deaths are heat and cold waves, river and coastal floods, droughts, wildfires and windstorms (27). The future changes in these hazards in Europe under climate change have been assessed under a business-as-usual scenario of climate change (the so-called SRES A1B scenario) (26):
- Heat waves show a progressive and highly significant increase in frequency all over Europe in near future climate. This increase is most pronounced in Southern Europe, where current 100-year events could occur almost every year in the 2080s. In Southern Europe, up to 60 % of the territory could be annually exposed to a current 100-year heat wave by the end of the century.
- Cold waves show an opposite trend with current cold extremes tending to mostly disappear in Europe in more distant futures. The current 2-year event may occur less than every 100 years by the end of the century almost everywhere in Europe.
- Stream flow droughts may become more severe and persistent in Southern and Western Europe: the current 100-year events could occur approximately every 2 to 5 years by 2080 in Southern and Western Europe, respectively. This results from the reduced precipitation and increased evaporative demands with higher temperatures. Northern, Eastern and Central Europe show an opposite tendency with a strong reduction in drought frequency caused by higher precipitation that outweigh the effects of increased evapotranspiration (28).
- Most of Europe, especially Western, Eastern and Central regions, could experience an increase in the frequency of extreme wildfires: current 100-year events will occur every 5 to 50 years by the end of the century. Interestingly, Southern Europe shows a decrease in the frequency of very extreme events, which is likely due to the expected reduction in net primary productivity of terrestrial ecosystem that may limit the fuel availability and, ultimately, the propagation of large wildfires (29).
- Western Europe shows a consistent rise in future river flood hazard: current 100-year events could manifest every ~30 years in 2080s (30). A modest but significant decrease in river flood frequency is projected in Southern, Central and Eastern regions, in the latter because of the strong reduction in snowmelt induced river floods, which offsets the increase in average and extreme precipitation.
- Coastal floods show a progressive and pronounced increase in recurrence along Europe’s coastlines chiefly caused by sea level rise: current 100-year events may manifest every 2 to 8 years, or even sub-annually in Eastern Europe, in the 2080s.
- It is highly uncertain how windstorms may change. Projections indicate that areas with increases in windstorm hazard are mainly located in Western, Eastern and Northern Europe, while Southern regions present slight reductions in frequency as observed in previous studies (31).
Looking at the combination of these 100-year extreme events, results suggest that entire Europe could face a progressive increase in overall climate exposure, with a prominent spatial gradient towards southwestern regions where heat waves, droughts and wildfires are particularly effective (26).
Potential key hotspots that are potentially prone to an increase in exposure to multiple hazards are mainly located along coastlines and in floodplains where windstorms and floods will be likely relevant in combination with temperature-related hazards. More exposed regions include the British Isles, the North Sea area, north-western parts of the Iberian Peninsula, as well as parts of France, the Alps, Northern Italy and Balkan countries along the Danube River (26).
Vulnerabilities - Global socioeconomic developments recent decades
Disaster risks are rapidly increasing around the world: many regions are experiencing greater damage and higher losses than in the past. Annual total damage (averaged over a 10-year period) has increased tenfold between 1976–1985 and 2005–2014, from US$14 billion to more than US$140 billion. Average population affected each year has risen from around 60 million people (1976–1985) to over 170 million (2005–2014) (17).
Increasing exposure to flooding is the main cause of the steeply rising trend in global river flood losses over the past decades. In fact, various analyses of historical loss databases have not yet been able to derive a clear signal of climate change in these increasing losses (19).
Vulnerabilities - Global assessment socioeconomic developments 21st century
Disaster risk is a combination of three components: the hazard (the potentially dangerous naturally occurring event, such as a storm, heat wave or flood), exposure (the population and economic assets located in hazard-prone areas), and vulnerability (the susceptibility of the exposed elements to the natural hazard) (18). Hazard, exposure, and vulnerability are not static. They change in time. Future projections of those changes must be part of investments in disaster preparation today. Exposure, for instance, increases as population grows in hazardous areas, and as improved socioeconomic conditions raise the value of assets (16).
Increasing exposure to flooding is the main cause of the steeply rising trend in global river flood losses over the past decades. In fact, various analyses of historical loss databases have not yet been able to derive a clear signal of climate change in these increasing losses (19). This trend will continue to rise: between 2010 and 2050, estimated global population exposed to river and coastal flood is expected to increase from 992 million to 1.3 billion (20). Similar observations have been made for other climate-related disasters. For instance, the estimated share of the world population facing water scarcity increased from 20% in 1960 to 50% in 2000 (21). The IPCC (18) has high confidence that “increasing exposure of people and economic assets has been the major cause of long-term increases in economic losses from weather- and climate-related disasters.”
The impact of urbanization, excluding climate change
The impact of urbanization on the extent of urban areas exposed to flood and drought hazards has been assessed, without factoring in the potential impacts from climate change. The results of this assessment are summarized below (7).
Urban areas in coastal zone
In 2000, over 10% of total global urban land was located within the low-elevation coastal zones (LECZ, defined as ‘‘the contiguous area along the coast that is less than 10 m above sea level’’) that covers only 2% of the world’s land area. Most of the urban land in the LECZ was primarily located in the developed countries in Northern America and Western Europe along with China. By 2030, however, most of the urban land within the LECZ will be found in the developing countries. From 2000 to 2030, globally the amount of urban land within the low-elevation coastal zones is projected to increase by 230%; for Western and Eastern Europe this increase is projected to be 100% and 7%, respectively, resulting in 13% (Western Europe) and 2% (Eastern Europe) of the urban area being located in LECZ in 2030, respectively (7).
Urban areas exposed to high-frequency floods (coastal and river)
With respect to high-frequency flood zones, including exposure to both coastal and river floods, in 2000 about 30% of the global urban land was located in these zones; by 2030, this will reach 40%. For Western Europe these numbers are 34% (2000) and 34% (2030), and for Eastern Europe 9% (2000) and 10% (2030) (7).
A broad shift is projected in the urban exposure from the developed world to the developing world from 2000 to 2030. The emerging coastal metropolitan regions in Africa and Asia will be larger than those in the developed countries and will have larger areas exposed to flooding. By 2030, India, Southern Asia, and South-eastern Asia are expected to have almost three-quarters of the urban land under high-frequency flood risk (7).
Urban areas in dry-lands
The urban extent in dry lands will also increase strongly. Across all regions and all dry-lands, most urban expansion is expected to occur in semi-arid regions of China. The most urban expansion in hyper-arid regions is expected in Northern Africa. For Western Europe the (projected) percent urban extent in dry-lands are 5% (2000) and 6% (2030), and for Eastern Europe 11% in both years (7).
Urban areas exposed to both floods and droughts
Overall, without factoring in the potential impacts from climate change, the extent of urban areas exposed to flood and drought hazards will increase, respectively, 2.7 and almost 2 times by 2030. Globally, urban land exposed to both floods and droughts is expected to increase over 250%. In particular, Southern Asia, India, and South America used to have the most urban land in 2000 in areas exposed to both frequent floods and to recurrent droughts. By 2030, Mid-Latitudinal Africa is expected to join these three regions in having the largest urban extents exposed to both floods and droughts. The largest increase in the amount of urban land exposed to both floods and droughts is expected in Southern Asia (7).
Vulnerabilities - Global assessment climate change impacts flood risk
Projections flood frequency
Global river flood risk for the end of this century has been projected based on the outputs of 11 global climate models and a number of scenarios of changes in the concentration of greenhouse gasses (1). Flood risk change was expressed in the change of the return period from 1971-2000 (current situation) to 2071-2100 (future projection) of a river discharge having a 100-year return period in 1971-2000. This projection provides the potential risk of flooding, irrespective of non-climatic factors such as land-use changes, river improvements or flood mitigation efforts such as the construction of dams.
According to these results, the frequency of occurrence increases (the return period decreases) across large areas of South Asia, Southeast Asia, Northeast Eurasia, eastern and low-latitude Africa, and South America. For Europe, according to these results, flood frequency increases in West and Northwest Europe, including the UK, Ireland, the Low Countries and most of France. In contrast, flood frequency decreases in many regions of northern and eastern Europe, Anatolia, Central Asia, central North America and southern South America. Globally, flood frequency increases in 42% and decreases in 18% of the land grid cells. The spread in the results of the return periods for different models is large, however, and generally stretches out from an increase to a decrease of flood frequency. Remarkably, all projections for the Rhine River indicate an increase of the frequency of the discharge that has a 100-year return period in 1971-2000 (1).
Projections flood risk
Global warming will intensify the hydrological cycle of the planet, leading to more extreme rainfall accumulation and a widespread increase in river flood risk in all continents. Under a high-end scenario of climate change, global flood impacts are projected to rise at an average rate of 2.4 million people and 3 billion EUR per year, exceeding a fourfold increase in flood risk by the end of the century due to climate change only (33).
These projections summarize the findings of a study on flood risk in large river basins with an upstream drainage area larger than 5000 km2, for all continents, and for the world as a whole, for three specific warming levels: 1.5°C, 2°C, and 4°C compared to preindustrial levels (33). The two lower levels refer to the Paris Agreement where it was agreed to join efforts in keeping the increase in the global average temperature to well below 2°C above preindustrial levels and to pursue efforts to limit the temperature increase to 1.5°C. In this study climatic projections from seven downscaled general circulation models (GCMs) were used to estimate changes in the expected damage and population affected by river floods under these three warming levels. The study focused on climate change effects only: effects of socioeconomic changes (population, GDP, land use) were not included. Impact estimates refer to population estimates of 2015 and damage in EUR in 2010 values. Flood protection standards (34) were included. With respect to the relation between inundation depth and the corresponding direct economic damage per unit surface (35), a distinction was made between five sectors: residential, commercial, industrial, infrastructures, and agriculture. Impacts from flash floods, pluvial floods, and coastal floods were not included.
The study includes regions representing 73% of the world population and 79% of the global GDP and takes the impact over the baseline period 1976-2005 as a reference. Central estimates of global flood risk in this baseline period total 54 million people affected and 58 billion EUR (75 billion USD) of damage per year. The findings of this study indicate that, compared with 1976-2005, the expected damage and population affected by river floods increase by 120% and 100% in the study area, respectively, when global warming rises by 1.5°C compared to preindustrial levels. For 2°C global warming the estimated increases for expected damage and population affected by river floods are 170%. For 4°C global warming the numbers are even 500% increase in damage and 580% increase in population affected (33). Somewhat lower estimates of the number of people exposed to river flooding annually were presented by (46): about a 100% increase (from ca 25 million to ca 50 million people/year, median estimate) for 2°C global warming by 2050 (staying more or less the same from 2050 to 2100). The projection of (46) for 5°C global warming compares quite well to the results of (33): about a 600% increase (to ca 150 million people/year, median estimate) for 5°C global warming by 2100.
Currently, flood risk is relatively high in Asia and Africa, which together account for 95% of people affected and 73% of the economic damage for the baseline period 1976-2005. Largest future increase in flood risk was found for U.S., Asia, and Europe. Largest absolute impacts were found in China, where current estimates of 9 million people affected and 25 billion EUR damage per year are projected to rise with global warming, reaching 40 million and 110 billion EUR per year at 4°C warming. A strong (more than 20-fold) increase in flood risk was also found for India and Bangladesh at 4°C warming, which puts them in the top of countries with largest population affected. Remarkably, projected increase in flood risk is relatively high for the European Union as a whole as well, in spite of the relatively high standards of flood protection in the EU countries (33).
Vulnerabilities - Global assessment peak flows 1.5 °C and 2 °C global warming
The Paris targets agreed upon in 2015 aim to limit global warming to ‘well below 2 °C and to pursue efforts to limit the temperature increase to 1.5 °C above pre-industrial levels’. The impacts of these global warming scenarios on global peak river flows was studied for peak flows with a 100 year return period (55). In this study, daily river flow data were calculated from a multi-model ensemble of four atmosphere-only general circulation models. This was done for three 10 year simulation periods: (1) the reference period from 2006 to 2015, (2) a future decade that is about 1.5 °C warmer than pre-industrial levels, and (3) a future decade that is about 2.0°C warmer than pre-industrial levels.
In most world regions peak river flows will occur more frequent
According to this study, in most regions of the world peak river flows that currently occur once every 100 years occur more frequent at 1.5 °C global warming, and this frequency further increases at 2.0 °C warming. These regions include South America (Amazon catchment), central Africa, central-western Europe, the south of the US (Mississippi river area), central Asia, and Siberian catchments. In these regions, the frequency of the current 1-in-100 year flow typically shifts to once in 70-90 years at 1.5 °C, and once in 50 years at 2.0 °C global warming. The current 1-in-100 year flow becomes even more common in most of central and eastern China, the southern part of the Amazon catchment, the Blue Nile, and in northern India: once in 50-60 years at 1.5 °C warming, and once in 25-35 years at 2 °C warming. Frequency changes are small for North America and eastern Europe.
In some world regions peak river flows will occur less frequent
Some world regions see a decrease in the frequency of high flows between the present day and 1.5 °C or 2.0 °C warming. These regions are located in the northern part of South America, areas of the western coast of the United States (Colorado River) and Canada, South Africa (Orange River), Scandinavia, and in some parts of eastern Europe. In most of these regions, according to this study, the current 1-in-100 year flow decreases in frequency to approximately 1-in-150 years, with little further decrease at 2.0 °C.
Why peak river flow frequency will change
The authors of this study relate these changes to a decreasing role of snowmelt in the generation of high flows as snow accumulation reduces, although there are some exceptions (South and East Coast of the US and the south of South America), and a more dominant role of extreme precipitation. The latter results in a progressive increase in the frequency of high flows in large parts of the world, despite a reduction in the number of rain days. This suggests an intensification of hydrological extremes in these regions. In fact, more frequent peak river flows and more frequent droughts can go hand in hand: in some world regions both an increase in high flow frequencies and a decrease in mean river flow accompanied by major drought hazards is projected for 2.0 °C global warming (56).
Vulnerabilities - Global assessment risks 1.5 °C, 2 °C and 3 °C global warming
Future projections of the impact of climate change on river flooding at a global scale often focus on changes in vulnerability or exposure, and in direct economic damage. Global estimates on loss of human lives, and projections of economic damage that also include indirect economic effects (welfare losses) are still scarce, however. A recent study published in Nature Climate Change fills the gap (57). In this study, population exposed to flooding, numbers of casualties, direct economic damage, and subsequent indirect impacts (welfare losses) have been estimated under a range of temperature (1.5°C, 2°C and 3°C global warming) and socio-economic scenarios. Current vulnerability levels were kept constant, and no future adaptation was assumed, since it is practically impossible to estimate how progress on adaptation will take place in different parts of the world. The estimates are based on two scenarios of socioeconomic growth. According to the authors, their results ‘offer the most complete picture of the consequences of floods on society’.
Population exposed to flooding
At present (reference period (1976-2005 ), each year about 58 million people are exposed to river flooding globally. This number would increase by 50% to 60% if the ambitious mitigation target of 1.5°C is reached, and by 76% to 102% in a 2°C warmer world. At 3°C warming, the increase would be 120% to 188%. These ranges reflect the outcomes of the two scenarios of socioeconomic growth (57).
Numbers of fatalities
Global flood mortality shows a more pronounced rise with warming. In the current climate, global river flood mortality is nearly 5,700 fatalities per year. Under 1.5°C global warming this mortality would increase by 70% - 83% (9,700 - 10,400 casualties). For 2°C global warming an increase by 103% - 134% (11,500 - 13,300 casualties) is estimated. At 3°C warming, the increase would be 180% - 265% (15,900 - 20,800 casualties) (57).
At present, the scientists estimate global direct river flood damages to be €110 billion per year on average. The three levels of global warming would increase this damage by 160% - 240% (1.5°C), 320% - 520% (2°C), and 620% - 1,000% (3°C), respectively (57).
The increase in direct flood damages leads to welfare losses for all regions at all warming levels. The global welfare reduction with respect to a scenario without climate change is projected to reach at least 0.27%, 0.40% and 0.53% at 1.5, 2 and 3°C warming, respectively. Welfare losses strongly differ from one region to another. Some advanced economies, such as Japan, South Korea and North America, are barely affected, whereas highly populated developing regions, such as China and south Asia (including India), may undergo welfare losses much higher than the global average. In fact, the greatest overall flood impacts are projected forAsia at all analysed warming levels (57).
Vulnerabilities - Global assessment climate change impacts stream flow seasonality
River discharge volumes generally vary throughout the year following a seasonal pattern. This pattern characterizes a certain river. Timing and magnitude of high-flow and low-flow periods result from local seasonal cycles of precipitation, evaporation demand, and snow accumulation and melt, and are characteristic for a certain region and thus a certain river basin. Climate change may alter the seasonal stream flow regime in many regions through a variety of (potentially interfering) mechanisms including shifts in the temporal and spatial precipitation pattern, changes in snowmelt timing due to rising temperature, or increasing evaporation demand (36).
For eleven large river basins climate change impacts on stream flow seasonality have been assessed between the present-day situation (1981-2010) and end-century conditions (2070-2099). These basins include Rhine and Tagus in Europe, upper Amazon in South America, upper Mississippi in North America, Lena, Ganges, upper Yellow, and upper Yangtze in Asia, Blue Nile and Niger in Africa, and Darling in Australia (36). The assessment was based on a large number of climate projections from five global circulation models (GCMs) under four scenarios of climate change (the concentration pathways RCP2.6, RCP4.5, RCP6.0, RCP8.5), feeding into nine regional hydrological models. The findings are changes in long-term average monthly stream flow in the individual basins. Only shifts of at least one month could be inferred; seasonality shifts at sub-monthly time scales could not be identified within the current study design.
Stream flow seasonality hardly shifts
The findings of this study indicate that stream flow seasonality hardly changes between now and the end of the century for most of the studied river basins (36). The timing of high-flow and low-flow periods more or less stays the same. The Lena and Niger are an exception. The Lena, an arctic river, strongly reacts to spring-summer snow melting; projected increases in air temperatures in the northern latitudes lead to advancing of the snowmelt season and an earlier beginning of the flood season of the northern rivers. In the Niger basin stream flow seasonality is projected to shift due to a later start of the wet season over the semi-arid African Sahel due decreasing precipitation amounts in July and August (37).
Current stream flow seasonality is amplified
In many basins, the current seasonality of stream flow seems to be amplified by climate change, however. All basins influenced by monsoonal precipitation (Ganges, Yangtze, Yellow) consistently show increased stream flow volumes during the high-flow season. In the Lena basin increased snow accumulation leads to an amplification of the snowmelt flood peak. In the Rhine basin, model results for the high-end scenario of climate change indicate increasing winter stream flow and decreasing summer low-flows. For the low-end scenario of climate change stream flow doesn’t seem to change in the summer low-flow period while winter high-flows are projected to increase. A substantial decrease of stream flow volume is projected for the Tagus basin for all months under this high-end scenario of climate change due to decreasing precipitation and increasing evaporation. For the low-end scenario of climate change little or no stream flow changes are projected for the Tagus (36).
Direct human impacts on rivers may exceed impacts climate change
The authors stress that uncertainties in these model projections are high for all studied basins due to uncertainties in climate models, hydrological models, and scenarios of climate change. Besides, the study only focused on climate change impacts on the natural stream flow regime. In many regions of the world, the natural flow regime is already significantly altered by human activity through e.g. flow regulation, water abstraction and transfer, and land-use change (38). Projected population growth and economic development are expected to increase the demand for land and water resources, and will probably further intensify human interference on the stream flow regime. According to the authors of this study, these direct anthropogenic impacts may exceed the projected climate-induced alterations (39).
EU assessment for a constant protection level for all countries
Without flood defences, almost 6 % of the European population would be living in the 100 year flood area (coastal and river floods) and the corresponding economic loss could be €236 billion (data for 2010). Estimated flood protection reduces economic damage substantially by 67 to 99 % and the number of people flooded is reduced by 37 to 99 % for the 100 year event (6).
The impact of climate change and socio-economic development in the European Union has been assessed under the assumption of uniform flood protection levels up to the 50-, 100- and 250-year flood event. This was done for 12 climate experiments derived from a combination of 4 GCMs and 7 RCMs, covering the period 1961–2100, under a medium–high emission scenario (A1B) (2).
Under the assumption of protection against a 100-year flood event, the current EU expected annual population affected is ca. 200,000. In the absence of measures to keep flood protection at the 100-year flood level, the combined effect of climate and demographic changes would increase this number to approximately 360,000 people by the 2080s. At a 100-year flood protection level, presently less than 0.1% of the population in all countries is annually affected by floods. By the end of this century, the highest average relative impact on population at this flood protection level is projected for Austria (0.15%), Hungary (0.13%), the Netherlands (0.13%), Slovenia (0.18%) and the UK (0.13%). In the Netherlands, however, protection standards by far outweigh the assumed 100-year protection level; hence the true number of people affected in the Netherlands is likely overestimated (2).
Assuming protection against a 100-year flood event, the estimated EU expected annual damage for the period 1961–1990 is €5.5 billion. In the absence of measures to keep flood protection at the 100-year flood level (no adaptation), annual damages are projected to reach €98 billion/year (constant 2006 prices, undiscounted) by the 2080s due to the combined effects of socio-economic and climate change. The largest share of these damages arises from socio-economic development. With adaptation, keeping flood protection at the 100-year protection level, the avoided damages (benefits) would amount to €53 billion/year by the 2080s. Assuming less stringent protection up to the current 50-year event, EU damages amount to €174.2 billion/year by the end of this century. For a protection level equal to the current 250-year event, on the other hand, EU damages would total €41.3 billion/year by the 2080s (2).
Modelling the potential increase in flood risk for Europe by the 2050s suggests that economic losses could increase from an estimated 4.2 billion euros per year in the 2000–2012 period, to 23.5 billion in the 2050s, under a business-as-usual emissions scenario (3). Interestingly, the bulk of the increase (about two-thirds) could be due to socio-economic development rather than climate change itself (3).
For most countries present damages under a 100-year flood protection level are well below 0.5% of the national GDP. In general, higher relative impacts are observed in Eastern European countries, especially in Hungary and Slovakia (0.8% and 0.6%, respectively). By the end of this century, relative economic impacts are projected to increase for all EU countries except Poland and the Baltic States. In relative terms, Eastern European countries will still be most severely affected by flooding, especially Hungary (1.36%), but also Slovakia (0.87%), the Czech Republic (0.81%), and Romania (0.79%) (2).
About 82.3% of the damages relate to residential areas, 6.8% to industry, 4.9% to commerce, 4.7% to agriculture and only 1.3% to industry (2).
Benefits of upgraded flood protection
At the country level significant benefits were identified for the United Kingdom, France, Italy and Hungary from upgrading protection levels to the future 100- year flood event. However, also Romania, the Czech Republic, Slovakia, Slovenia and Bulgaria would see large benefits relative to their GDP. Some countries in Eastern Europe would potentially have to spend a significant share of their current GDP to abate the future impacts from flooding in view of socio-economic and climate changes – notably Slovenia (0.7%), Bulgaria (0.6%), Romania and Hungary (both close to 0.45% of current GDP) (2).
European assessment for realistic protection levels for each country
Estimates based on a combination of climate and flooding models indicate that river floods affect some 216,000 people every year in Europe (14). The estimated annual damage for Europe is 5.3 B€ (referring to the period 1976–2005). Climate change may strongly increase the annual damage and number of exposed people. Under an upper-end scenario of climate change (the high level RCP 8.5 scenario of greenhouse gas concentration, corresponding to over 4°C warming before 2100), the socio-economic impact of river floods in Europe is projected to increase by an average of 220% by the end of the century, due to climate change only (14).
Central estimates of population annually affected, both due to climate change and socio-economic developments, are within 500,000 and 640,000 in 2050 and within 540,000 and 950,000 in 2080. Larger variability is foreseen in the future economic growth and consequently in the expected damage of flooding, with central estimates at 20–40 B€ in 2050 and 30–100 B€ per year in 2080 (14). These results are based on realistic flood protection levels in European countries as mapped by (15).
According to these calculations, the increase of expected annual damage in 2080 is highest for Italy, Hungary, Austria and Slovakia. The projected increase of the number of annually affected people is highest for Belgium, Austria, Slovenia and Slovakia. In Finland and in Lithuania, annual damage and number of people affected may decrease between now and 2080 due to a reduction in snowmelt- driven floods (14). The estimates above refer to large rivers only; the impact due to flash floods, surface water flooding and coastal floods is not accounted for.
Differences in the assumed flood protection standards dramatically change the results of impact assessments. The use of a sound map of flood protection levels (15) is a key component that improves the presented evaluation of the flood risk at national level, as compared to assuming a constant protection level for all of Europe (2: see above).
Economic impacts of climate change on river floods
Without adaptation the expected annual damage due to river floods in the EU at 2 °C global warming is estimated to rise from approximately EUR 4–5 billion/year (currently) to EUR 32 billion/year by the middle of the century (current values, undiscounted). Half of this increase is attributable to climate change, and the other half to socio-economic development (43).
How to adapt to higher river flood peaks?
River flood risk increases due to climate change and socio-economic developments unless effective adaptation measures are taken. Flood risk is the combination of hazard, exposure and vulnerability. Hazard is the extent of the flood, exposure the population and economic assets located in hazard-prone areas, and vulnerability the susceptibility of population and assets to the flood. Adaptation reduces flood risk by acting on one of these risk components. The effectiveness of four types of adaptation measures has been assessed for a scenario of 4°C global warming by the end of the century and rapid economic growth (22). These four measures are
Increase of flood protection levels, for instance by raising dikes.
- Reduction of peak flows, by setting up areas within or aside the river network that can be flooded in a controlled manner when the river stage reaches critical levels, by using reservoirs and basins to (temporarily) store or retard part of the flood discharge, or by afforestation and river renaturation.
- Reduction of vulnerability, including the implementation of early warning systems, dry and wet flood proofing, and floating buildings (23). This measure does not reduce the frequency of flooding events but rather the consequences of the flooding (number of people or value of assets effected).
- Relocation: reducing the exposure of people and assets at risk of flooding by moving them to areas with negligible risk.
The assessment shows that only increasing the level of flood protection is not sustainable in the long term, even though this measure is often highly cost-effective. This is largely due to the fact that a higher level of flood protection results in the loss of flood memory, where people and business increasingly settle in flood-prone areas (again) because the last flood is already so long ago, thus increasing exposure (24). This is called the ‘levee effect’: strong dikes ‘attract’ people and investments, and floods become more catastrophic if dikes do fail.
On the other hand, empirical evidence suggests that recurrent flooding is usually associated with decreasing vulnerability (25), due to the enhanced resilience and coping capacity acquired by the society during previous events. This so-called ‘adaptation effect’ is the reverse of the ‘levee effect’.
According to this assessment, adaptation efforts should give priority to measures targeted at reducing the consequences of hazardous events, rather than trying to avoid their occurrence. In particular, relocation and vulnerability reduction measures should be further developed. These measures have two interesting features. First, they reduce the impacts of all floods and thus strengthen the resilience of societies and ultimately the ‘adaptation effect’. Second, uncertainties in future climate projections (strongly related to hazard) do not complicate the design of these adaptation measures (focused on resilience) (22).
Flood risk reduction via land-use planning
Zoning policies can be used to limit the exposure to flooding of people and assets. Zoning regulations entail the determination of areas with a certain flood risk (i.e. the 100-year flood zone) and setting up certain land-use requirements for these zones. Such requirements could constitute, for instance, a complete ban, restricting certain uses, requiring certain building standards, giving recommendations and providing information to inhabitants in certain zones (8). In many countries (e.g. Germany, The Netherlands, UK), municipalities play an important role in flood risk management as they can specify measures for the minimisation of the damage potential for flood-prone areas in land-use plans (9). Their land- use plans tell which land use is allowed on each plot, and flood issues could, theoretically, be incorporated, but this is not always the case in practice.
Spatial planning can also play a role in limiting fatalities by optimising the possibility to reach safe places in case of flooding, be it within the flooded region (vertical evacuation, for instance to higher floors or designated flood shelters) or out of the affected region (horizontal evacuation). In addition, spatial planning can facilitate the evacuation of people away from threatened areas by making sure the main road network is elevated and thus able to be used longer in case of flooding (10). Old levees or local embankments can potentially be used for this and may have an extra compartmentalisation effect (11). Such compartmentalisation could limit the flood extent and thus fatalities and damage as well (10).
Flood risk reduction via private damage-reducing measures
Two types of building precautionary measures aim at minimising damage (12):
- wet flood proofing: flood-adapted use and equipment of buildings. Examples of wet flood proofing are the following: to adapt the building use, which means that cellars and endangered floors are not used cost intensively; to adapt the interior fitting which means that in endangered floors only waterproofed building material and movable small interior decoration and furniture are used; or to safeguard possible sources of contamination, such as an oil tank of a heating system.
- dry flood proofing: sealing, reinforcement and shielding. Examples of dry flood proofing measures are: adapting the building structure via an elevated configuration; to waterproof seal the cellar, e.g. by constructing the basis and walls of buildings out of concrete that is non-permeable; or to deploy mobile flood barriers such as temporary flood guards.
The references below are cited in full in a separate map 'References'. Please click here if you are looking for the full references for Europe.
- Hirabayashi et al. (2013)
- Rojas et al. (2014)
- Jongman et al. (2014), in: Huntingford et al. (2014)
- IPCC (2014)
- Kundzewicz et al. (2013), in: IPCC (2014)
- Mokrech et al. (2015)
- Güneralp et al. (2015)
- Merz et al. (2007), in: Kreibich et al. (2015)
- Böhm et al. (2004), in: Kreibich et al. (2015)
- Kreibich et al. (2015)
- Klijn et al. (2010); Koks et al. (2014), both in: Kreibich et al. (2015)
- ICPR (2002), in: Kreibich et al. (2015)
- Mediero et al. (2015)
- Alfieri et al. (2015)
- Jongman et al. (2014)
- Global Facility for Disaster Reduction and Recovery (2016)
- www.emdat.be, in: Global Facility for Disaster Reduction and Recovery (2016)
- IPCC (2012), in: Global Facility for Disaster Reduction and Recovery (2016)
- Kundzewicz et al. (2014); Visser et al. (2014), both in: Global Facility for Disaster Reduction and Recovery (2016)
- Jongman et al. (2012), in: Global Facility for Disaster Reduction and Recovery (2016)
- Veldkamp et al. (2015), in: Global Facility for Disaster Reduction and Recovery (2016)
- Alfieri et al. (2016)
- Strangfeld and Stopp (2014); Kreibich et al. (2015); Pappenberger et al. (2015), all in: Alfieri et al. (2016)
- Di Baldassarre et al. (2015), in: Alfieri et al. (2016)
- Wind et al. (1999); Kreibich and Thieken (2009); Jongman et al. (2015), all in: Alfieri et al. (2016)
- Forzieri et al. (2016)
- Guha-Sapir et al. (2014); NatCatSERVICE (2015), both in: Forzieri et al. (2016)
- Forzieri et al. (2014), in: Forzieri et al. (2016)
- Migliavacca et al. (2013a), in: Forzieri et al. (2016)
- Rojas et al. (2012), in: Forzieri et al. (2016)
- Nikulin et al. (2011); Outten and Esau (2013), both in: Forzieri et al. (2016)
- Hundecha et al. (2016)
- Alfieri et al. (2017)
- Scussolini et al. (2016), in: Alfieri et al. (2017)
- De Moel et al. (2016), in: Alfieri et al. (2017)
- Eisner et al. (2017)
- Biasutti and Sobel (2009); Patricola and Cook (2010), both in: Eisner et al. (2017)
- Döll et al. (2009); Grill et al. (2015), both in: Eisner et al. (2017)
- Arrigoni et al. (2010); Wang and Hejazi (2011); Patterson et al. (2013); Haddeland et al. (2014), all in: Eisner et al. (2017)
- Blöschl et al. (2017)
- European Environment Agency (2017)
- Roudier et al. (2016), in: European Environment Agency (2017)
- Hodgkins et al. (2017)
- Hartmann et al. (2013), in: Hodgkins et al. (2017)
- Milly et al. (2002), in: Hodgkins et al. (2017)
- Lowe et al. (2017)
- Thober et al. (2018)
- Andréasson et al. (2004); Arheimer and Lindstrom (2015); Alfieri et al. (2015); Roudier et al. (2016); Donnelly et al. (2017), all in: Thober et al. (2018)
- Rojas et al. (2012); Alfieri et al. (2015), both in: Thober et al. (2018)
- Roudier et al. (2016); Donnelly et al. (2017), both in: Thober et al. (2018)
- Rajczak et al. (2013); Alfieri et al. (2015), both in: Thober et al. (2018)
- Kay and Jones (2012); Alfieri et al. (2015), both in: Thober et al. (2018)
- Gosling et al. (2016); Alfieri et al. (2015); Rojas et al. (2012), all in: Thober et al. (2018)
- Guha-Sapir et al. (2017), in: Thober et al. (2018)
- Paltan et al. (2018)
- Schewe et al. (2014); Lehner et al. (2017), in: Paltan et al. (2018)
- Dottori et al. (2018)
- Berghuijs et al. (2019)
- Blöschl et al. (2017); Hall et al. (2014, 2015), all in: Berghuijs et al. (2019)
- Berghuijs et al. (2016); Blöschl et al. (2017); Stephens et al. (2015), all in: Berghuijs et al. (2019)
- Amatulli et al. (2018), in: Berghuijs et al. (2019)
- IPCC (2019a)
- Ganguli and Merz (2019)
- Ward et al. (2018), in: Ganguli and Merz (2019)
- Kew et al. (2013); Paprotny et al. (2018); Petroliagkis et al. (2016); Reeve et al. (2008); Svensson and Jones (2001); Svensson and Jones (2004), all in: Ganguli and Merz (2019)
- Zscheischler et al. (2018), in: Ganguli and Merz (2019)
- Kemter et al. (2020)
- Paprotny et al. (2018), in: Ludlow and McGovern (2020)
- Slater et al. (2021)