The present study was conducted in Lahore district, which is the capital of Punjab province and second most populated area of Pakistan. According to Zahran et al. (2023a) the most recent census of 2017 showed that Lahore’s total pollution increased from 6.3 million in 1998 to 11.2 million at an annual increment of 3%. Lahore district has a total area of 1842 km² and is situated on the left bank of the Ravi River. It is bounded on the northwest by the Ravi River and on the east by the Bambanwali Ravi Bedian Dipalpur (BRBD) Canal. The study area is geographically situated in the northeast area of Punjab between 31°15′ to 31°43′ N and 74°10′ to 74°39′ E.

Lahore is known for its cultural diversity and historical importance because of the city’s abundance of Mughal-era monuments, gardens, and parks. However, substantial growth of urban infrastructure in this metropolitan area, especially the increase in vehicular emissions, has altered the city’s landscape and badly impacted the gorgeous locations. The city of Lahore is divided into nine administration townships that are overseen by the Town Municipal Administration (TMA) and army cantonment area. The wintertime extends from November to March, and the chilly months of winter wherein the temperature ranges from 0°C to 3°C are December and January. The normal daily temperature ranges from 40°C to 45°C, with the highest temperature of about 48°C being recorded in the months of May and June (Tariq and Mumtaz, 2023). Lahore has frequent precipitations of about 600 mm in the months of July and August during the monsoon period; the average annual rainfall values for 2005, 2010, 2015, and 2020 were 580, 660, 850, and 940 mm, respectively, while the average rainfall over the period of 2005–2020 was 662 mm/year. The study area for the present study is shown in Figure 1.

Multiple datasets were used in the present study to prepare different factors. Data from both satellite and ground measurements are included in these datasets. Table 1 lists the specifics of the data and details about the sources.

The comprehensive methodological flowchart for the present study is shown in Figure 2. After acquisition, the data were processed to obtain the derivatives. Moreover, the images were classified to obtain LULC maps at 5-year intervals and validated using multiple software (e.g., QGIS, GEE, ArcGIS). After preparation of the dataset, the Kendall tau test was applied to determine the associations between the variables, and the GWR analysis was performed to find the spatial correlations between the dependent and explanatory variables to extract the results.

LULC classification analysis was performed using a multitemporal dataset from the Landsat series 5, 7, and 8 at intervals of 5 years (2005, 2010, 2015, and 2020). This interval was chosen to capture the changes in the LULC patterns over time and their impacts on the groundwater resources. The Landsat images were selected based on their spatial resolution (30 m) and availability for the study area. The LULC data were processed using the random forest (RF) algorithm in the Google Earth Engine (GEE) platform. The RF algorithm was chosen for its high accuracy and ability to handle large datasets. The training dataset consisted of 70% of the total samples, and the remaining 30% were used for validation. The number of trees generated was set to 100, and the maximum depth of each tree was set to 10. To ensure data integrity and accuracy, error lines present in the Landsat 7 imagery were removed using QGIS software. The processed data were then exported to ArcGIS 10.8 for further analyses.

The geospatial modeling techniques used in this study, including RF and Kappa coefficient (KC), were chosen on the basis of their abilities to handle large datasets and provide accurate results. QGIS, GEE, and ArcGIS were used for the data processing, analyses, and visualization, respectively, owing to their capabilities for handling geospatial data and performing spatial analyses.

In the study area, the spatial distributions of the four primary LULC classes is illustrated for the years 2005, 2010, 2015, and 2020 in Figure 3 and Table 6. These show that the territory covered by the bare ground decreased from 434.63 km² in 2005 to 161.18 km² in 2010, 57.09 km² in 2015, and 81.30 km² in 2020. The present temporal-based investigations revealed that urbanization in the study area increased with time, i.e., 341.85 km² in 2005, 506.25 km² in 2010, 756.76 km² in 2015, and 950.29 km² in 2020, as shown in Table 6. From the results shown in Table 6, it is determined that the vegetation trend shows a dramatic decline in the study area. The vegetation in the research region decreased from 1,265.45 km² in 2005 to 1,229.09 km² in 2010, 1,084.89 km² in 2015, and 1,023.91 km² in 2020. The spatiotemporal results of this study reveal that the surface water bodies has areas of 32.48 km² in 2005, 31.46 km² in 2010, 18.91 km² in 2015, and 10.42 km² in 2020. The results of the spatiotemporal analyses of the LULC show that the water bodies, vegetation cover, and barren land have been gradually decreasing while the urban environment has consistently increased in the study area.

The present study reveals that urbanization and climate changes have significantly impacted groundwater resources in Lahore district of Pakistan. The spatiotemporal LULC maps (Figures 3A–D) show a decline in vegetation and an increase in urban areas over time, leading to decreased GWLs. The built-up index expansion is inversely related to the GWL degradation, indicating that the growing urban environment is the primary cause of groundwater decline. The GWL maps (Figures 3E–L) show significant decreases over time, with more pronounced declines during the post monsoon seasons. The spatial maps (Figures 3A–L) demonstrate direct relationships between the LULC changes and decline in GWLs.

The study area is located in a semi-arid, subtropical environment. Here, the monthly mean rainfall is low, and Figure 4 illustrates the monthly fluctuations of precipitation. The significant amounts of precipitation occur during the monsoon seasons in July and August, which substantially contribute to the yearly recharging of aquifers. The annual average rainfall trends from 2005 to 2020 are shown in Figures 4A–D.

Figures 4E–H show the spatial patterns of the LST over Lahore region for the years 2005, 2010, 2015, and 2020, respectively. The analytical outputs of the present study shown in Figure 4E illustrate that the average temperature range was 27°C–32°C in 2005. Figures 4F, G show that the mean LSTs for 2010 and 2015 were 28°C–33°C and 28°C–35°C, respectively. Figure 4H reveals that the average LST for 2020 was 29°C–36°C. The spatiotemporal results of the LST imply that the LST increased with time in the study area. It was observed that the temperature was significantly more severe in the urban environment of the study area relative to the rural regions, indicating the presence of the urban heat island (UHI) effect with apparent severity. The recorded UHI effects have been increasing over the past 20 years, which may be attributed to elements like the growing human population, increase in anthropogenic activities, and changes in the LULC.

To analyze the potential combined impacts of changes in LULC and climatic conditions on the geographical and temporal variations of GWLs between 2005 and 2020, four specific time points were chosen: 2005, 2010, 2015, and 2020. To compensate for the insufficient number of wells in the study region, a total of 27 monitoring wells were randomly selected along with their bordering areas to ensure more accurate results. Figure 6 displays the quantile–quantile (QQ) plots of the pre and post monsoon GWLs at various time intervals. A QQ plot is a visual tool for evaluating the conformity of a dataset to a specific distribution, such as the normal distribution. QQ plots are useful for visualizing changes in the distributions of GWLs over different years and offer insights into the form and variability of the data distribution over time through comparisons of the empirical quantiles of the data to theoretical quantiles of a normal distribution.

The results of the present study show that LULC has significant influence on the groundwater resources. We illustrate that the vegetation decreased over time while the urban environment grew rapidly, adversely disturbing the groundwater table in the study area. In contrast, the number of impermeable surfaces in the Lahore region of Pakistan has been increasing over time. This shows that groundwater resources declined as the built-up index expanded, indicating an inverse relationship between the impervious areas and GWL degradation.

The spatiotemporal LULC maps of the study area for 2005, 2010, 2015, and 2020 are shown in Figures 3A–D; the corresponding groundwater depth maps for the pre and post monsoon seasons are shown in Figures 3E–G and Figures 3H–L, respectively. By comparing the spatial maps in Figures 3A–C with those in Figures 3D–G and Figures 3H–L, it is seen that there was a direct relationship between LULC change and declining GWLs. The QQ plots (Figure 6) show that the study area’s pre and post monsoon GWLs are significantly affected by climate changes. These results clearly indicate that urbanization and climate changes are synergistically impacting groundwater resources, leading to declines in GWLs and groundwater quality. Furthermore, the results of the Kendall’s tau test (Table 7) reveal significant associations between the climate variables and groundwater. Specifically, there is a substantial negative correlation (−0.333) between groundwater quantity and annual average temperature, indicating that rising temperatures are contributing to groundwater depletion. On the other hand, there is a favorable association (0.666) between the yearly groundwater depth and mean precipitation, suggesting that increased precipitation can help mitigate groundwater decline.

Most of the water used in Lahore is sourced from the groundwater table rather than other sources. To meet the expanding water needs of metropolitan areas and the farming industry, constant water extraction has been investigated over the past 20 years. The spatiotemporal study results of the climatic variables are shown in Figure 4. The yearly average rainfall and temperature in the research region increased steadily over the period of 2005–2020. However, owing to the growing urban environment, the infiltration of precipitation into the soil was disrupted along with the LST. This resulted in loss of vegetation, which in turn hindered the replenishment of groundwater reservoirs.

The GWR analysis was performed in the study area to examine the spatial correlations between the dependent and explanatory variables (within one modeling approach). The dependent variable here was groundwater, while the explanatory variables were the population and built-up area. The findings of the GWR analysis, which illustrate how the local R² varies throughout the five main regions (Tehsils) in Lahore, are shown in Figure 5. The results show an interesting spatial pattern and positive significance values. For Lahore, the entire area was split into four classes in the GWR analysis: 0.55–0.56, 0.57–0.60, 0.61–0.66, and 0.67–0.71. The variations in the built-up areas in the Raiwind Tehsil ward are strongly connected with changes in the population and GWL, as shown in Figure 5. The local R² values for the northern, northeastern, and southern portions of Lahore are between 0.55 and 0.60. The model’s importance is demonstrated by the global R² of 0.84. Overall, these findings imply that limiting the population increase is necessary to ensure sustainability of urbanization.

However, a difference was observed in the relationship between GWL and urban or built-up areas, which has strong impacts on urban planning of Lahore city from an environmental perspective. First, the results show that urbanization processes, which are shown by growing built-up areas and population density, have clear effects on how the groundwater moves. The strong link observed in the Raiwind Tehsil ward shows how important it is to target interventions in areas that are urbanizing rapidly to manage groundwater resources well. Further, the moderate to strong correlations seen in different parts of Lahore show that local factors are very important in shaping groundwater movements. This shows the importance of handling groundwater in a manner tailored to each situation rather than simply using global or regional models. The large global R² value implies that the model can explain most of the changes in the GWLs across Lahore. This shows that GWR analysis is a useful method for finding links that are spread out over different areas of a city.

Overall, the results show the importance of developing plans for integrated land use and water resources management to ensure that Lahore’s urbanization will be sustainable. Limiting the population growth and controlling the urban sprawl are important steps in protecting groundwater supplies from the damage that come with urbanization and promotion of long-term sustainability. By highlighting the specific impacts of urbanization and climate changes on groundwater resources, this study aims to provide valuable insights for policymakers and stakeholders to develop effective strategies for sustainable groundwater management.