Effective management of the watershed in response to historical climate change using a GIS-based multi-criteria decision analysis (MCDA)

Watershed management involves the coordinated management of land and water to mitigate the problems of droughts, floods, soil erosion, water pollution, etc. This management includes the selection and implementation of the most suitable watershed management alternatives to meet the human requirements for food, fiber, water, shelter, and recreation by conserving the water resources, plants, and soil of the catchment (Tomer 2004; Tomer et al. 2014). Hydrological processes of the watershed are influenced by dynamic factors of climate change and land use cover change (LUCC). Climate change deals with a long-term changing trend (either increasing or decreasing) of climatic parameters like temperature, precipitation, etc. (IPCC 2014). Climate change raises the temperature (causes global warming), impacts the precipitation pattern, and enhances the extreme events of droughts and floods which adversely affect water resources development (Bhanger & Memon 2008). That is why climate change is perceived as changes in the runoff yield of the watershed and causes the problem of water shortage due to the decreasing trend of runoff (Melkonyan 2015). While LUCC as the result of human activities disturbs the hydrological cycle by influencing transpiration, evaporation, and interception (Tomer & Schilling 2009) and affects the water yields of the watershed. Hence, in order to address the adverse impacts of climate change and LUCC, there is a need to formulate strategies that may ensure a balance between social, environmental, and economic factors by actively engaging watershed stakeholders.

A hydrological model, the Soil and Water Assessment Tool (SWAT), has been widely recommended to evaluate the impacts of climate change and LUCC on hydrological processes, at the basin scale (Zhang et al. 2008, 2015; Daniel & Abate 2022; Daniel 2023), because it investigates the relationship between hydrological process and climate (Jothityangkoon et al. 2001). Impacts of climate change in the SRB by employing the SWAT model have been assessed and outcomes revealed a reduction in runoff (Shahid et al. 2018). Similarly, the hydrological process and land use changes have been investigated with the application of the SWAT model in the watershed of Simly Dam, located in the SRB, and findings revealed a decrease in surface runoff (Abbas et al. 2015; Ghoraba 2015). This situation entails the active participation of the stakeholders of the watershed to conserve water, and the implementation of different management alternatives, according to the scenario of climatic change. De Bruin et al. (2009) and Haque (2016) considered the multi-criteria decision analysis (MCDA) as a productive technique to design climate adaption strategies for watershed management. Potential rainwater harvesting areas have been identified by using GIS-based MCDA Toosi et al. (2020). MCDA, with two weighting methods of Analytical Hierarchy Process (AHP) and FIM (Factor Interaction Method), has been used to identify suitable locations for underground reservoirs in the northern area of Pakistan (Jamali et al. 2014). GIS-based AHP and multi-criteria decision-making (MCDM) have been used to locate the runoff zones on the basis of different physical properties in Gujarat, India (Rana & Suryanarayana 2020). AHP and direct weighting approach have also been used to model the managed aquifer recharge site suitability analysis (Rahman et al. 2012; 2013).

Since there is a need to address watershed management coping with the adverse effects of dynamic factors of climate change and LUCC. Several studies have evaluated the impact of climate change and land use in the SRB of the Pothwar region (Bhatti et al. 2014; Samie et al. 2017, 2020; Shahid et al. 2018; Ghafoor et al. 2022; Ismail et al. 2022). These studies have focused on evaluating the historical and future climate change and land use change impacts using various hydrological modeling approaches and datasets from both historic records and GCMs (general circulation models). The studies mentioned above did not engage the stakeholders in watershed management through management alternatives. Few studies (Javaid et al. 2016; Israr et al. 2022; Javed & Siddiqui 2023) that carried out the MCDA for different management practices were conducted in the Pothwar region but not specifically in the SRB and studies did not cope with the problems of climate change. This study stands apart from previous studies with the implementation of a framework by designing real and hypothetical cases based on the climate periods and land use maps. This study also varies from literature studies regarding the involvement of watershed stakeholders by applying MCDA to design climate change resilient and mitigation strategies while considering the impacts of climate change.

In the SRB, a reduction in rainfall and an increment in temperature has been observed due to climate change (Usman et al. 2021). Due to the fast rate of growing population in the study area water demand has increased, causing the problem of shortage in water availability (Shirazi et al. 2020). Urbanization and deforestation have increased in the whole watershed due to which soil erosion has increased causing sedimentation in small dams and ponds (Tariq & Aziz 2015). So, to compete with this problem, watersheds demand appropriate management via suitable alternatives that align with current climate change conditions and that is why this study accomplished: (i) investigation of impact due to historic climate change and LUCC on water availability; (ii) selection/assessment of suitable alternatives for watershed management using MCDA; and (iii) evaluation of site suitability for selected suitable alternative of the watershed. The findings of the study will facilitate making climate change adaption strategies, for watershed management through suitable alternatives, according to current diversified conditions.

A statistical framework was integrated with the SWAT model to assess the climate change and LUCC impact on the watershed's hydrological cycle. Framework repercussion in terms of runoff was used as an input to MCDA. AHP was subsequently employed to adopt the best watershed management alternatives. Overall previous studies have primarily focused on the application of the model to investigate the hydrological processes rather than using MCDA by engaging the opinion of the experts of the watershed. This covers the integrated watershed management by applying the WHS and VC for soil and water management.

The Soan River Basin (SRB) is situated in Pakistan's sub-Himalayan Pothwar region. It covers an area of about 6,540 km², extending 274 km in length, with elevations ranging from 281 to 2,256 m above mean sea level (a.m.s.l). The basin is located between 72° 24′ 0″ to 73° 30′ 0″E and 32° 36′ 0″ to 33° 54′ 0″N. It starts from Murree hills and passes through metropolitan areas like Rawalpindi, Fateh Jung, Pindi Ghab, Talagang, and Mianwali. Along its course, the Ling Stream flows through Lehtrar and Kahuta, merging with the SRB near Sihala, close to the Kak Pul bridge on the Islamabad Highway. Additionally, the SRB is joined by two other streams, namely the Korang River and the Lai stream, before and after the Soan Bridge, respectively. Eventually, the stream reaches the proposed Kalabagh dam site near Pirpiyahi, where it arrives in plain areas and ends at the hydrological station of Dhok Pathan before joining the Indus River near Jand. About 30% area of SRB is hilly and the remaining 70% comprises rolling plains. The SRB has nine main tributaries mainly fed by monsoon and also some minor ones contributed from snow-melt. Its left bank tributaries are steeper and larger than the right bank tributaries. The basin's total yield has been evaluated at 1.525 MAF (million acre-feet). The topography of the area presents significant opportunities for expanding irrigation potential within the Basin. The hilly area receives more rainfall than the plain areas from 400 to 1,710 mm, respectively. The region experiences a continental, subtropical climate with hot summers and a semiarid to subhumid climate with cold winters. The mean monthly maximum temperature ranges from 35 to 41 °C, while the mean monthly minimum temperature ranges from 19 to 30 °C. In winter, the temperature in the watershed drops below zero at higher altitudes like Murree, and in the summer, the temperature can reach up to 47 °C at lower altitudes like Chakwal and Jouharabad. The land use pattern of SRB consists of agricultural land, barren land, residential land, forest, and water. The soil in the SRB exhibits a range of types, including clay loam to silty clay loam, with excellent drainage. Almost 60% of the population belongs to rural areas and the main source of income is agriculture. Major crops are wheat, vegetable, oilseeds, sorghum, maize, groundnut, fruit plantation, millets, chickpea, fodders and sugarcane. To fulfill agricultural water requirement a large number of rainwater harvesting structures (WHS), mini dams, soil water conservation structures, and ponds have been established by private and public sectors (Ashfaq et al. 2014). To meet the growing water demand of Islamabad, Simly Dam was constructed in 1983 on this river. During the last few years population is highly influenced due to unexpected fluctuations in streamflow of the SRB.

Historical climatic data of rainfall, and maximum and minimum temperatures, spanning from 1981 to 2020 on the daily basis of four stations (Murree, Islamabad, Chaklala and Chakwal) lying in SRB, was obtained from Water and Power Development Authority (WAPDA) and Pakistan Meteorological Department (PMD). The slope map was derived from the Digital Elevation Model (DEM) obtained through Shutter Radar Topographic Mission (SRTM). A soil map was obtained from the Food and Agriculture Organization (FAO) of the United Nations. Land use maps were obtained from the satellite imagery of Landsat 7 and 8 for the years 2000 and 2020, respectively. Supervised classification was performed by using a GIS interface to create the maps of land use. All of the spatial data was projected into the projected coordinate system of WGS_1984_UTM_Zone_43N for Pakistan. Streamflow data of the Dhok Pathan station was obtained from the Irrigation and Power Department. Qualitative data about the watershed's management alternatives were acquired from different water experts of different water-related departments and these experts were also requested to assign weights for the competing criteria and the alternatives (Figure 1).

In the current study, Mann–Kendall (MK) and Sen slope were used to detect the trends of climatic parameters and runoff because this trend analysis has been applied in many hydro-climatic studies (Daniel & Abate 2022; Iqbal et al. 2022; Daniel 2023). The Sen slope method was used to assess the magnitudes of changes and MK non-parametric test was applied to identify the trend in hydro-climatic variables. A complete description about MK and Sen slope has been given in Yang et al. (2017).

40 years of climatic data were used, from 1981 to 2020, to utilize the above mentioned framework. The data were divided into two parts, 1981–2000 and 2001–2020. Two land use maps were prepared for the years 2000 and 2020. The above-described framework developed the four scenarios (two Real Cases as S1 and S4 and two Hypothetical Cases as S2 and S3) and SWAT model simulations were carried out for all four scenarios to calculate the impacts. These scenarios were given in Table 1.

During the comparison of the SWAT model output for all the above four scenarios the impacts of climate change and LUCC were quantified using Equations (1)–(3).

Sensitivity analysis was carried out to evaluate the impacts of input parameters on model output. This analysis prioritizes the most sensitive parameters during calibration, which can ultimately result in reduction of model uncertainty. To calibrate the SWAT model simulated flows, 18 parameters were selected from literature studies specifically conducted in Pakistan and also in other regions (Khatun et al. 2018; Rahman et al. 2022; Aawar & Khare 2020). The selected parameters with their description are given in Table 2.

MCDA provides several alternatives or options in operational research problems, and analysts have to evaluate these alternatives with the best suitable solution in the decision-making process. It is most widely used in the field of water resources as described in the section introduction. The MCDA procedure involves the following steps.

Defining a problem is the basic step to proceed with the MCDA. A literature survey reveals that climate change has led to reduced rainfall and runoff in the SRB (Shahid et al. 2018). Therefore, it can be inferred that climatic variation is the main constraint in all agriculture and water-related sectors. Water experts, as key stakeholders, aided in identifying and prioritizing criteria and alternatives for watershed management in response to climate change.

A set of objectives, which are associated with attributes and yield the guidelines to identify alternatives, is referred to as criteria. In light of the challenges emerging in the SRB due to climate change, the identification of criteria for prioritizing watershed management alternatives was based on comprehensive literature studies and expert input from water professionals. Criteria used to select the best suitable alternative were contribution to store water, contribution to reduce floods and soil erosion, contribution to groundwater recharge, contribution to sustainable and profitable agriculture, contribution to improve surface water quality, contribution to economy, degree of community acceptance, inexpensive labor and cost, availability and simplicity of the technology, and contribution to management policy.

An alternative is a set of attributes that satisfies all the constraints. The study revealed a decline in rainfall and runoff within the SRB, prompting the identification of watershed management alternatives to facilitate the formulation of suitable strategies and policies for mitigating water scarcity. To facilitate the implementation of AHP, a meticulous assessment of the literature and expert input led to the selection of ten alternatives, namely: land use change, reclamation of degraded lands, vegetative cover (VC) for water and soil management, water harvesting, crop change, terracing, area closure, waterways maintenance, agroforestry, and new crop varieties.

Succeeding the selection of the most suitable alternatives for watershed management, site identification was conducted through suitability analysis using GIS-based weighted overlay (WOA) analysis. By using MCDA, WHSs and VC for soil and water management were declared as top prioritized alternatives. Input layers of rainfall, runoff, slope, LULC, and soil type were used as criteria to perform suitability analysis for WHS site identification. Here the AHP was also used to assign percentage weights for each criterion and these weights were used to generate a suitability map of WHS using WOA. All input raster layers for WHS were reclassified to suitability classes (suitability scores) given in Table 3. These suitability scores were based on the literature (Durbude & Venkatesh 2004; FAO 2016; Grum et al. 2016; Al-Ruzouq et al. 2019; Badhe et al. 2020; Kar et al. 2020; Badapalli et al. 2021; Hoque et al. 2022).

Similarly, input raster layers of rainfall, slope elevation, LULC, soil type, potential evapotranspiration, soil organic content, temperature, and aspect were used for site identification of VC for soil and water management. Again, the AHP was also used to assign percentage weights for each criterion and these weights were used to generate a suitability map using WOA. All input raster layers for WHS were reclassified to suitability classes (suitability scores) given in Table 4. These suitability scores were based on the literature (McVicar et al. 2010; Akinci et al. 2013; Stanchi et al. 2013; Otgonbayar et al. 2017; Hassan et al. 2020; AL-Taani et al. 2021; Binte Mostafiz et al. 2021).

However, both maximum and minimum temperatures have an increasing trend moving in the SRB. One of the reasons for the increase in temperature is the land use change pattern because it is directly linked with human activities which cause atmospheric changes. In the SRB forest land and vegetation cover have been converted into buildup areas for household, industrial units with the aim of job opportunities and into roads for transportation to meet the higher income. All these factors increase atmospheric temperature by increasing the concentration of greenhouse and carbon dioxide gases (Shirazi et al. 2020).

The land use of forest has reduced by an area of 206 km² (3.15%), the settlement has increased by an area of 454.84 km² (6.96%), water has increased by an area of 1.5 km² (0.02%), bare soil has reduced by an area of 1,340.25 km² (20.49%) and agricultural land has increased by an area of 1,089.31 km² (16.66%) during 2000–2020. Details of the total change in land use are given in Table 5.

The major class of land use change is the reduction of bare soil from 51.71% (3,381.48 km²) to 31.22% (2,041.23 km²) from 2000 to 2020, respectively, with a total change of 20.49% (1,340.25 km²). This was followed by an increase in agriculture land from 24.06% (1,573.65 km²) to 40.72% (2,662.96 km²) from 2000 to 2020, respectively, with the total change of 16.16% (1,089.31 km²). From these changes in land use of the SRB, it is indicated that there was the transformation of bare soil to the increment of agricultural land.

According to Gill & Mushtaq (1998), the increment in agricultural land in the SRB is because of Barani Village Development Program (BVDP). In this development project, 161 ponds and 200 mini dams were constructed due to which agriculture activities were increased. Majeed et al. (2010) analyzed the agricultural benefits due to mini dams. It was estimated that before the BVDP project 61% of farmers were not using their land for agriculture and only 33% were doing rainfed agriculture. After the completion of this project, 85% farmers started agriculture activities and started to grow high value crops with high cash return value to increase the earnings of farmers. Shirazi et al. (2020) observed that the population rate and people's movement trend toward cities, to improve their living standards, is increasing due to which urban area is expanding and vegetation cover is also converting into built-up areas.

To account for the uncertainties of model simulations p-factor should close to 1 and r-factor should close to 0. From the values of p-factor and r-factor, given in Table 6 for both real cases, it was stated that the results are acceptable. Findings ensure from similar studies done by Thavhana et al. (2018); Leta et al. (2022); and He & Bao (2021). From the results of sensitivity analysis the most sensitive parameters were found, given in Table 7, based on the values of t-stat and p-value because larger the t-stat and smaller the p-value gives more sensitivity to the parameter (Abbaspour et al. 2007). For real case 1 most sensitive parameters were ESCO, GE_DELAY, SLSUBBSN, CH_K2, CN2, SOL_AWC, EPCO, and CANMX. Also, the most sensitive parameters for real case 2 were CN, CANMX, ALPHA_BF, CH_K2, SMFMX, and HRU_SLP. The least sensitive parameters for real case 2 were SOL_BD and EPCO. The results confirm similar studies about the parameter's sensitivity conducted by Almeida et al. (2018), Mengistu et al. (2019), and Nazari et al. (2020).

So, the performance of the SWAT model for runoff in both real cases was declared as ‘good performance’ according to the value ranges of NSE and R² according to Moriasi et al. (2007). So, these findings indicate that the SWAT model is an effective runoff simulation tool for assessing the effects of climate change and land use change on catchments.

To assess the impacts of climate change and LUCC, simulated flows were used of four scenarios. An intercomparison between S1 with S4 revealed that runoff has decreased, and evapotranspiration (ET) has increased from 1980 to 2020.

Scenarios S1 and S3, as well as S2 and S4, were used to estimate the effects of climate change on runoff and ET. For this purpose, Equation (1) was applied to assess the impacts of climate change. The results showed that due to climate change, runoff has reduced by 69.18% and ET has increased by 57.92% as shown in Table 8.

Impacts of land use change on runoff and ET were estimated by taking the difference among scenarios S1 and S2, as well as scenarios S3 and S4. For this purpose, Equation (2) of land use cover change was used. So, results showed that due to land use change, runoff has reduced by 30.82% and ET has increased by 42.08% shown in Table 8. An increase in agricultural land or vegetation is one of the potential causes of the ET increase in the SRB. The building of small dams in the basin of SRB can also be attributed to runoff reduction.

From the above results, it can be concluded that runoff has reduced, and ET has increased under the impacts of climate change and land use change from 1981 to 2020. A comparison was made between the impacts of climate change and land use change on runoff and ET, it was noted that climatic factors have more impacts than land use factors in SRB. According to this finding, climate change has been the most important factor influencing runoff, while LUCC had a smaller impact.

The observed variations of river flow in Pakistan are an outcome of unpredicted precipitation trends. It has been noted that rainfall has decreased in the SRB; therefore, reduction in runoff was being observed in this basin. Due to increase in population water consumption increased and it reduced the runoff. Reservoirs and ponds construction is also attributed to decrease in runoff and a number of small dams were observed in the SRB. That is another reason for decrease in the runoff.

The observed increase in agricultural land has resulted in increasing rate of ET. Because of mini dam construction, water storage has increased, due to which more water is exposed to sunlight, also causing an increase in ET. It has also been observed that built-up area has increased, due to which less infiltration and more ET occur.

Using AHP, a comparison matrix was calculated among the criteria, given in Table 9, which showed that the contribution to store water accounted for 21.29% of the total weight, was the most prioritized criteria which was followed by contribution to reduce floods and soil erosion (19.39%), contribution of sustainable and profitable agriculture (15.92%), contribution to economy (11.83%), degree of community acceptance (10.14%), contribution to improve surface water quality (8.18%), contribution to ground water recharge (5.09%), availability and simplicity of technology (3.74%), contribution. to management policy (2.65%), and inexpensive labor and cost (3.40%).

So, from the overall priority matrix shown in Table 10, the WHS was the most suitable and highly prioritized watershed management alternative with the priority percentage of 22.46%. VC for soil and water management was the second most prioritized alternative with the priority percentage of 17.05%. These two alternatives were followed by terracing (11.31%), reclamation of degraded land (8.87%), new cash crop varieties (8.28%), agroforestry (7.95%), crop change (7.90%), water ways maintenance (6.53%), land use change (5.07%), and area closure (4.56%). Water harvesting has been the most important technique with regard to climate change to design the climate change adaptation strategies and policies, (FDRE-NAP 2019). So, now the the point was how to obtain the maximum benefits of WHS and VC according to our requirement. In that sense, suitable site identification was the process to get more benefits (Dile et al. 2016).

The number of studies were conducted to assess the impacts of climate change and LUCC on hydrological processes in different basins of the world such as Bengawan Solo River, Indonesia (Marhaento et al. 2021), Eastern Baltic Sea region (Marhaento et al. 2021), Loess Plateau of China (Li et al. 2009), Upper Ganga basin (UGB) in India (Chawla & Mujumdar 2015), Haihe River Catchment, China (Yang & Tian 2009), Southeastern United States (Hung et al. 2020) and Upper Blue Nile River basin in Ethiopia (Mekonnen et al. 2018). All these studies revealed a significant and insignificant decrease in runoff due to climate change and LUCC. Furthermore, these studies also quoted that climatic factors have more impacts on hydrological processes than land use. However, the above quoted literature was found similar and consistent with the current study. So, to compete this water shortage, various studies were conducted in the world such as drought-prone areas of Bangladesh (Hoque et al. 2022), Arsha and Balarampur Blocks, Purulia (Kar et al. 2020), Upper Sina River Catchment, Ahmednagar India (Badhe et al. 2020) and upper Geba watershed, Ethiopia (Badapalli et al. 2021). These studies revealed WHSs as the best management alternatives to meet the water shortage and also identified the suitable sites by employing the suitability analysis through GIS-based MCDA. Furthermore, various studies were conducted in the world, such as Bangladesh (Binte Mostafiz et al. 2021), Mongolia (Otgonbayar et al. 2017), Azad Jammu and Kashmir (Hassan et al. 2020) and Ma'an Governorate, Jordan (AL-Taani et al. 2021) to control the soil erosion, to recharge the ground water, to control the water pollution and to increase the agriculture development. These studies depicted the suitable land sites for agriculture like VC grass, shrubs, etc. with the utilization of MCDA and AHP based GIS. Current study findings were consistent with the previous literature studies by identifying the suitable sites of VC for soil and water management. This study provides the solution to compete the climate change and LUCC impacts with the integration of watershed management by applying best management practices and alternatives.

The growing deficit of water raised the need for watershed management because hydrological processes of watersheds are influenced by the factors of climate change and LUCC. Preliminary analysis of historical rainfall and temperature data showed that rainfall has a decreasing trend and temperature has an increasing trend in the SRB. This study was deliberated to design the watershed management strategies to compete with climate change impacts via best suitable alternatives. For this, the framework was utilized with the application of the SWAT model to assess the impacts of climate change and LUCC. MCDA was applied by involving the stakeholders to make the intercomparison between criteria and alternatives with the help of AHP. Suitability analysis was performed to identify the suitable sites for the selected alternatives.

The LUCC findings revealed the reduction in forest and bare soil with an area of 3.15 and 20.49%, respectively, and expansion in the area of settlement, water and agricultural land by 6.96, 0.02, and 16.66%, respectively, from 2000 to 2020. Framework results showed that runoff has reduced 69 and 31% due to climate change and LUCC, respectively. ET has increased 58 and 42% due to climate change and LUCC, respectively. AHP showed that WHS and VC have been found to be the most suitable alternatives with the priority percentage of 22.46 and 17.05%, respectively, and suitability analysis found their suitability for 70–90% of the basin area. So, it can be concluded that climatic factors have more effects than land use factors on hydrological processes. Thus, there should be construction of WHSs to control the floods and drought and vegetative cover should be adopted to control the soil erosion. This study will facilitate in selecting suitable climate adaptive watershed management alternatives through integrating the expert's perspectives about climate change, to provide continuous supply of water for food production and human use, and protective measures for flood and drought control. This study can assist the policymakers to design climate change adaptation strategies, plans, and policies for the future development of climate-sensitive watershed management.

This included the rainfall, maximum and minimum temperature to assess the climate change and considered the remaining parameters as constant. This study was about the historical climate change not the future projections of climate change. Only site identification was performed, in terms of area, for watershed management alternatives but did not quantify how much runoff would be conserved, and soil erosion would be controlled. So, future studies could consider the GCMs and RCMs to investigate the impacts of future climate change. Future research could also consider how much water can be conserved, and soil erosion can be controlled by adopting such measures with respect to future development. Future research studies could also consider the integration of socio-economic factors in the decision-making strategies.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.