1.1. Objectives

The study’s overarching goal is to analyse the resilience of groundwater systems in Asian cities to climate change and anthropogenic development.

The specific objectives are:

1. To evaluate regional climate models (RCMs) for projections of future climate change in Asian cities.
2. To forecast future land usage and land cover in Asian cities and examine possible future climatic scenarios.
3. To assess the spatiotemporal distribution of groundwater recharge in Asian cities under several climatic change scenarios and anthropogenic development.
4. To estimate the aquifers’ groundwater levels in Asian cities under various scenarios of climatic change and anthropogenic development.
5. To develop groundwater resiliency indicators that consider climate change and human growth scenarios and analyse groundwater resiliency in Asian cities.

2.1. Study area

The study area includes four rapidly expanding Asian cities: Bangkok and its vicinity, Ho Chi Minh City, Lahore, and Kathmandu Valley (Figure 1). The four Asian cities were selected based on their groundwater dependence which ranges from 30–100%.

Bangkok and its environs comprise seven provinces in Thailand: Bangkok, Samut Prakan, Samut Sakhon, Nakhon Pathom, Nonthaburi, Pathum Thani, and Phra Nakhon Si Ayutthaya. Bangkok is in the humid tropics and has a tropical wet-and-dry climate throughout the year. The average temperature is 30°C, and the annual rainfall is 1500 mm. The water supply for the city’s demand comes from both surface and groundwater sources. Groundwater is mostly used in the water supply systems near Bangkok. In the study area, 11.3 million (in 2015) people are living at a density of 300 to 3,600 people/km². Eight confined aquifers exist in Bangkok and its environs, with Phra Pradaeng, Nakhon Luang, and Nonthaburi being the most exploited because of their higher production, accessibility, and excellent water quality. In the mid-1950s, the extensive use of groundwater was started, and there was a continuous increase in groundwater use until 1997. Land subsidence (1.0 cm/yr between 2006 and 2012), groundwater level depletion and recovery, and groundwater quality deterioration are the main groundwater-related problems.

Ho Chi Minh City (HCMC), the biggest city in Vietnam, is located 1760 kilometres south of Hanoi. HCMC had a population of around 7.35 million (in 2015) and a total area of 2095 km², making up roughly 8.79% of its total populace and 0.6% of Vietnam’s total land. The average humidity is 75%, and the average temperature is 27°C. The climate is classified as tropical, more precisely tropical wet and dry. The total annual rainfall is 1946 mm, 1130 mm, lost due to evaporation. Both the surface and groundwater resources are used to deliver water. The fast growth in groundwater abstraction began in 1990 when Vietnam’s economic policies were implemented. It is planned that groundwater’s portion of the overall water supply, which was 44.7% in 2012, will be reduced to 15.5% in 2015 and 2.7% in 2020. Major groundwater-related problems include a decline in groundwater level (rate: 0.5–1.0 m/yr), saltwater intrusion in some areas, groundwater quality degradation (NO₃ content is four times the permissible value; iron content is 100 times the permissible value), and land subsidence of a few centimetres per year in heavily pumped areas.

Kathmandu, Nepal’s capital and largest metropolitan city, is in central Nepal and has a total area of 656 km². The Kathmandu Valley’s population is 2.8 million (in 2015) and the population density is 29 persons per km², which includes both the urban and rural areas of Kathmandu Valley. Kathmandu Valley is in the mild temperate zone, with an elevation range of 1205–2713 masl and a significant contrast in typical summer (28–30°C) and winter (10°C) temperatures. On average, it receives 1455 mm of rain annually, 65% of which falls from June to August. Kathmandu Valley has a closed reservoir system with a gentle slope towards the Valley’s centre. The groundwater flow is assumed to be slow, particularly in the deeper aquifer. The water supply system is hugely dependent upon monsoon rain, the stream, and the rainfed recharge system, as 25–30% of the water demand is mitigated through groundwater abstraction. Due to an unfavourable geological characteristic of the Valley, surface runoff is high in volume during the monsoon period. The natural recharge in the swallow aquifers during the high monsoon periods occurs along the basin margin, although the presence of the clay bed significantly restricts natural recharge in deep aquifers.

With a total area of 1772 km², Lahore is one of the largest metropolitan regions in South Asia. The population was 10.37 M in 2015, and a population density of 14,000 persons per km². Lahore has a subtropical semi arid climate, with an average annual rainfall of 712 mm for the 30 years from 1971 to 2000. The majority of the yearly rainfall (roughly 75%) falls between June and September, recharging the groundwater by around 40 mm on average each year. The average yearly temperature is roughly 24.3°C, with monthly variations between 33.9°C (in June) and 12.8°C (in January). The only source of water in the city is groundwater. The Water and Sanitation Agencies (WASA) tube well infrastructure is used by about 89% of Lahore’s entire land area. These tube wells, which have various capacities and are situated between 150 and 200 m deep, continuously pump 800 cubic feet of water per second (cusecs). Water levels between 2003 and 2011 are falling at a rate of 0.57 m/yr to 1.35 m/yr due to excessive groundwater consumption, and well productivity is also declining.

2.2. Methods and data

The overall methodological framework required to accomplish the stated objectives is shown in Figure 2.

“DEM is a digital elevation model; LULC is land use/cover; P is precipitation; T is Temperature; RH is Relative Humidity; WS is Wind Speed; SR is Solar Radiation; Q_obs is Observed River Discharge; Cal is Calibration; Val is Validation; GWL_obs is Observed Groundwater Level; GW is Groundwater; RCM is a Regional Climate Model; GwRe is the groundwater resiliency; GwR is groundwater recharge; GwL is groundwater level; and n is the base year”

The key objective of the study is to assess groundwater resiliency in four Asian cities; Bangkok and its vicinity, Ho Chi Minh City, Lahore, and Kathmandu Valley under climate change and human development scenarios Firstly, the evaluation of Regional Climate Models (RCMs) in each city was done based on their ability to replicate the observed climate datasets. The top 5 best-performing RCMs, along with two Representative Concentration Pathway (RCP) scenarios (RCP 4.5 and RCP 8.5), were then used to project the future climate of the study area. From the time perspective, the RCP scenarios differ in radiative forcing and greenhouse gas intensity. The main objective of the RCP scenarios is to identify future uncertainties without forecasting them, with different socioeconomic hypotheses used in their development (Moss et al., 2010; Rogelj, Meinshausen, & Knutti, 2012). RCP 4.5 is described by the Intergovernmental Panel on Climate Change (IPCC) as a moderate emission scenario in which emission peaks around 2040 and then declines. RCP 8.5 is the highest baseline emission scenario in which emission continues to rise throughout the 21st century. RCP 8.5 gives a much more rapid warming and more pronounced change in important indicators such as river flow, temperature, and precipitation. On the other hand, the RCP 2.6 scenario will result in the least amount of global warming and only limited climate change and the difference between RCP 2.6 and 4.5 will remain relatively small until the end of the century. Therefore, in this study, two scenarios: RCP 4.5 and RCP 8.5 are used. A quantile mapping technique was used to remove the bias from the climate data sets. The land use change model Dyna-CLUE was used to project the future land use change of the study area. The baseline climatic data, future projections from climate models, and land use change scenarios conceived by Dyna-CLUE were fed into the hydrological model SWAT and WetSpass to forecast future groundwater recharge. The hydrological modelling of Bangkok and its vicinity was done using the WetSpass model, whereas, for three other Asian cities, the SWAT model was used. The WetSpass model was chosen in Bangkok since it was challenging to include the city within a hydrological watershed. The future groundwater abstraction computed using population and water demand forecasting and future groundwater recharge was then input into calibrated steady-state groundwater model GMS-MODFLOW to obtain future groundwater level. Finally, based on the findings, a groundwater resilience indicator was created, which was then utilised to create a groundwater resilience map for the research region.

The data used for the study include the time series observations (rainfall, wind speed, maximum and minimum temperature, solar radiation, river discharge, relative humidity, groundwater level and groundwater abstraction), physical characteristics of catchments (DEM, soil map, land use map, slope, aspect ratio, distance from road, distance from river, population density), hydrogeologic properties of aquifers (specific storage (S_s), hydraulic conductivity (K_x, K_y), specific yield (S_y)) and future climate projection from climate models (adopted from Neupane, Shrestha, Ghimire, Mohanasundaram, & Ninsawat, 2021).

2.3. Land use change scenarios

Three land use change scenarios, namely: high urbanisation (HU), medium urbanisation (MU) and low urbanisation (LU) scenarios, were considered for all four Asian cities to project the future land use map. In high urbanisation scenarios, there will be a significant increase in built-up land or urban areas. In medium urbanisation scenarios, there will be a medium increment in a built-up area and low urbanisation scenarios, forest conservation is given priority and there will be no change in built-up land. A detailed description of the future land use scenario is presented in Supplementary Table 1.

2.4. Groundwater abstraction scenarios

To calculate future groundwater abstraction, three scenarios (like land use change scenarios) were analysed for all four Asian cities. The groundwater abstraction scenarios for Bangkok and its vicinity, Ho Chi Minh City and Lahore are mentioned below:

• High Urbanisation: Groundwater abstraction will be increased in future by 10%, 20%, 30% and 40% in 2035, 2055, 2075 and 2095, respectively (we assume that there will be very high increase in the built-up areas due to rapid population growth and hence the abstraction will increase in future).
• Medium Urbanisation: Groundwater abstraction will be decreased in future by 10%, 20%, 30% and 40% in 2035, 2055, 2075 and 2095, respectively (we assume that there will be very small increment in urban land and small reduction in agricultural land, and there will be some policies intervention to prevent groundwater abstraction or transboundary water diversion from different river basin).
• Low Urbanisation (safe-yield pumping): Groundwater abstraction will be decreased in future by 15%, 30%, 45% and 60% in 2035, 2055, 2075 and 2095, respectively (we assume that there will be no change in the area occupied by built up land or urban land and agricultural land. This suggests that the area occupied by forest will be increased, therefore in low urbanisation groundwater abstraction will be decreased in future).

Likewise, the groundwater abstraction scenarios for Kathmandu Valley are mentioned below:

• High Urbanisation: Groundwater abstraction meets 40% of overall water demand (is the pessimistic scenario where the groundwater abstraction further increases in future).
• Medium Urbanisation: Groundwater abstraction meets 20% of overall water demand (is the business as-usual scenario).
• Low Urbanisation (optimistic): Groundwater abstraction meets 10% of overall water demand (is the optimistic scenario which designates the reduction in groundwater abstraction for the implementation of policies to prevent groundwater abstraction or transboundary water diversion from different river basins).

The reason behind two different scenarios between Kathmandu and other three Asian cities is because of very less groundwater abstraction in Kathmandu as compared to other three Asian cities. And if we adopt similar methodology as of three Asian cities, there will not be much change in groundwater abstraction volume for Kathmandu.

2.5. Groundwater resiliency indicators

Based on the existing resilience document and definition (as discussed in literature review section) following methods have been suggested for the resilience indicators and resilient index calculation of a groundwater system.

2.6. Quantification of uncertainties

In climate change impact assessments studies, uncertainty quantification is an important issue because meteorological variables, such as precipitation and temperature must be projected several decades into the future. It is quite impossible to remove the uncertainties, however we can quantify it using several tools and techniques. This study also follows some uncertainty quantification techniques such as evaluation of regional climate models and finding the best RCMs for the study region. The main source of the uncertainty in climate impact analysis is emission scenarios and climate models and the uncertainty coming from the hydrological and groundwater model is very small compared to it coming from climate models. As we are reducing and quantifying the uncertainties from climate models it also helps in reducing the uncertainties from the models. We have used a physically-based hydrological model and this kind of model helps to reduce the parameter uncertainties. Also, the input data are the main source of uncertainties in modelling work. Since we have used the most efficient input data sets from respective organisations of each city, it also helps to reduce uncertainties.

3.1. Evaluation of RCMs

The evaluation of Regional Climate Models (RCMs) in each city was done based on their ability to replicate the observed climate datasets. This study assesses the capability of 21 RCMs from the Coordinated Regional Climate Downscaling Experiment (CORDEX) to simulate climate extremes in the rapidly developing Asian cities (Bangkok and its surroundings, Ho Chi Minh City, Kathmandu Valley, and Lahore) that are highly vulnerable to climate change. A detailed description of the evaluation of RCMs can be assessed by Neupane et al. (2021). Table 2 shows the list of top five better performing RCMs in four Asian cities, which in turn were used to project the future climate of respective Asian cities.

3.2. Projected future climate of Asian cities

The future climate of four Asian cities is projected using the top five better-performing RCMs obtained from the evaluation study (Table 2) for three future periods: 2010 to 2039 or near future (NF), 2040 to 2069 or mid future (MF), and 2070 to 2099 or far future (FF) relative to the baseline period (1976–2005) under two emission scenarios (RCP 4.5 and RCP 8.5). A quantile mapping technique was used to remove the biases from future climate data.

3.2.1. Temperature trends

The study’s findings indicate that future temperatures are predicted to rise in all four Asian cities for both RCPs scenarios, which aligns with the global temperature trend. By the end of the twenty-first century, the average annual maximum temperature for Bangkok and the surrounding area is predicted to rise by 0.4°C to 1.5°C under RCP 4.5 and by 0.5°C to 3.1°C under RCP 8.5. Similarly, it is anticipated that the average annual minimum temperature would rise by 0.8°C to 2°C in RCP 4.5 and 0.9°C to 4.3°C in RCP 8.5.

By 2099, the average annual maximum temperature in Ho Chi Minh City, Vietnam, is projected to rise by 0.52°C to 2.71°C under RCP 4.5 and 0.75°C to 5.13°C under RCP 8.5. Similarly, it is anticipated that the average annual minimum temperature would rise by 0.73°C to 5.13°C in RCP 4.5 and 0.97°C to 6.68°C in RCP 8.5.

The average annual maximum temperature is predicted to rise in the Kathmandu Valley by 0.22°C to 2.44°C by the end of the twenty-first century under RCP 4.5 and by 0.60°C to 5.38°C under RCP 8.5. Similar to this, it is estimated that the average annual minimum temperature would rise by 0.48°C to 3.52°C in RCP 4.5 and 0.57°C to 5.60°C in RCP 8.5.

By 2099, Lahore’s average annual maximum temperature is anticipated to rise by 0.16°C to 2.85°C under RCP 4.5 and by 0.31°C to 4.43°C under RCP 8.5. Similarly, it is anticipated that the average annual minimum temperature would rise by 0.46°C to 3.72°C in RCP 4.5 and 0.62°C to 6.07°C in RCP 8.5. The yearly maximum and lowest temperature trend for four Asian cities is shown in Figure 3.

3.2.2. Rainfall trends

As for rainfall, all four Asian cities are expected to receive more rainfall in future. All RCMs and both RCP scenarios agree that rainfall is expected to increase annually throughout all time periods. For Bangkok and its vicinity, the relative increase in future rainfall ranges from 12.04% to 36.64% for RCP 4.5 scenario and 17.19% to 57.42% for RCP 8.5 scenario. For Ho Chi Minh City, the relative increase in future rainfall ranges from 10.18% to 24.83% under RCP 4.5 scenario and 11.52% to 28.40% under RCP 8.5 scenario. Likewise, the relative increase in future rainfall for Kathmandu Valley ranges from 21.44% to 25.12% for RCP 4.5 and 18.97% to 37.41% for RCP 8.5 scenario. For Lahore, the relative increase in future rainfall ranges from 11.39% to 15.67% under RCP 4.5 scenario and 13.1% to 15.38% under RCP 8.5 scenario. The maximum increase in future rainfall is expected for Bangkok and its vicinity, Thailand. Projection shows that the amount of rainfall is going to increase significantly during both dry and wet seasons for all four Asian cities. For four Asian cities, Figure 4 displays the yearly rainfall trend and an overview of the absolute change in annual, dry- and wet-season rainfall.

3.3. Projected future land-use of Asian cities

Future land-use change in all four Asian cities was projected using the Dyna-CLUE model. Three land use scenarios (HU, MU and LU) were developed for Bangkok and its vicinity, Ho Chi Minh City, Kathmandu Valley and Lahore to analyse its impact on groundwater. For Bangkok and its vicinity, in a high urbanisation scenario, agricultural land decreases from 69% of total land area to 33% from 2015 to 2099. Whereas built-up area increases from 15% to 52% of total land area. For medium urbanisation scenario, built up area increases from 15% of total land area to 25% from 2015 to 2099 and in low urbanisation scenario, forest area increases from 7% to 25% of total land area from 2015 to 2099 whereas, built up area is constant in all the future period.

For Ho Chi Minh City, in a high urbanisation scenario, agricultural land decreases from 32% of total land area to 2% from 2015 to 2099. Whereas built-up area increases from 26% to 62% of total land area. For medium urbanisation scenario, built up area increases from 26% of total land area to 35% from 2015 to 2099 and in low urbanisation scenario, forest area increases from 29% to 50% of total land area from 2015 to 2099 whereas, built up area is constant in all the future period.

Likewise for Kathmandu Valley, in a high urbanisation scenario, agricultural land decreases from 10% of total land area to 2% from 2015 to 2099. Whereas built-up area increases from 21% to 51% of total land area. For medium urbanisation scenario, built up area increases from 21% of total land area to 25% from 2015 to 2099 and in low urbanisation scenario, forest area increases from 62% to 75% of total land area from 2015 to 2099 whereas, built up area is constant in all the future period.

In Lahore, agricultural land decreased from 95% of total land area to 75% from 2015 to 2099 under a high urbanisation scenario. In contrast, built-up area rises from 4% to 25% of total land area. Built-up area increases from 4% to 12% of total land area in the medium urbanisation scenario from 2015 to 2099, while forest area increases from 0.1% to 25% of total land area in the low urbanisation scenario from 2015 to 2099.

Figure 5 shows the land use map of four Asian cities in different future periods under a high urbanisation scenario. The land use map of four Asian cities for medium and low urbanisation scenario is represented in Supplementary Figures 1 and 2, respectively.

3.4. Groundwater recharge trends

For all four Asian cities, the groundwater recharge for the reference period 2001–2005 (Bangkok and its vicinity), 1989–2002 (Ho Chi Minh City), 1979–2005 (Kathmandu Valley) and 2000–2005 (Lahore) and for three future periods: 2010 to 2039 or near future (NF), 2040 to 2069 or mid future (MF), and 2070 to 2099 or far future (FF) were approximated seasonally and yearly (wet season from May–October and dry season from November–April). In the high and medium urbanisation scenarios, groundwater recharge is predicted to decrease, whereas it will rise in the low urbanisation scenario in both RCP scenarios. Future recharge in groundwater is anticipated to rise further during the dry season. The decline in future groundwater recharge is expected to be significant in the wet season.

For Bangkok and its vicinity, in a high urbanisation scenario, according to RCP 4.5 and RCP 8.5, the rate of fall in groundwater recharge will be between 28.3 mm/yr to 44.7 mm/yr and 22 mm/yr to 41.1 mm/yr, respectively. In a similar manner, future groundwater recharge for the medium urbanisation scenario is anticipated to decrease from 22.6 mm/yr to 33.3 mm/yr for RCP 4.5 and RCP 16.7 to 30.2 mm/yr for the RCP 8.5. In contrast, the expected increase in future groundwater recharge for low urbanisation scenarios ranges from 19.6 mm/yr to 35.8 mm/yr and 27.4 mm/yr to 40.1 mm/yr for RCP 4.5 and RCP 8.5, respectively.

In the high urbanisation scenario for Ho Chi Minh City, according to RCP 4.5 and RCP 8.5, the rate of fall in groundwater recharge will be between 22.1 mm/yr and 50.6 mm/yr and 29 mm/yr to 57.8 mm/yr, respectively. In a similar manner, future groundwater recharge for the medium urbanisation scenario is anticipated to decrease from 18.9 mm/yr to 46.1 mm/yr for RCP 4.5 and from 13.9 mm/yr to 42.1 mm/yr for RCP 8.5. In contrast, the expected increase in future groundwater recharge for low urbanisation scenarios ranges from 16 to 33.9 mm/yr for RCP 4.5 scenarios and 22.4 to 36.8 mm/yr for RCP 8.5 scenarios.

In Kathmandu Valley, for a low urbanisation scenario, the anticipated upsurge in future groundwater recharge ranges from 57.3–104.9 mm/yr and 65.7–116 mm/yr for RCP 4.5 and RCP 8.5, respectively. For the high urbanisation scenario, the drop in future groundwater recharge is anticipated to vary from 43.7–115.8 mm/yr and 28.2–104.1 mm/yr for the RCP 4.5 and RCP 8.5 scenario, respectively. For medium urbanisation scenario, groundwater recharge is expected to increase by 10.3 mm/yr and 1.4 mm/yr in near and mid future, whereas it is expected to decrease by 13.4 mm/yr in far future under RCP 4.5 scenario and for RCP 8.5 scenario, groundwater recharge is projected to increase by 15.1 mm/yr, 5.8 mm/yr and 2.4 mm/yr in near, mid, and far future, respectively.

For Lahore, in a high urbanisation scenario, the drop in future groundwater recharge is anticipated to differ between 12.4–34.3 mm/yr and 8.9–28 mm/yr for the RCP 4.5 and RCP 8.5, respectively. Future groundwater recharge for the medium urbanisation scenario is anticipated to decrease from 6.4 mm/yr to 23.2 mm/yr for RCP 4.5 and RCP 4.8 to 20 mm/yr for RCP 8.5. Whereas, for a low urbanisation scenario, the expected upsurge in future groundwater recharge ranges from 11.3 mm/yr to 27.2 mm/yr for RCP 4.5 scenario and 20.3 mm/yr to 34.4 mm/yr for RCP 8.5.

The findings show a progressive rise in groundwater recharge from the near future to the distant future in low urbanisation scenarios, with the lowest recharge found in high urbanisation scenarios. This is due to increased impervious surface induced by rapid urbanisation (Adhikari, Mohanasundaram, & Shrestha, 2020). It may also be established that changes in land-use patterns contribute much more to or less to groundwater recharge than changes in climate.

Figure 6 depicts the combined consequences of climate change and land use change under RCP 4.5 and RCP 8.5 scenarios and high urbanisation scenario on groundwater recharge across three future time periods: NF, MF and FF relative to the baseline period in four Asian cities. Medium and low urbanisation scenariosis presented in Supplementary Figures 3 and 4, respectively.

3.5. Groundwater level trends

Following a comparison with the baseline year 2001, projections were made for four future periods: 2035, 2055, 2075, and 2095 under two RCP scenarios and three land-use change scenarios (high, medium, and low urbanisation). Three abstraction scenarios (similar to land-use scenarios) were also examined in order to compute future groundwater abstraction, which was then utilised to forecast future groundwater levels in all aquifer layers.

For Bangkok and its environs, the results show that average groundwater levels are anticipated to fall under high urbanisation scenarios and all RCPs scenarios. Whereas, for medium and low urbanisation scenarios and both RCPs scenarios it is expected to increase in future. The amount of change in groundwater level through the Bangkok city is not uniform. All RCPs cause a fall in groundwater level in the city and high urbanisation scenario with the central part experiencing noteworthy reduction in groundwater level as compared to western and eastern part. Under the high urbanisation scenario, the highest fall in groundwater level was 30.5 m and 29.5 m by 2095 in the NB aquifer for RCP 4.5 and RCP 8.5 scenarios, respectively. Under all RCPs, the groundwater level in Bangkok increased for medium and low urbanisation scenario with the increment higher in western and eastern part as compared to central part and maximum increase in groundwater level by 31.06 m and 32.06 m by 2095 was seen in NB aquifer for RCP 4.5 and RCP 8.5 scenarios, respectively under low urbanisation.

In Ho Chi Minh City, the average groundwater level is expected to fall under high urbanisation scenarios and all RCPs. In contrast, it is expected to rise in the future for medium and low urbanisation scenarios, as well as both RCPs. The amount of change in groundwater level is not uniform throughout Ho Chi Minh City. Under all RCPs and high urbanisation scenarios, the city experiences a decrease in groundwater level, with the central part experiencing a significant decrease in groundwater level as compared to the northern (Chu Chi) and southern parts (Can Gio). Under the high urbanisation scenario, the upper middle Pleistocene aquifer experienced the greatest decrease in groundwater level of 56.9 m and 53.5 m by 2095, respectively. Ho Chi Minh City experienced an upsurge in groundwater level for all RCPs, medium and low urbanisation scenarios, with the increase being greater in the northern (Chu Chi) and southern (Can Gio) parts of the city than the central part, and the maximum increase in groundwater level by 21 m and 22.5 m by 2095 was seen in the middle Pleistocene aquifer for RCP 4.5 and RCP 8.5 scenarios, respectively, under low urbanisation.

The average groundwater level in Kathmandu Valley will decline in the future under all three pumping scenarios: high, medium, and low urbanisation, as well as both RCP scenarios. The average decline in groundwater level was greater in the high urbanisation scenario than in the medium and low urbanisation scenarios. The pace of decline in groundwater level is not uniform throughout the Valley. The fall in groundwater level in the Valley’s centre is considerable, and it is considerably greater than in the northern and eastern parts. The maximum decrease in groundwater level of 70 m and 66.7 m by 2095 was seen in well M4 (Lubhoo) for RCP 4.5 and RCP 8.5 scenarios, respectively under high urbanisation scenario. Whereas minimum decrease of 1.5 m and 1.3 m by 2095 was seen in well GK4 (Nayapati) for RCP 4.5 and RCP 8.5 scenarios, respectively under low urbanisation scenarios.

The average groundwater level in Lahore is expected to fall under high urbanisation scenarios and all RCPs. In contrast, it is expected to rise in the future for medium and low urbanisation scenarios, as well as both RCP scenarios. The pace of decline in groundwater level in Lahore is not uniform. The fall in groundwater level in the city’s centre is significant and much greater than in the outskirts. Under the high urbanisation scenario, the maximum decrease in groundwater level was 62.3 m and 57.2 m by 2095 in well Dhobi Ghat for RCP 4.5 and RCP 8.5 scenarios, respectively. Ghazi Mohala Children Park Ghari Shahu saw increases of 16 m and 18.9 m by 2095 for RCP 4.5 and RCP 8.5 scenarios, respectively under low urbanisation scenarios.

These regional disparities in groundwater level may be related to the amount of groundwater pumping caused by the intensity of the pumping wells. The core section of Asian cities is more densely inhabited, has the most abstraction wells, and draws the most water for home, industrial, and other commercial activity. Also, the centre areas of Asian cities are built-up, restricting groundwater recharge to aquifers.

Figure 7 shows the average change in groundwater level for both RCP scenarios (RCP 4.5, RCP 8.5), three land-use change and pumping scenarios (low, medium, and high urbanisation) for four future periods: 2035, 2055, 2075 and 2095 relative to the baseline in four Asian cities and Figures 8 and 9 shows absolute change in future groundwater level with respect to observed groundwater level in 2035, 2055, 2075 and 2095 for aquifer layers under high urbanisation scenario and RCP 4.5 (a) and 8.5 scenarios (b) in Bangkok and its vicinity and Ho Chi Minh City, Kathmandu Valley and Lahore. The results under medium and low urbanisation scenarios for Bangkok and its vicinity and Ho Chi Minh City are presented in Supplementary Figures 5 and 6, respectively.

3.6. Spatial variation in groundwater resiliency

Based on the results of groundwater recharge from hydrological models and groundwater level from groundwater models, a groundwater resilience map of Asian cities was created. In this research groundwater resiliency is specified as the percentage recuperation over total depletion. The groundwater resiliency map of all four Asian cities was developed for five time periods (2025, 2035, 2055, 2075 and 2095) under three land use and pumping scenarios (low, medium, and high urbanisation scenario) and climate change scenarios (RCP 4.5 and RCP 8.5).

Under high urbanisation scenarios, it is anticipated that less land area will fall into the “very highly resilience class” in Bangkok and the surrounding area, while more land area will fall into the “not resilient class” as time goes on. This is true for all climate change scenarios and for all three aquifer layers. Central part of Bangkok and its vicinity (Bangkok metropolis, Pathum Thani, Nonthaburi, Samut Sakhon, Ayutthaya) are not resilient to climate change and human development. Due to the dense habitation in the city’s centre, the aquifers receive less groundwater recharge. Besides this the central part of the Bangkok consists of underlying clay layer which forbid the water to enter the aquifers. Also, groundwater abstraction in these areas is very high due to high population density and industrial areas. Western part of Bangkok and its vicinity (Nakhon Pathom) is resilient to climate change and human development. Western part of the city is the major groundwater recharge zone for the Bangkok aquifer system. Under medium and low urbanisation scenarios the findings show an expected expansion in proportion of area under “very highly resilient class”.

In Ho Chi Minh City, under high urbanisation scenarios, there is projected diminution in percentage of area under “very highly resilience class” while the area under “not resilient class” is increasing as we move to the future, and this is valid for all climate change scenarios and all four aquifer layers. Central part of Ho Chi Minh City (District 1-12, Binh Chan, Hoc Mon, Binh Tan) is not resilient to climate change and human development. For the Lower Pleistocene aquifer all part of HCMC is not resilient to climate change and human development. Due to the dense habitation in the city’s centre, the aquifers receive less groundwater recharge. Also, groundwater abstraction in these areas is very high due to high population density and industrial areas. Northern part of Ho Chi Minh City (Chu Chi) and Southern part (Can Gio) is resilient to climate change and human development for three aquifer layers. Northern part of the city is the major groundwater recharge zone for HCMC aquifer system. Southern part consists of forest area (conservation area), increasing groundwater recharge to the aquifers. Under medium and low urbanisation scenarios the findings show an expected expansion in percentage of area under “very highly resilient class”.

According to the data, the percentage of the Kathmandu Valley that falls into the “very highly resilient” class is predicted to decline over time, while the percentage of the “not resilient” class is predicted to rise. For areas with high urbanisation compared to those with medium and low urbanisation, there was a considerable decline in the area categorised as “very highly resilient” and an expansion in the area under “not resilient”. Central part of the Kathmandu Valley (Lubhu, Kathmandu airport, Baneshwor, Swayambhu, Patan, Baluwatar) are not resilient to climate change and human development. Due to the dense habitation in the Valley’s centre, the aquifers receive less groundwater recharge. Also, groundwater abstraction in these areas is very high due to high population density and industrial areas. Northern part of the Valley (Budhanilkantha, Sundarijal, Shakhu, Danchi, Gokarna) are resilient to climate change and human development. Northern part of the Valley is the major groundwater recharge zone for Kathmandu aquifer system. This part is a semi urban area with more agricultural area and forest as compared to built-up hence the groundwater recharge in these areas is high.

According to the high urbanisation scenario projections for Lahore, the proportion of area falling into the “very highly resilience” class is anticipated to decline over time, while the percentage of land falling into the “not resilient” category is projected to rise. Central parts of Lahore (Allama Iqbal town, Johar town, Shahadara, Badshahi mosque, Garhi Shahu) are not resilient to climate change and human development. Due to the dense habitation in the city’s centre, the aquifers receive less groundwater recharge. Also, groundwater abstraction in these areas is very high due to high population density and industrial areas. The outskirts of the city (Lokhori, Jahman, Padhana, Wahqa, Manga mandi) are resilient to climate change and human development. Outskirts part of the city is a semi urban area with more agricultural land hence the groundwater recharge in these areas is high.

Figure 10 shows the groundwater resilience map of Bangkok and its vicinity and Ho Chi Minh City in five different time periods 2025, 2035, 2055, 2075, and 2095 under high urbanisation scenarios and RCP 4.5 (a) and RCP 8.5 (b) scenarios. For medium and low urbanisation scenarios, the results are presented in Supplementary Figures 7 and 8, respectively. Figure 11 shows the groundwater resilience map of Kathmandu Valley and Lahore in five different time periods 2025, 2035, 2055, 2075, and 2095 under high, medium and low urbanisation scenarios and RCP 4.5 (a) and RCP 8.5 (b) scenarios.