Introduction

Cities are home to about 57% of the world’s population and are key to achieving the United Nations’ Sustainable Development Goals (UN 2015, 2018). Water is essential for the safe and stable operation of urban systems and the security of water supply is fundamental to urban sustainability (Flörke et al. 2018; Koop et al. 2022; McDonald et al. 2011b). However, human pressure and climate change are seriously impacting both the quantity and quality of urban water supplies (Chung et al. 2021; He et al. 2021; McDonald et al. 2016). In terms of quantity, urban population growth, socioeconomic development, and booming industrial, agricultural, and domestic water demands are leading to a continuous increase in water scarcity (Hanasaki et al. 2013; He et al. 2021; Wada et al. 2016). In addition, climate change is changing the spatial and temporal distribution of water resources, challenging the reliability and security of urban water supply worldwide (Dong et al. 2020; Greve et al. 2018). In terms of quality, a wide variety of substances, including nutrients, chemicals, sediments, etc., from human activities and landscape dynamics such as urbanization, manufacturing and heavy industry, agriculture, and mining are threatening the quality of downstream urban water supplies (Best 2018; Herrera et al. 2017; Veldkamp et al. 2017). Water quantity and quality are also interrelated as water scarcity increases the concentration of pollutants, while water pollution reduces the availability of clean water (McDonald et al. 2011a; van Vliet et al. 2021; Wan et al. 2016). Hence, it is urgent to comprehensively assess the impacts of human pressures and climate change on global urban water scarcity and quality risks in an integrated way, to identify effective measures to maintain urban water security, safeguard the livelihoods of urban dwellers, and achieve the UN Sustainable Development Goals especially Goals 6 Clean Water and Sanitation, 14 Life Below Water, and 15 Life on Land.

In recent years, several studies have assessed the security of global urban water supplies. However, most studies have focused on a limited number of cities and do not provide a comprehensive picture of global urban water security risks. The few global-scale studies that do exist also have shortcomings. For example, Padowski and Gorelick (2014) assessed the water supply vulnerability of 70 large cities worldwide in 2010 and 2040; Flörke et al. (2018) assessed the water scarcity of 482 cities between the baseline (1971–2000) and the 2050s; He et al. (2021) assessed global urban water scarcity in 2016 and 2050. But these studies did not consider the quality risk to urban water security. Although studies by McDonald et al. (2011a), 2016) did assess the quality, these studies analyzed historical periods only. An integrated assessment of scarcity and quality risks to future water security of the world’s cities is now urgently required.

In this study, we comprehensively assess the water scarcity and quality risks for the world’s large cities (i.e., population > 1 million) for 2015 and 2050. First, we assess the water scarcity of 304 large cities according to water demand and water availability, and evaluate the quality of water supply in terms of the population density, cropland fertilization, and landscape patterns in source watersheds. Then, the impacts of human pressures and climate change on urban water risks are further analyzed. Finally, we discuss the implications of our findings for relevant policies and landscape sustainability.

Materials and methods

Estimation of risks to urban water security

We quantified the risks to water security for the world’s large cities for circa 2015 and 2050 in terms of both scarcity and quality (Fig. 1). To assess the future risks, we adopted the scenario framework from the Scenario Model Intercomparison Project (ScenarioMIP), part of the International Coupled Model Intercomparison Project Phase 6 (CMIP6) (O’Neill et al. 2016). The ScenarioMIP has been developed to better link the Shared Socioeconomic Pathways (SSPs) and Representative Concentration Pathways (RCPs) to support research on climatic and socioeconomic interactions. We selected the four ScenarioMIP Tier 1 scenarios (i.e., SSP1&RCP2.6, SSP2&RCP4.5, SSP3&RCP7.0, and SSP5&RCP8.5). Among them, SSP1&RCP2.6 represents the sustainable development pathway of low radiative forcing level, low climate change mitigation challenges, and low social vulnerability. SSP2&RCP4.5 represents the business-as-usual pathway of moderate radiative forcing and social vulnerability. SSP3&RCP7.0 represents a higher level of radiative forcing and high social vulnerability. SSP5&RCP8.5 represents a rapid development pathway and very high radiative forcing (O’Neill et al. 2016).

Fig. 1
figure 1

Flow chart.The bold text represents the key steps. The non-bold text represents input or output of these steps. UN United Nations; CWM city water map; UWB urban water blueprint

Full size image

First, we measured the scarcity risk of each city based on the water stress index (WSI) of the catchment in which the city is located (Table 1). The WSI equals the ratio of water use to water supply and has been widely used to quantify water scarcity (Mekonnen and Hoekstra 2016; Wada et al. 2014). Limited by data availability, we directly used the WSI in 2014 calculated by He et al. (2021) to quantify the water scarcity risk, and used the WSI in 2050 under four scenarios of SSP1&RCP2.6, SSP2&RCP4.5, SSP3&RCP7.0, and SSP5&RCP8.5 from He et al. (2021) to quantify the future water scarcity risk. The WSI in 2014 from He et al. (2021) was calculated on the basis of the AQUEDUCT 3.0 dataset published by the World Resources Institute (Hofste et al. 2019), in which the effects of reservoir storage, inter-basin transfers, groundwater recharge and upstream depletion on downstream water supply were considered. The WSI for 2050 was estimated based on the simulation results of global water demand and the simulation results of runoff under 10 global climate models (GCMs), taking into account the differences in socioeconomic development under four scenarios and the uncertainty in water supply from various models. Following previous studies (He et al. 2021; Hofste et al. 2019; Mekonnen and Hoekstra 2016), we classified the scarcity risk into four levels: low (i.e., no risk) (WSI < 0.4), medium (0.4 ≤ WSI < 0.8), high (0.8 ≤ WSI < 1.0) and very high (WSI > 1.0) (i.e., water demand exceeds water supply).

Table 1 Details of the data sources used in this study
Full size table

Second, we measured the water quality risk of each city based on the intensity of human pressures in its water source areas (McDonald et al. 2016). Since the widely used urban water source dataset – City Water Map does not fully collect information on urban water sources in China (McDonald et al. 2014), we supplemented water sources for 93 large cities in China with information from the Chinese Urban Water Blueprint and the List of Important Drinking Water Sources in China (MWR 2016; Shemie et al. 2016). With this modification, we obtained information on surface water sources for a total of 304 large cities worldwide, which accounted for 59.26% of all the world’s large cities (Supplementary Table 1). We considered four elements that affect water quality for which both historical and future data are available: population dynamics, urban expansion, crop fertilization, and change in pasture land. Based on these four factors, we modified the human footprint index (HFI) (Sanderson et al. 2002; Shrestha et al. 2021; Venter et al. 2016), which provides a comprehensive measure of the intensity of human pressure, and used it as an indicator of water quality risk. The modified HFI for the i-th urban water source in the n-th year (\({HFI}_{i}^{n}\)) is equal to the sum of weighted population density index (\({PDI}_{i}^{n}\)), urban land proportion (\({UL}_{i}^{n}\)), crop fertilization index (\({CFI}_{i}^{n}\)), and pasture land proportion (\({PL}_{i}^{n}\)), which can be expressed as:

$$HFI_{i}^{n} = W_{{PD}} \cdot PDI_{i}^{n} + W_{{UL}} \cdot UL_{i}^{n} + W_{{CF}} \cdot CFI_{i}^{n} + W_{{PL}} \cdot PL_{i}^{n}$$
(1)
$${PDI}_{i}^{n}=\text{m}\text{i}\text{n}(\text{1,0.3333}\cdot \text{log}\left({PD}_{i}^{n}+1\right))$$
(2)
$${CFI}_{i}^{n}=\text{m}\text{i}\text{n}(1,{CF}_{i}^{n}/{CF}_{max}^{2015})$$
(3)

where \({W}_{PD}\), \({W}_{UL}\), \({W}_{CF}\), and \({W}_{PL}\) are the weights of the four elements mentioned above and are set at 10, 10, 8 and 4 respectively according to the strength of influence of each element (Sanderson et al. 2002; Venter et al. 2016). \({PD}_{i}^{n}\) denotes the population density (persons per km2) in the urban water source area, \({CF}_{i}^{n}\) denotes the amount of crop fertilization per unit area (kg N per km2), \({CF}_{max}^{2015}\) denotes the maximum crop fertilization per unit area (kg N per km2) among all urban water sources in 2015. This weighted summation is based on the assumption that the pressures caused by these four categories of human activities can be superimposed, and the differences in the weights represent different degrees of pressure, i.e., population growth and urban expansion have the greatest pressure, cultivation has the second greatest pressure and grazing has the least pressure.

Crop fertilization (i.e., nitrogen) data and pasture land data were obtained from the Land-Use Harmonization 2 (LUH2) dataset (Table 1). Four sets of urban land data were used, sourced from LUH2, the History Database of the Global Environment (HYDE), the study by Chen et al.(2020), and the study by He et al.(2021) (Table 1). Three sets of population data were used including HYDE, Worldpop, and the Socioeconomic Data and Applications Center (SEDAC) dataset (Gao 2020) (Table 1). Where multiple datasets existed, we calculated the HFI based on the arithmetic mean of these data and analyzed the uncertainty of the HFI based on different combinations of data (4 sets of urban land data × 3 sets of population data, for a total of 12 combinations). Based on the relationship between HFI and water pollution risk categories (Fig. 2a), we classified the water quality risk into four levels (De Stefano et al. 2017): low (i.e. no risk) (HFI < 6) (less than 0.5 mg N L−1), medium (6 ≤ HFI < 8) (0.5 to 1.0 mg N L−1), high (8 ≤ HFI < 10) (1.0 to 2.0 mg N L−1) and very high (HFI > 10) (over 2.0 mg N L−1).

Fig. 2
figure 2

Validation of results on quality risk. a, correlation between modified human footprint index calculated in this study and pollution risk estimated by Transboundary Waters Assessment Programme (De Stefano et al. 2017). Dots represent the transboundary river basins. b, correlation between modified human footprint index calculated in this study and water pollution index from Numbeo pollution survey (https://www.numbeo.com/pollution/). c, correlation between modified human footprint index calculated in this study and drinking water pollution and inaccessibility index from Numbeo pollution survey. Dots represent surveyed cities. Only the cities with Numbeo survey samples > 100 were used for correlation analysis

Full size image

Based on the results of the above assessments, we obtained water scarcity risk levels and water quality risk levels for 304 large cities worldwide under four scenarios for the years 2015 and 2050 (Supplementary Table 1). We have validated the results on quality risk using simulation from a process-based model and questionnaire results based on residents’ perceptions, which testifies as to the reliability of our results (Fig. 2). Based on this information, we analyzed the current status of urban water risks and future changes in urban water risks at global, national, and city scales.

Contribution analysis

In assessing the influencing factors of water scarcity risk, we calculated the impacts of changes in water demand caused by socioeconomic development and changes in water availability caused by climate changes on the WSI. The calculation can be expressed as:

$${\varDelta WSI}_{i,WD}=\frac{{TWW}_{i}^{2050}}{{AWR}_{i}^{2014}}-{WSI}_{i}^{2014}$$
(4)
$${\varDelta WSI}_{i,WA}=\frac{{TWW}_{i}^{2014}}{{AWR}_{i}^{2050}}-{WSI}_{i}^{2014}$$
(5)

where \({\varDelta WSI}_{i,WD}\) represents the impact of changes in water demand on the WSI of the catchment of the i-th city, \({\varDelta WSI}_{i,WA}\) represents the impact of changes in water availability on the WSI of the same catchment. \({WSI}_{i}^{2014}\) represents the WSI in 2014 for this catchment. \({TWW}_{i}^{2014}\) and \({TWW}_{i}^{2050}\) represent the water demand in 2014 and 2050, respectively. \({AWR}_{i}^{2014}\) and \({AWR}_{i}^{2050}\) represent the water availability in 2014 and 2050, respectively. In assessing the influencing factors of water quality risk, we calculated the impacts of population density dynamics, urban land dynamics, crop fertilization dynamics, and pasture land dynamics on the HFI. The calculation can be expressed as:

$${\varDelta HFI}_{i,PD}={W}_{PD}\cdot ({PDI}_{i}^{2050}-{PDI}_{i}^{2015})$$
(6)
$${\varDelta HFI}_{i,UL}={W}_{UL}\cdot ({UL}_{i}^{2050}-{UL}_{i}^{2015})$$
(7)
$${\varDelta HFI}_{i,CF}={W}_{CF}\cdot({CFI}_{i}^{2050}-{CFI}_{i}^{2015})$$
(8)
$${\varDelta HFI}_{i,PL}={W}_{PL}\cdot ({PL}_{i}^{2050}-{PL}_{i}^{2015})$$
(9)

where \({\varDelta HFI}_{i,PD}\), \({\varDelta HFI}_{i,UL}\), \({\varDelta HFI}_{i,CF}\), and \({\varDelta HFI}_{i,PL}\) represent the impacts of population density dynamics, urban land dynamics, crop fertilization dynamics, and pasture land dynamics on the HFI for the water source of the i-th city, respectively. \({PDI}_{i}^{2015}\), \({UL}_{i}^{2015}\), \({CFI}_{i}^{2015}\), and \({PL}_{i}^{2015}\) represent the population density index, urban land proportion, crop fertilization index, and pasture land proportion for the water source of this city in 2015, respectively. \({PDI}_{i}^{2050}\), \({UL}_{i}^{2050}\), \({CFI}_{i}^{2050}\), and \({PL}_{i}^{2050}\) represent the above four variables in 2050, respectively.

Results

Current risks to water security for the world’s large cities

Of the 304 cities evaluated, 273 (89.8%) faced risks to water security in 2015 (Fig. 3b). Among these, 262 (86.1%) faced quality risk, 85 (27.9%) faced scarcity risk, and 74 (24.3%) faced dual risks (Fig. 3b). Furthermore, 69 (22.7%) faced very high water quality risk, 48 (15.8%) faced very high water scarcity risk, and 7 (2.3%) were exposed to both very high quality risk and very high scarcity risk (Table 2, Supplementary Table 1).

Fig. 3
figure 3

Current risk to water security of large cities. a, spatial patterns of cities facing different levels of water risk (cities with population above 10 million in 2015 were labeled). b, number and proportion of cities facing risk at the global scale. c, proportion of cities facing risk at the national scale (all countries were grouped into two income-levels, and the four countries with > 10 large cities are also listed). Please refer to Supplementary Table 1 for risk levels for all estimated cities

Full size image
Table 2 Number of cities facing risk to water security from 2015 to 2050
Full size table

Of the 27 megacities (i.e., population > 10 million) assessed, 25 (92.6%) faced risks to water security (Table 3; Fig. 3a). Among them, 23 (85.2%) faced quality risk, 10 (37.0%) faced scarcity risk, and 8 (29.6%) faced dual risks. 11 megacities including São Paulo, Mumbai, and Dhaka faced very high water quality risk, five megacities (Delhi, Beijing, Los Angeles, Moscow, and Bangalore) faced very high scarcity risk, and Bangalore faced very high risk to both water scarcity and quality.

Table 3

Table 3 Risk to water security for megacities from 2015 to 2050
Full size table

The proportion of cities facing risks to water security was relatively low in high-income and upper-middle-income countries and high in low-income and lower-middle-income countries (Fig. 3c). Of the 218 large cities assessed in high-income and upper-middle-income countries, 192 (88.1% of the total) faced at least one kind of risk and 48 (22.0%) faced dual risks. In contrast, of the 81 large cities in low-income and lower-middle-income countries, 77 (95.1%) faced at least one kind of risk and 24 (29.6%) faced dual risks. Of the countries with more than 10 large cities assessed, the United States, a high-income country, had 24 (68.6%) large cities facing at least one kind of risk and 5 (14.3%) facing dual risks. In India, which is a lower-middle-income country, all 31 cities assessed faced at least one kind of risk and 18 (58.1%) faced dual risks, which was much higher than the global average and the average for high-income countries.

Future risks to water security for large cities

From 2015 to 2050, 270–303 (88.8–99.7%) large cities were projected to face rising risks to water security, including 207–297 (68.1–97.7%) cities with rising scarcity risk, 177–236 (58.2–77.6%) cities with rising quality risk, and 114–220 (37.5–72.4%) cities with risk rising to both water scarcity and quality (Figs. 4a, 5 and 6). By 2050, the number of cities facing risks to water security was projected to be 280–285 (92.1–93.8%), including 267–271 (87.8–89.1%) cities facing quality risk, 92–126 (30.3–41.4%) cities facing scarcity risk, and 79–110 (26.0–36.2%) cities facing dual risks (Table 2, Supplementary Table 1). The numbers of large cities facing very high quality risk and very high scarcity risk were projected to be 80–100 (26.3–32.9%) and 53–77 (17.4–25.3%), respectively. 13–27 (4.3–8.9%) cities were projected to face both very high quality risk and very high scarcity risk (Table 2, Supplementary Table 1).

Fig. 4
figure 4

Changes in risk to water security of large cities from 2015 to 2050. a, changes in risk of large cities at the global scale. Bars present the average numbers of cities facing increase in risk using the ensemble mean of runoff from GCMs and the mean of population density and urban land proportion from different datasets under four scenarios. Crosses present the numbers of cities facing increase in risk using population density and urban land proportion from each dataset under four scenarios. b, changes in water scarcity risk of large cities at the national scale. c, changes in water quality risk of large cities at the national scale. Two income-levels and four countries with estimated cities > 10 were listed. Please refer to Supplementary Table 1 for changes of risk levels for all estimated cities

Full size image

25–27 (92.6–100.0%) megacities were projected to face rising risks to water security, including 19–23 (70.4–85.2%) with rising quality risk, 19–27 (70.4–100.0%) with rising scarcity risk, and 13–22 (48.1–81.5%) with risk rising to both water scarcity and quality (Fig. 5). Under the business-as-usual scenario (i.e., SSP2&RCP4.5), Bangalore, Karachi and Delhi were projected to face both very high water quality risk and very high scarcity risk, while 12 megacities including Tokyo, Paris and São Paulo were projected to face very high quality risk, and Moscow, Los Angeles, Mexico City and Beijing were projected to face very high scarcity risk (Table 3).

Fig. 5
figure 5

Changes in risk to water security of large cities from 2015 to 2050 under the four socio-economic and climate change scenarios. Cities with a population >10 million in 2015 are labeled

Full size image

The magnitude of rising risks to urban water security was higher in low-income and lower-middle-income countries than in high-income and upper-middle-income countries (Fig. 4b, c). The average changes in scarcity risk index and quality risk index for large cities in low-income and lower-middle-income countries were projected to be 0.22–1.64 and 0.41–1.20, respectively, which is much higher than that (0.08–0.52 and − 0.07–0.27) for large cities in high-income and upper-middle-income countries (Fig. 4b, c). Of the countries with more than 10 large cities assessed, India was projected to have the highest increase in risk indices, i.e., 0.06–2.03 and 0.20–1.28, respectively (Fig. 4b, c).

Factors influencing risks to water security for large cities

Increase in water demand will be the main cause of rising water scarcity risk for the estimated cities (Fig. 7). Under SSP2&RCP4.5, of the 287 cities with rising scarcity risk, 276 (96.2%) were projected to be driven by increasing water demand and only 11 cities, including Los Angeles, Delhi, Karachi, Melbourne, and Yekaterinburg, were driven by climate change. Under other scenarios, the proportion of cities with water scarcity risk driven by increasing water demand was also projected to exceed 90%.

Population growth in source watersheds was the main reason for the increase in water quality risk for cities, which was followed by cropland fertilization, urban expansion, and pastureland expansion (Fig. 7). Under the four scenarios, the numbers of cities with increasing quality risk driven by population growth in source areas were projected to be 101 (57.1%), 124 (52.5%), 107 (53.0%), and 133 (66.8%), respectively. The increase in cropland fertilization in source watersheds was the secondary reason, and the numbers of cities with it being the primary driver of risk were projected to be 47 (26.6%), 89 (37.7%), 63 (31.2%), and 34 (17.1%), respectively. Urban expansion in source watersheds was the tertiary driver, with the number of cities being 26 (14.7%), 17 (7.2%), 20 (9.9%), and 32 (16.1%). The expansion of pastureland in source watersheds made little contribution to the increase in quality risk, primarily affecting less than 6% of cities under all four scenarios.

Discussion

Major findings

This study provides a comprehensive assessment of future water supply risks in terms of both scarcity and quality for 304 large cities worldwide. Our results show that about 90% of these cities faced risks in 2015, with about three times as many cities facing quality risk (86.1%) as those facing scarcity risk (27.9%), and 24.3% of 304 cities facing dual risks. Unlike the water-scarce cities, which were predominantly located in drylands, cities facing water quality risk were widely distributed around the world (Fig. 3a). In contrast, although a large number of studies have focused on urban water scarcity, research on the quality risk for urban water supply is still in a nascent stage (Liu et al. 2022; Michalak et al. 2023; Turner et al. 2021). Our findings on the widespread quality risk to urban water security evidence the importance of in-depth research on this topic.

We found that lower income countries had a higher proportion of cities facing risks to water security (Fig. 3c). For example, nearly 60% of cities in India faced dual risks compared to a quarter of cities globally. Moreover, the vast majority of cities facing very high risk, e.g., Bangalore, Jakarta, Lagos, Manila, and Dhaka, were also located in lower-middle-income and low-income countries (Fig. 3a; Table 3). In addition, the large cities in lower income countries had higher increases in water risk indices (Fig. 4b, c). In contrast to higher income countries that can reduce water risks through solutions such as virtual water trading and water treatment, lower income countries cannot afford these measures (He et al. 2021; Rachunok and Fletcher 2023). As a result, urban water supply risks tend to be greater in lower income countries.

As a result of increasing human pressures between 2015 and 2050, 88.8–99.7% of large cities were projected to face rising risks to water security. Although the number of cities facing rising scarcity risk was similar to the number of cities facing rising quality risk, there would be obvious differences in their spatial distribution (Fig. 5). Under four scenarios, China would have more cities facing rising scarcity risk, but few cities facing rising quality risk. Cities facing rising quality risk would be mainly located in India, African, South American and North American countries. This is mainly because the future growth in China’s urban population would increase urban water demand, but the decrease in the total population would reduce the pressures on water source areas and therefore reduce the water quality risk (Fig. 4c) (Gao 2020; Wang et al. 2022). Even so, the quality risk to urban water security in China would still be a cause for concern by 2050 (Fig. 6). In contrast, the total population including urban population would increase in India, African, South American and North American countries (Gao 2020; Wang et al. 2022).

Fig. 6
figure 6

Spatial patterns of cities facing different levels of risk in 2050 under four socio-economic and climate change scenarios. Cities with population above 10 million in 2015 were labeled)

Full size image

Policy implications

The above findings show that human pressures and climate change will further increase risks to water security for the world’s large cities in the future, threatening the achievement of SDG 6 ‘Clean Water and Sanitation’ and SDG 11 ‘Sustainable Cities and Communities’. In order to address the rise in risks to urban water security, the following four directions are suggested.

Conservation of urban water source areas is urgently needed. We found that most cities faced water quality risk due to degradation of their source watersheds (Table 2) (Abell et al. 2007; McDonald et al. 2016). However, the existing protected areas mainly focus on biodiversity conservation, and these biodiversity reserves do not align well with important urban water source areas (TNC 2016). Therefore, there is an urgent need to establish protected areas for important water source areas to limit human pressures and promote ecological restoration in these areas, in order to secure urban water supply (Jung et al. 2021; Mulligan et al. 2020; TNC 2016).

Water-based urban planning and governance need to be promoted. The results of the attribution analysis show that the main reason for the rise in urban water scarcity is the increase in water demand (Fig. 7). Therefore, for water-scarce cities, water conservation and wastewater reclamation should be actively promoted, while limiting the size of urban population, industry and green space within cities are also needed in order to reduce urban water demand (Liu et al. 2020; Liu and Wu 2022). For cities such as Karachi, which are sensitive to climate change (Fig. 7), the possible impacts of climate change on water availability also need to be taken into account in urban planning.

Fig. 7
figure 7

The effects of different factors on water risk indices from 2015 to 2050 under the four socio-economic and climate change scenarios. Only top 50 cities in terms of increase in risk indices were listed

Full size image

Infrastructures that underpin urban water supply and drainage should be improved to ensure clean urban water supply and reduce urban wastewater pollution (Chung et al. 2021; Vorösmarty et al. 2021). Improving the level of purification of urban water supply facilities can effectively reduce the threat of source watershed degradation (Liu et al. 2022). The construction of reservoirs, inter-basin water transfers, desalination plants, and water reclamation plants can help cities cope with water shortages (Padowski et al. 2016). Dams can reduce sediments, improvement of agricultural infrastructures can reduce the discharge of nutrients, and wastewater treatment facilities can reduce the discharge and concentration of urban wastewater to minimize the pollutants’ impacts on rivers and downstream areas (Larsen et al. 2016; Yu et al. 2019). It is important to note that when constructing infrastructure, especially dams, reservoirs, and inter-basin transfers, the social, economic and environmental effects of the project need to be comprehensive assessed to reduce their negative impacts (WWF 2009). In addition to artificial infrastructures, nature-based solutions with environmental co-benefits are worthy of promotion, such as wetlands that can purify water and provide a range of other ecosystem services (Cohen-Shacham et al. 2016).

Transregional compensation systems for water-related services are also needed. Most of the water used by cities comes from source watersheds outside the cities’ administrative areas, or even from other countries (TNC 2016). Agricultural and industrial development in water source areas can increase the water supply risks for downstream cities (Vorosmarty et al. 2010). To enhance urban sustainability, the socioeconomic development of urban water source areas must be limited. To alleviate this inter-regional trade-off, transregional compensation has become an important approach (Gao et al. 2019; Romulo et al. 2018). Nowadays, there are still obvious deficiencies in the scope of application and standards for compensation of water-related services. In the future, it is necessary to further improve such compensation schemes at different scales according to the supply, demand, and flow of water-related services (Romulo et al. 2018).

Implications for landscape sustainability

Green spaces and water bodies are the main landscape elements that provide ecosystem services and enhance human well-being in cities (Wu 2013, 2014, 2021), so increasing the area of green spaces and water bodies is the most commonly used measure to improve urban landscape sustainability. However, this measure can increase water consumption and exacerbate urban water scarcity in drylands (Liu and Wu 2022). In particular, during extreme droughts, urban vegetation can suffer a fatal blow (Kumar et al. 2023). Thus, rather than blindly increasing the area of green spaces and water bodies in cities facing water scarcity, more attention should be paid to optimizing the spatial configuration of green spaces and water bodies so that more city dwellers can enjoy the ecosystem services they provide.

Changes in the landscape patterns of urban water sources can directly affect water-related ecosystem services, which in turn affect the quantity and quality of urban water supply. Although some urban water source areas are already under strict protection, nearly 90% of the world’s large cities still face water quality risks from degraded water source landscapes (Abell et al. 2007; McDonald et al. 2016). In addition, the scope of urban water source areas is expanding with the implementation of water transfer projects around the world (TNC 2016; WWF 2009), creating new challenges for urban water source protection and urban water security.

Therefore, future studies should pay more attention to improving the spatial configuration of green spaces and water bodies in water-scarce cities, identifying the importance of urban water source areas for urgent protection according to the supply, demand, and flow of water-related ecosystem services, and assessing the effectiveness of different protection measures, in order to provide theoretical foundations and practical ways to reduce urban water risks.

Uncertainty

In this study, uncertainties mainly affect two aspects. First, limited by data availability, we only considered four major factors influencing the water quality risk with continuous historical and future data, i.e., population growth, urban expansion, agricultural fertilization, and pasture expansion, and did not consider other sources of pollution such as atmospheric deposition, mining and oil/gas extraction (Turner et al. 2021), which would underestimate the extent of water quality risk. Second, we used a modified HFI in our assessment of water quality risk, instead of a model for simulating pollutant generation, transport and purification processes, which led to uncertainty in the results. However, the reliability of our results was validated by the results from a process-based model and the questionnaire results based on residents’ perceptions (Fig. 2). In addition, we only assessed the quality risk for urban water supply and this may not be representative of the quality of tap water used by residents, which is also affected by urban water supply infrastructure (e.g., capacity of the water treatment plant) (Liu et al. 2022).

Conclusions

Scarcity and quality risks of water supply are key elements affecting urban landscape sustainability. This study integrated the WSI and the modified HFI to comprehensively evaluate changes in water scarcity and quality risks under four scenarios of socioeconomic development and climate change for 304 large cities around the world from 2015 to 2050. About 90% of cities were found to be at risk in 2015, with three times as many cities at quality risk as that at scarcity risk. Cities in lower income countries faced higher levels of water risks. The vast majority (88.8–99.7%) of cities were projected to face rising risks between 2015 and 2050, with one-third of cities facing dual risks in 2050. Cities in lower income countries were projected to experience higher magnitude of rising risks. Increasing water demand will be the main cause of rising scarcity risk, but for a few cities, such as Los Angeles and Karachi, declining water availability due to climate change will be the main cause. The growth of population and crop fertilization in water source areas will be the main causes of rising quality risk. To address the rising water risks and improve urban landscape sustainability, conservation of urban water source areas, water-based urban planning and governance, improvement of urban water supply infrastructures, and ecological compensation regimes for water-related services are urgently needed.