Climate change water vulnerability and adaptation mechanism in a Himalayan City, Nainital, India

Urban water management is a growing concern in India’s rapidly urbanizing cities. Population growth and climatic variability are exuberating the impact on surface and underground water supply. Understanding the causes and the extent of water vulnerability is required for developing effective strategies for water insecurities. This study attempts to assess the water vulnerability across different wards of a touristic city of Himalaya-Nainital using IPCC approach considering the three dimensions: exposure, sensitivity, and adaptive capacity. Seven indicators, mostly spatial, i.e. edaphic (aspect, elevation) and climatic (land surface temperature) besides some water infrastructural (distance to water distribution) and population, were considered for development of vulnerability index using Analytical Hierarchy Process for assigning weights. These indicators were simple to extract and easy to obtain and mostly available from secondary sources and were capable to account the variability at micro-level. Moreover, the current adaptation mechanisms for water security were also derived through conducting surveys by randomly selecting households across the wards. Staff House and Harinagar wards were the most vulnerable. The survey results that the adaptation mechanism should be managed at individual and organization level. Policy measures such as optimum use of water, grey water recycling, spring rejuvenation, rain water harvesting, and leakage proof infrastructure with intervention of new technologies, may be adopted and implemented for reducing the water vulnerability in the city along with the public participation. The appropriate measures for water vulnerability would further provide support for improving the facilities to the tourists in the city thereby improved economic opportunities to the locals.


Introduction
About two-thirds of the world's population live under serious water shortage for at least 1 month of the year, and half a billion face severe water scarcity every year (Mekonnen and Hoekstra, 2016). Escalating water insecurity under rapid urbanization, climate change under current water scarcity with water governance is a global concern (Flörke et al., 2018). Moreover, the current population growth in expanding cities will increase the water demand, resulting in over-exploitation of surface water and ground water resources (Hoekstra et al., 2018) and therefore ensuring the cities have an ample source of water becomes exceedingly necessary (Padowski and Jawitz, 2012). This problem is critical as by the end of 2050, around 66% of the world population will be residing in urban areas with high proportion to the Asia and Africa (UNDESA, 2014). With a population of 1.3 billion and an anticipated growth of 1.7 billion by 2050, providing clean and safe water to vast majority of populace in India is difficult (Kumar, 2019). The problem is severe as climate change has already impacted the water accessibility and is most likely to exacerbate water availability and accessibility in the future (IPCC 2014). Water crisis is already recognized as one of the global high-risk issues (WEF, 2017) and the urban areas of the remote and fragile Himalayan mountains are facing acute problem for water supply through municipalities . The increased water demand due to increasing Responsible Editor: Philippe Garrigues * Rajiv Pandey rajivfri@yahoo.com ecotourism (Cook and Bakker, 2012) coupled with climate change  would further broadened the water crisis in cities of the Himalayan region.
In the Himalayan region of India, urban growth is instrumental towards the improvement of infrastructure in the remote areas (Tiwari and Joshi, 2016); however, the haphazard urban growth in the fragile mountains is instrumental for disturbing the ecosystem thereby disrupting the hydrological regimes causing depletion of natural resources, drying of springs, and reduction in ground water recharge (Tiwari et al., 2018). Moreover, majority of the population in Himalaya for water depends on groundwater either from nearby spring or through digging wells and bore wells and majority of urban areas are lacking the infrastructure for satisfying water demand of the population from municipal sources . Change in the water demand and supply due to urban sprawl; land-use changes due to population expansion and impact of climate change are projected to further threaten the water sustainability (Schnoor, 2015).
Excessive reliance on ground water in the Himalayan region will have deleterious consequence keeping in view the inherent fragility of aquifers in mountainous regions and thus, mountain-specific urban planning and accountability of the stakeholders is required to address the grim future of water resources . Because of vast socioeconomic, geographic, and meteorological differences existing among cities, a generalized national level water policy may not be appropriate to streamline the water infrastructure for distribution and supply of water to the urban population. Therefore, a systematic assessment of criticality of water under the current pace of urbanization and climate change at the city level is required with consideration of stakeholders' perception at spatiotemporal scale matching with municipal water infrastructure schemes (Dong et al., 2020). Vulnerability analysis tends to be a reasonable approach for assessing the crisis as the information provided may be useful for the policy makers in decision-making for addressing the water crisis (Pandey et al. 2014).
Vulnerability is defined as the propensity or predisposition to be adversely affected and encompasses a variety of concepts and elements including sensitivity or susceptibility to harm and lack of capacity to cope and adapt (IPCC, 2014). Though various vulnerability assessment frameworks exist for assessing climate change and water vulnerability, there is no specific methodology with adequate indicators for assessment of water vulnerability at micro-level for the hilly region. Those prevailing methodologies applied at macro-level will not be suitable for micro-level household analysis since factors and characteristics differ at the micro-level framework (Pandey et al., 2014). Vulnerability may be accessed through a composite index, which is a unit less relative construct for measuring the susceptibility of the system, using various relevant system specific indicators of the three dimensions of vulnerability (Abson et al., 2012;Giri et al., 2021). However, conceptualizing vulnerability and choosing appropriate indicators are key challenges for vulnerability assessment (Füssel, 2007). Therefore, site-specific micro-level operative model framework for water crisis management is required to formulate in order to create awareness among the policy makers and authorities to plan their responses in implementation of water management policies (Nguyen et al., 2020). Moreover, results of such evaluation at the city level would add value to local level water planning, which is essential to strengthen the local governance ability for facilitating the informed decision-making by the policy makers (Ojha et al., 2020). Present evaluation of water vulnerability differs with exiting literature due to its specificities for site-specific micro-level indicators including the various interacting factors for water management such as geology and forests besides the infrastructure along with stakeholder-based perception evaluations for understanding the adaptation mechanisms.
Hill town and cities are degrading in variety of aspects such as micro-climate, vegetation, recharging of aquifers, and water table, leading to affect the environment of the town and city and also the surrounding regions (Kumar, 2013). Therefore, an evaluation for water vulnerability under current stresses was attempted in Nainital city which surrounds the Nainital Lake in the lower Himalayas, to supplement information about the water management. Nainital Lake is the major source of water supply for drinking and other domestic purposes in the town of 40,000 local inhabitants and tourists including recreational activities (Purushothaman et al., 2012). Moreover, various anthropogenic activities such as illegal construction, litter, domestic discharge, and recreational usage of lake are detrimental for the lake's quality and quantity (Purushothaman et al., 2012). The current scenario of water issues in the city and differential distribution of water in the different wards was evident from the interview and the field survey conducted during the year 2020-2021. The study hypothesized that the municipal wards have differential vulnerability and are governed by physical and infrastructural resources besides population density. The study objectives were to estimate ward-level vulnerability and identify adaptation mechanism at stakeholder levels. The study was primarily based on remote sensing data for majority of the key indicators and the results are also represented using GIS maps for better visualization and understanding besides primary information were also collected from the local population to understand the coping mechanism. The key indicators for evaluation were less in number and easily assessable and primarily based on secondary sources and therefore facilitating to the dynamic nature of vulnerability. The water vulnerability can be reduced by applying the stakeholders perspectives for water management at the city and therefore stakeholders' perspective were also evaluated through accounting their views specifically in terms of water management through rain water harvesting,

Theoretical framework
Present study considered "vulnerability as the susceptibility to be harmed" and is measured using IPCC framework based on the three dimensions of vulnerability exposure, sensitivity, and adaptive capacity (Pandey et al., 2014;IPCC 2007;Omerkhil et al., 2020;Giri et al., 2021). Many theoretical frameworks for assessing vulnerability depending upon the context were developed with varying level of scope in terms of scale, dimensions, and components (Jamshed et al., 2020a, b, c). For example, MOVE framework, a general as well as integrative and holistic framework to systematize and assess vulnerability, risk and adaptation, which underlines the key factors relationships to the exposure of a society or system to a stressor, the susceptibility of the system or community exposed, and its resilience and adaptive capacity (Birkmann et al., 2013). However, vulnerability in the present context was derived from a mixture of physical units by considering the physical characteristics of the system, i.e. urban region by considering edaphic factors along with using political ecology connect by considering population, land use, and land cover besides governance in terms of water supply distribution. The water vulnerability of the city can be broadly socioeconomic and biophysical in nature. From a socioeconomic perspective, looking specifically at a city's population, the vulnerability encompasses with various factors such as awareness of climate change, relative mobility, socioeconomic status, and length of residence time (Lindley et al., 2011), and in a mountainous city with low population, the awareness, constrained mobility along with residence time, and population density were considered the prime factors for the study. The physical fabric of a city encompasses through edaphic factors including the quality and location of physical infrastructure which also temperature and LST (Mutiibwa et al., 2015). Rise in urban temperature due to variety of reasons such as urbanization, population, climate change poses urban heat, and water availability, as a use to cool and various other purposes for human system, is important for reliable water supply as a measure to adapt the increasing urban heat. Land surface temperature (LST) as the skin temperature of the land is an important parameter for determining hydrology between the ground and the atmosphere in urban system (Adanu et al., 2020). Sensitivity is defined as the factor that cause a system or population to be harmed or deprived of water, and thus, population density, a measure of water requirement (Lee and Choi, 2018), and edaphic factors such as slope and elevation (Gupta et al., 2020;Sekhri et al., 2020) were considered indicators for sensitivity besides land use and land cover (LULC) as proxy to modify water availability (Adanu et al., 2020); i.e., less urbanization at gentle slope at low altitude lead to improvise the condition for improved water availability by reducing the evapotranspiration. LULC change has direct impacts on water yield (Li et al., 2018) by modifying the underlying mechanisms of transferring rainfall to the surface water balance and the partitioning of precipitation into evapotranspiration, runoff, and groundwater flow (Foley et al., 2005). Slope has indirect impacts on water availability (Tse-ring et al., 2010). Steep slope can increase the flow rate of a waterbody which could lead to increase in soil erosion; moreover, slope has a significant effect on the infiltration rate of groundwater; the amount of infiltration decreases with increase in the slope (Fox et al., 1997).
The capacity of water availability and facility for household use was considered adaptive capacity in the present context, and therefore, keeping in view of mountainous city, distance to water body, i.e. lake (Das et al., 2020), and distance to road (Pandey and Jha, 2012) were considered indicators for adaptive capacity (Table 1). It was presumed that the shorter the distance between infrastructure and the water supply and distribution system, the lower the infrastructure requirements besides reducing the chances of leakage. Moreover, the road infrastructure also reflects better maintenance of water distribution systems under the vulnerable state such as leakage and pipe bursting during winter.
Overall, seven indicators for defining different dimensions of vulnerability were included where exposure, sensitivity, and adaptive capacity included one, three, and four indicators, respectively, keeping in view of policy perspective based on less number of parameters (Table 1). As the study area was small, climatic factor could not be obtained through remote sensing for exposure, and hence, land surface temperature (LST) was considered a proxy of temperature. Sensitivity included four indicators such as land use land cover, slope, population density, and elevation which had positive effect on the water vulnerability of the city. The adaptive capacity included two indicators, namely distance to lake and distance proper storage and sustainable use of water at individual and household level, and role of water supply institution. The results will support the decision makers to re-orient policies with inclusion of more pronounced role of public and therefore implying appropriate measures in various wards to reduce the water vulnerability. The proposed theoretical framework in this study can be used in future studies focussing on micro-level water vulnerability assessment through customized modification.
an indicator of governance mechanism (Carter et al., 2015). The humans, possessions, livelihoods, and habitats are exposed to climate through creating potentially harmful environment for water distribution and infrastructure, and keeping in view of local level of variability, land surface temperature (LST) as a proxy of element of climate, i.e. temperature, was considered. Despite the fact that the two temperatures differ in their physical meanings and responses to atmospheric circumstances, there is a strong link between near surface air to road, since these two indicators can aid people to cope with the water stress in the wards. The relationship of each indicator with vulnerability is illustrated in Table 1, where positive sign shows the direct relationship of indicator with vulnerability whereas negative sign indicates the inverse relation of indicator with vulnerability. Adaptation mechanism for biophysical and governance units may be visualized directly through the perception of locals by collecting the information through evaluating their perception. The overall experience of the locals in relation to water management with their various dimensions such as water crisis, infrastructure, institution, and locals' role for water use optimization may be used to reduce effectively the water vulnerability.

Materials and methods
Profile of study area topographic depressions, and debris cones. The slope mainly varies between 20 and 40°, occasionally over 50°at some places and covered with slope of wash materials, with the lacustrine deposits at topographic depressions (Sah et al., 2018). The town has a tropical climate with a mean temperature of ∼25°C in summer and near to 0°in winter with snowfall (Gupta et al., 2016). Pine, Oak, Walnut, Chinar, Hisalu, Kaphal, Buruns, and other temperate forests species are main vegetation in the town (Fig 1).
Nainital is a popular tourist destination in Uttarakhand with a population of 41,377 as per the Census of India 2011 and distributed in fifteen wards. Although tourism boosts the economy of the locals but excessive tourism has become a critical issue in the city affecting the quality of the lake water besides reduced water supply to the residents due to provisions for hotels (Barbhuiya, 2020). The Nainital Lake forms the city's primary source of water, as all the infiltration wells and shallow tube wells are directly recharged by lake water and have been utilized for varied purposes starting from tourism, irrigation, potable water, and to fisheries at times. The dependence of the individuals of the town on the lake for their livelihood and for water supply has increased during peak tourist season thereby exhibiting high stress on the water bodies.

Dataset collection
The different datasets utilized in the study for vulnerability analysis is from secondary sources and specified in Table 2. The indicators as per theoretical framing was ward wise population, land surface temperature (LST), land use land cover (LULC), slope, elevation, distance to road, and distance to lake. Except ward wise population, all other variables have  (Koech et al., 2020) Slope Sensitivity due to steeper slopes Positive (Lee and Choi, 2018;Gupta et al., 2020;Sekhri et al., 2020) Elevation Sensitivity due to topography. Higher elevation requires better infrastructure and high pressure for water distribution Positive (Gupta et al., 2020;Sekhri et al., 2020) Adaptive Capacity Distance to lake Distance to lake is proxy for infrastructural availability of water and more the distance, high the vulnerability Negative (Das et al., 2020) Distance to road Distance to lake is proxy for availability of water and the more the distance, the higher the vulnerability The study area Nainital town is situated in Kumaon Lesser Himalaya in Uttarakhand state of India. The town with varied topography spreads in a total area of 11.73 km 2 and lies at 29°2 2′ 49′′ N Latitude and 79°27′ 48′′ E Longitude with the average elevation at 2084 m above mean sea level. This area is bounded by various hills such as Naina Peak, Sher-ka-Danda, Deopata, Lariyakata, Alma, and Ayarpatha. The kidney-shaped Nainital Lake is the major source of water supply to the town, and other sources encompass of various springs such as Pardadhara and Sipahidhara. The chief drainage of the area is Balia Nala which is fed by the Nainital Lake (Khanduri, 2019). The major geomorphic features of the region are characterized by mountain peaks, hill slopes, been retrieved from remote sensing, whereas ward wise population data was obtained from Nagar Palika Parishad, Nainital. The tools Google Earth Engine and ArcGIS 10.1 software were used in this study for data collection and analysis.

Data processing in GIS
The exposure component was determined by LST, due to the coarser resolution of the climatic data causing insignificant variation in the climatic data in different closely connected wards and the absence of the metrological station in the nearby area. LST was considered and was obtained from Landsat data (with 30m resolution) derived through Google Earth Engine. The thermal bands of Landsat TM, ETM+, and Landsat 8 are commonly used for temperature analysis (Cristobal et al., 2018). Code was run in Google Earth Engine to obtain annual mean of LST for the period 1990 to 2020. Layers having no data value or gap value were filled with the help of focal statistics tool and by using raster calculator; mean was obtained in ArcGIS 10.1 software.
The sensitivity dimension consisted of land use land cover classification (LULC), which was obtained using LISS IV data with 5.8m resolution. Google Earth Engine was used for the supervised classification as well as for reprojection and clipping. SVM (support vector machine) classifier, a supervised non-parametric method present in Google Earth Engine, was used to extract the land cover features. The merit of the approach was its ability to generalize well in cases where training dataset is limited and achieve higher accuracy than other conventional methods, such as (ML) maximum likelihood (Mantero et al., 2005;Mountrakis et al., 2011). A total accuracy of 0.83 was obtained with the assessment being determined by comparing the classified image with higher resolution data using a confusion matrix.
Other indicators of sensitivity included slope and elevation that were derived through SRTM DEM with 30m resolution, whereas for population density, the dataset of population was obtained from the Nagar Palika in Excel format. The data was Fig. 1 Location map of the study area comprising of different wards of Nainital city converted to a shapefile by joining the Excel sheet with the ward level boundary of Nainital. The shapefile was then projected and resampled to 5 m resolution. Population density (PD) was then obtained by dividing the population of each ward with the respective ward area in the attribute table of the ward boundary and converted to a raster, using polygon to raster tool in ArcGIS 10.1 software ( Table 3).
The two indicators for adaptive capacity dimension comprised of distance to road (DR) and distance to lake (DL). The lake was digitized and the road network shapefile was downloaded from Open Street Map (OSM) and clipped to the ward boundary of Nainital using the ArcGIS 10.1 software. Euclidean distance tool was used for estimation of distance to lake and distance to road from the ward boundary. It was made sure that all the raster layers were resampled to 5 m, and based on field observation and expert knowledge, weights were assigned to each indicator in this study using the methods of AHP and pairwise comparison matrix.

Survey and data collection
Based on the theoretical framework for adaptation, a questionnaire was designed for household interview and a survey was conducted during January to March 2020-2021. Household interview was conducted keeping in view of obtaining a comprehensive view on various water issues at household level following the guidelines of the government under the COVID crisis. The questionnaire contained four sections for accounting various issues of water including perception for climatic trends and adaptation protocol for addressing water vulnerability within the limited resource mix. The first section includes general information about the respondents' group. Second section of the questionaire was devoted for the challenges of water with focus on quality for uses; groundwater recharge and water scarcity and crisis. The third section mainly included questions on changes in climatic parameters and impacts of climate change in the last decade. The fourth section primarily dealt with the adaptation mechanism by managing the stakeholder role, i.e. public and water institution. Before administering the questionnaire, pre-testing was done and redundant questions were deleted. We adopted a random sampling approach to select nine to ten households from each fifteen wards of the city. After briefing on the study and receiving consent, the interview was conducted with the head of households. Based on respondents' preference, the interview was conducted in the local language for at most 30 min with 148 randomly selected groups. The missing value was around 6% and replaced with the mean responses. The Cronbach's alpha for the data was 0.66, which ensures the reliability of the data.

Vulnerability assessment
The overall methodology to assess water vulnerability is based on several studies (Jamshad et al., 2020c;Pandey et al., 2014) and categorized into four broader steps: (1) identification of appropriate indicators associated with all the three dimensions of vulnerability, and their normalization; (2) assigning weight to each indicator after rational and scientific evaluation; (3) integration of all the indicators at its dimension level; and (4) estimation of water vulnerability index for spatial representation of vulnerability in each ward. The methodological steps adopted for the present study is reported in Fig 2. The indicators were selected as per the theoretical settings. Assigning weight to the indictors is one of the crucial points in vulnerability assessment (Papathoma et al., 2019), as each indicator contributes differentially to vulnerability and has its own positive and negative impact on water vulnerability (Pandey et al., 2014). Several methods are available in literature to allocate weights to the indicators; however, Analytical Hierarchy Process (AHP) is a widely used method for assigning weight in water requantitative data (Saaty, 1980). AHP basically involves four steps: in the first step, a pairwise comparison matrix is constructed for each criterion; in the second step, the resulting matrix is transformed into a normalized matrix; third step involves the calculation of the average of the values in each row to get the corresponding weight; and the last step includes the determination and verification of the consistency ratio. AHP value was assigned by sixteen experts from various domains as urban developers, water managers, foresters, economist, and policy makers.
The relative significance of two indicators is estimated on a scale of 1 to 9, where 1 indicates that the two indicators have equal importance and 9 specifies that one indicator is significantly much more important than the other (Saaty, 2008). The relative significance of factor was adjudged based on the experts of the theme including the authors. All the weights were compiled into a pairwise comparison matrix with a diagonal value 1 and reciprocal weights on the lower left triangle. Each comparison matrix yielded an Eigenvector, which was used to assign weight to each of the factors (Filho et al., 2020). For example, in Table 3, the first row has LULC value equal to LULC; thus, it was marked as 1; and then LULC compared to slope, PD, DL, elevation, DR, and LST is marked on a scale from 2 to 9 as given in Table 1. The consistency ratio (CR) was used to assess the precision of the measured weights and the variations between pairwise measurements. The accepted consistency ratio is less than 0.1; otherwise, it is necessary to re-check subjective arbitrary assumptions and recalculate the weights (Saaty and Vargas, 2001). The consistency ratio (CR) is calculated by using the following equation: where CI is the consistency index and determined by the equation where n represents the size of the matrix and λ max is the average of the values. RI represents the random matrix to check the consistency of pairwise comparison matrix. The value of RI was computed by (Saaty, 1980).
The average values for comparison matrix were estimated in two steps, first by multiplying each column of the pairwise comparison matrix by the corresponding weight and in second step, sum of the rows was divided by its corresponding weight for estimating the final average values denoted by (λ max ) ( Table 4).
CI=0.03, RI=1.24, CR=0.02; (LULC land use land cover; PD population density; DL distance to lack; DR distance to road; LST land surface temperature) Based on the vulnerability relationship, the quantified indicators were standardized and indexed for vulnerability analysis (Pandey et al., 2018). The indicator which is positively influencing vulnerability, i.e. increase vulnerability, was transformed as: (Zhou et al., 2018). AHP is a multi-criteria decision-making method and used to provide a systemic methodology for assessing and combining the effects of various variables, including qualitative and The indicator which is negatively influencing vulnerability, i.e. decreases vulnerability, was transformed as: Where X i is the value of indicator X for ith household. Max represents the maximum value of all the values of X indicator. Min represents the minimum value of all the values of X indicator. The weight for each normalized indicator was estimated and reported in Table 5. After AHP analysis, each indicator layer was reclassified into three classes; value of 1 was assigned to the sub-criteria contributing minimum to vulnerability and a value of 3 to the sub-criteria contributing maximum to vulnerability (Table 5). In the case of indicators of adaptive capacity, the distance to road and distance to lake were classified into three classes and value 3 was assigned to the distance nearest to lake, as it would contribute to a higher adaptive capacity due to low requirement of infrastructure and better support for maintenance leading to lower vulnerability. Precisely, the values for adaptive capacity indicators were inversed.
Further, the indicator layers were multiplied with its respective weights and aggregated to derive component and overall vulnerability. IPCC framework was applied for the vulnerability analysis with consideration of the three dimensions. The bottom-up approach provides improved stakeholder perspective for adaptation mechanism (Pandey et al., 2018). This approach also provides an advantage of visualizing separate maps for each dimension of vulnerability and therefore facilitates better understanding about the leading factors responsible for the vulnerability (Gupta et al., 2020;Sekhri et al., 2020). The formula for estimating water vulnerability was as follow (Pandey et al., 2017) with the lower the value, the lower is the vulnerability.

Vulnerability
¼ f Exposure; Sensitivity; Adaptive capacity ð Þ ¼ Exposure þ jSensitivity−Adaptive capacityj The above calculation was performed using raster calculator in ArcGIS 10.1 software. The aggregated components were divided into grids with the help of fishnet tool in ArcGIS for estimating the differential vulnerability across the ward.

Result and discussion
Apart from the secondary sources, preliminary field survey was conducted through structured questionnaire to retrieve the general information regarding the water status and challenges in the city. The result of the data analysis revealed that the households of the city receive intermittent water supply from the municipality, and all households were equipped with metered connection. The water charge varies from minimum 200 to maximum 3000 depending upon the requirements of the households. More than 86% households reported that water supply was not sufficient for the uses of households. Most of the households store water in tanks with varying capacity from 500 to 3000 L in either one or up to four tanks. During crisis, almost 37% households collect water from nearby springs or water tank. In the wards, public water source was also available at a place which is at most 100 m vicinity to 50% households, survey revealed. Twenty three percent household reported that water quality was good; however, 76.4% reported normal water quality. Water taste was more or less normal; however, the water quality was ensured through filtering by 90% household and remaining 10% boil water for drinking purposes. Interestingly, all the respondents agreed that water quality of lake has been improved with the efforts of the responsible organizations and civil body; however, ground water recharge was perceived to reduce by 97.3% households. The water quality was attributed to the leakage in the pipelines, lack of effective regulation for sustainable water use and discharge, and lack of modern technology. Majority of households were in favour of modern technology for water supply and filtration, i.e. water distribution infrastructure and law and regulation for sustainable water use and discharge (Table 6). Almost all households reported that they face water scarcity during summer and they were also not practicing water harvesting in their houses. However, the households identified over-exploitation of ground water resources through either water withdrawal or concrete house construction as major problems for reduction in ground water recharge, along with urbanization and climatic parameters such as rainfall and impact of climate change particularly changes in temperature (Table 6). Water crisis and scarcity were also investigated during the survey to understand the possible causes. The households reported that tourism and urbanization were the major causes, in spite of the fact that tourism is a major economic activity in the city, reported the households. Climate change and unsustainable use of water resources were also one of the causes; however, the intensity of these causes was not as severe as tourism and urbanization, reported the respondents. Households were having a normal perception about the working and management of the water institution (Table 6).

Exposure component analysis
Major water challenges faced by the households were supply for limited time (38.5%), leakage (25%), long distance from water pump (16.2%), besides water cost (15.5%), and insufficient storage (4.7%), as reported by the respondents. Changes in the climatic conditions were also having an adverse impact on water issues, reported the households; moreover, respondents perceived changes in climatic parameters (Table 7). All households reported that the amount of snowfall and rainfall has been decreased during the last 10 years with changes in frequency of rainfall (94.6%). However, the frequency of hailstorm and frequency of drought could not be adjudged conclusively and about 50% households were reported inconclusive for both the parameters (Table 7). Conclusively, climate was attributed for changes in water availability.
In the hilly areas due to the typical scarcity of meteorological stations, LST may be the most useful variable (Voogt and Oke, 2003) in characterizing spatial-temporal patterns of near surface air temperature (Mutiibwa et al., 2015). Therefore, LST was considered a proxy to temperature and a key indicator for exposure dimension. Higher temperature was given a value of 3, while lower temperature was assigned a value of 1. The wards with high LST values were Awagarh, Mallital bazar, Sree Krishnapur, and Harinagar. The high value of  LST was found in these wards as they covered high urban area and barren land, since both have a direct relationship with land surface temperature (Tan et al., 2020). Moreover, area with open vegetation also showed high value of LST, as urban vegetation and LST are negatively correlated (Weng et al., 2004;Chen et al., 2006;Buyantuyev and Wu 2010). Further, replacement of rural land cover type with concrete urban elements has substantial environmental consequences, such as reduce evapotranspiration, increased surface runoff, and lower air and water quality (Yue et al., 2007). Therefore, higher value of LST may have a positive influence on vulnerability and may affect people since it indicates the inadequate surface water availability (Traore et al., 2021). Thus, wards with higher LST in low vegetation and high urban area, due to the presence of impervious surface, result in increased surface runoff and decline in ground water recharge (Srinivasan et al., 2013) (Fig 3).

Sensitivity component analysis
The sensitivity dimension was estimated based on four indicators (LULC, slope, elevation, and population density) revolving around biophysical and demographic factors, each of which is addressed under separate heading in the following paragraphs.
Land use land cover LULC has been divided into three classes, namely forest, urban, and open vegetation, where urban was given value of 3 since it has a higher impact on water resources and forest was given the lowest rating 1. In the study area, forest covers around 31km 2 whereas areas under urban and open vegetation were 1.78 km 2 and 0.99 km 2 , respectively. Maximum settlement (urban area) was observed near the lake (Fig 4a). The land use land cover may have a great impact on water resources for domestic and non-domestic purposes, as urban expansion will increase the demand of water supply for infrastructure development and for other domestic use, which would eventually put pressure on lake. Moreover, LULC affects water yield through altering infiltration, evapotranspiration, and groundwater recharge capacity that is mainly determined by the biophysical characteristics of LULC composition (DeFries and Eshleman, 2004;Li et al., 2018). Apart from this, LULC has an effect on the surface water quality due to various point and non-point source pollution under different land uses (Mallin et al., 2009). Moreover, vegetated land has relatively lower water yield due to higher evapotranspiration and water infiltration than built-up areas and bare land due to their impermeable surface (Li et al., 2018). Therefore, the area has high possibility of lower water yield due to high vegetation cover.
Slope The slope varies from 0 to 65 o , and based on literature, slope was classified in three classes ranging from 3 (steep slope) to 1 (gentle slope) with the view that steep slope can increase the surface runoff and reduce the groundwater infiltration (Fox et al., 1997). Slope with 0-15°were considered low, 15-25°as moderate, and 25-65 as high slope. Moreover, the steep slopes lead to rapid changes in climatic zones over small distances, posing a significant impact on water availability (Tse-ring et al., 2010). Our result shows area with steeper slope in Sree Krishnapur, Ayarpata, Upper Mall, and Sher ka Danda wards (Fig 4b).
Elevation Topography such as elevation also plays a critical role in sensitive analysis as people residing in higher elevation are more distant from the primary source of water that is Nainital Lake and also during the water crisis; alternative sources of water like water tank were generally not available in wards with higher elevation. Another impact of the elevation was requirement of high pressure for water supply. This was evident from the field questionnaires where most of the respondents agreed that during water shortage, there was no alternate supply of the water tank due to the topography of the area. The elevation was classified into three categories and higher elevation was given a value of 3. The study shows higher elevation for wards Sher ka Danda, Snow View, and in the upper region of Sainik School, Rajbhawan, and Aryarpata wards (Fig 4 c).
Population density The analysis of population data shows that maximum population density was obtained for Harinagar Ward and the least population density was obtained for Sree Krishnapur Ward. Increase in population has a direct impact Fig. 3 Indicator of exposure-land surface temperature on the water resources, as high population requires more water for domestic purposes and would lead to put more pressure on the water resources (Fig 4d).

Adaptive capacity analysis
The adaptive capacity dimension comprised of two indicators, distance to lake and distance to road. In the city, lake is the primary source of water; therefore, people residing near to the lake have more access to water and are better equipped with water sources, thus have high adaptive capacity as compared to people residing far away from the lake. The layer was categorized into three with the distance closest to the lake receiving a high value of 3, indicating that it has a high adaptive capacity (Fig 5a). Distance to major road is considered proxy for infrastructure (Byrne, 2014;Egyir et al., 2015), as people residing near the road had more easy access to water as compared to people living far from the road (Fig 5b). Therefore, in the study, a value of 3 and 1 was assigned to the wards near and far from the road, respectively. The result shows that wards Mallital, Rajbhawan, Aryarpata, Upper Mall, Tallital Bazar, and Sher ka Danda were nearer to lake and major roads and possess high adaptive capacity.

Estimation of water vulnerability index
Water vulnerability index was developed by aggregating all the components of exposure, sensitivity, and adaptive capacity (Fig 6). The map for the three dimensions was developed as per the methodology, by aggregating the respective indicators and reported in Fig 6. The respective values for each dimension for each of the wards have been stated in Table 9 in the Appendix. The result shows that Sree Krishnapur, Harinagar, Mallital Bazar, and Awagarh were the most exposed wards though the level of exposure was very low due to the low value of LST. The sensitivity was high for Staff House, Harinagar, Awagarh, and Mallital Bazar and was least for Narayan Nagar and Rajbhawan (Appendix Table 9). The adaptive capacity was maximum in Mallital Bazar, Tallital Bazar, Uppar Mall Road, and Awagarh; however, the poorest adaptive capacity was for Narayan Nagar (Appendix Table 9). Vulnerability was estimated and reported in Table 9 in the Appendix with Staff House and Harinagar being the most vulnerable wards, due to their high population density and high built-up area in these wards. The high population density increases water consumption thus putting pressure on the infrastructure. Moreover, the LST was also high leading to more exposure of urban heat, besides less infiltration and excessive surface runoff due to the existence of impervious surface. The least vulnerable was the Rajbhawan and Ayarpata wards. The In order to enhance the resilience of a community, respondents were suggested adaptation mechanism based on people participation as also suggested by Jana et al., 2021. The adaptation mechanism revolved around the improvement in provision of water through rain water harvesting, proper storage, and sustainable use of water at individual and household level. Moreover, at the water supply institution, the respondents were very critical for reduction of leakages, i.e. minimum loss of water while distribution through effective mechanism including proper upkeep and use of advance and modern technology. The city managers may also consider the recycling and reuse of water for various purposes keeping the barriers in view as Singapore has developed the water recycle system (van Rensburg, 2016). Respondents were also in favour of recycling of water besides educating the people for better and sustainable use of the water. Most of the households reported proper storage vulnerability map was divided into three categories: low, moderate, and high. Low category was assigned to wards having value between 0 and 0.17, moderate between 0.18 and 0.25, and higher vulnerability to wards falling between values 0.26 and 0.4. The most vulnerable wards were Mallital, Awagarh, Nainital club, Staff House, Tallital, and Haringar as their overall sensitivity and exposure were high. Wards that fall under moderate category were Sher ka Danda, Upper Mall, Sree Krishnapur, Sukhatal, and Sainik School ( Fig. 7; Appendix Table 9). The differential vulnerability among the wards were due to variety of factors such as inefficient water institution (Hoekstra et al., 2018); governance failure (Pahl-Wostl, 2017); lack of water recycling (van Rensburg, 2016), and lack of comprehensive and integrated water management (Hoekstra et al., 2018). and sustainable use of water as the best possible practice for water management one can do at the initial level. At institutional level, among the various suggestions, grey water recycling mechanism, revival of spring, and awareness among the stakeholders were given priorities, as reported by the households (Table 8). The governance failures at multiple levels of governance (Pahl-Wostl, 2017) require institutional and organizational reforms for water management which may also support to reduce the water vulnerability (Hoekstra et al., 2018). Moreover, the urban development strategies and housing policies, which influence the human exposure and vulnerability, must also be aligned for addressing the vulnerability (Birkmann et al., 2021). Moreover, size of the city is also a determinant for the vulnerability to the locals (Jamshed et al., 2020c), and therefore, appropriate and suitable strategies to counter the city size would further support to better equip to address the vulnerability. The integrated water approaches with all issues pertaining to water may be considered in comprehensive manner with their mutual interdependencies, multiplicity, and dynamisms. Moreover, the approach must also be considered the wise use of water with alignment to urban dynamics and urban design (Hoekstra et al., 2018).

Conclusion
In the given study area, Nainital Lake is the primary source of water for people residing in the town. Increasing human pressure, unsustainable land use practices, declining lake depth, and drying of springs are some of the existing challenges within the city. Consequently, this study focused on understanding the water vulnerability of Nainital city at ward level using GIS modelling approach.
To assess the water vulnerability of the different wards in the area, it was essential to select appropriate indicators that were substitutions of the component's exposure, sensitivity, and adaptive capacity. The proposed method included seven main indicators (land use land over, slope, population density, elevation, land surface temperature, distance to road, and distance to lake) to evaluate the vulnerability of 15 wards within Nainital town. Weights were provided to the indicators using AHP analysis and maps were produced that showed vulnerability categories within the wards using GIS spatial analysis tools. The analysis results into overall moderate vulnerability in most of the wards of the town; however, high value of vulnerability was observed in Mallital, Awagarh, Nainital club, Staff House, Tallital, and Harinagar wards, primarily due to high sensitivity in these wards. Hence, the focus of this study was to determine the water vulnerability at ward level, with its aim to guide the policy makers and the government to reorient the policies in terms of improved distributional and infrastructural support to reduce water wastage, awareness campaign for optimum use of water, and provisions for water harvesting and develop appropriate adaptation strategies according to the vulnerability level in each ward. The reduced water vulnerability would provide better services to the tourists and therefore improve the economic opportunities to the locals. The study has limitation as water system is a complex mixture with various interacting systems and sub-systems, which is generally not clear. The index has issues for availability of precise data for various parameters, or use of proxy data, perception-based survey besides the lack of micro-level-integrated details of past information such as details of water use for various purposes. Moreover, sometime expert opinion for weightage may lead to some wrong interpretation.