Assessing the impact of land use land cover changes on soil moisture and vegetation cover in Southern Punjab, Pakistan using multi-temporal satellite data

ABSTRACT

Change in land use land cover (LULC) is driven by human activities and drives changes that limit the accessibility of services and products for livestock and humans. This study aimed to develop the spatio-temporal changes in LULC, vegetation cover, and moisture index using multi-temporal satellite data in Southern Punjab, Pakistan. Analysis of satellite data was accomplished using ArcGIS 10.4 software. Based on ground-truthing, the supervised classification technique (maximum likelihood algorithm) was used to achieve the LULC classification. The LULC change analysis revealed that vegetation area converted to the build-up area by 7.17 % and bare soil converted to urban areas by 1.68 % from 2000 to 2021. Average NDVI values were calculated at 0.23, 0.17, 0.19, and 0.14 for 2000, 2007, 2014, and 2021, respectively. In the study area, average NDMI values were observed at 0.28, 0.25, 0.2, and 0.15 for 2000, 2007, 2014, and 2021, respectively. Based on our study, the general trend in Southern Punjab is a decrease in vegetation cover, moisture index, and forest area due to an increase in build-up areas. Assessments of LULC and NDMI changes, as well as estimates of the effect on the environment, are necessary for many policy decisions and planning.

KEYWORDS:

• Land use land cover
• change detection
• geographic information system
• normalized difference vegetation index
• normalized difference moisture index
• remote sensing

1. Introduction

The study of land use land cover (LULC) changes is very important in defining how the land is presently used and delivers a starting point for current and future planning (Abdullah et al., 2019; Hussain & Karuppannan, 2023; Yamamoto & Ishikawa, 2020). The LULC has become a fundamental component in present approaches for managing environmental changes and observing natural resources (Akram et al., 2022a, 2022b; Bhagyanagar et al., 2012; Orimoloye et al., 2018). The LULC changes occur when there are alterations in how land is used or in the composition of the land cover (Akram et al., 2018; Masood et al., 2022). These changes can be driven by various factors, including urbanization, deforestation, agricultural expansion (Cristóbal et al., 2018; Khaliq et al., 2022), industrialization, infrastructure development, natural events (like wildfires or floods), and climate change (Ahmed et al., 2022; Aredehey et al., 2018). Understanding LULC is crucial for a variety of reasons, including environmental management, resource planning, disaster risk assessment, biodiversity conservation, climate change analysis, and urban development (Chew et al., 2016; Din et al., 2022; Hussain, 2018). Remote sensing (RS) technologies, satellite imagery, geographic information systems (GIS), and other tools are used to monitor and analyze these changes over time (Afzal et al., 2023; Choudhury et al., 2018). Researchers, policymakers, and environmental organizations often study LULCC patterns to assess their impacts on ecosystems, habitats, biodiversity, hydrology, air quality, and the overall health of the planet (Ahmad, 2012; Akhtar et al., 2017; A. Ali et al., 2023; A. M. A. Athick et al., 2019; Hussain, Ahmad, et al., 2020; Hussain, Mubeen, Ahmad, et al., 2023).

Normalized difference vegetation index (NDVI) is a critical tool for assessing the health and density of vegetation across landscapes (Almazroui et al., 2017; Ayele et al., 2018). It provides a numerical representation of the greenness and vitality of plant cover based on the reflectance of near-infrared (NIR) and red light (Aboelnour & Engel, 2018; Forkel et al., 2013). High NDVI values indicate healthy vegetation, while low values suggest sparse or stressed vegetation (Ding & Shi, 2013; M. T. U. Rahman et al., 2017). NDVI is extensively used in agriculture to monitor crop health, estimate yields, and identify areas prone to drought or pest infestations (Gilani et al., 2015; Hussain, Mubeen, et al., 2020; Safder, 2019). In ecological studies, NDVI helps in assessing changes in forest cover, tracking vegetation recovery after disturbances, and understanding the impacts of climate change on ecosystems (Bento et al., 2018; Hashim et al., 2020; Hussain, Mubeen, Ahmad, Fahad, et al., 2021, b). Due to the procedure used to produce this product, its features have numerous characteristic artefacts that require detection and elimination to deliver a sound ecological explanation (M. Ali et al., 2019; Cai et al., 2018). Elimination of these artefacts is done using a quality layer, which delivers absolute quality data for each pixel in the 16-day composited NDVI data (A. A. S. M. Athick & Shankar, 2019; Hussain, Mubeen, Jatoi, et al., 2023; Mubeen et al., 2021; Omran, 2012).

Normalized difference moisture index (NDMI) has been considered one of the important indicators widely used in ecology, hydrology (Yuan et al., 2015), and current soil moisture, as well as plays an important role in the interactions between energy and water at the air-soil interface (Hussain, Mubeen, et al., 2020; Huyen et al., 2016; Nayak & Fulekar, 2017). The NDMI is a significant parameter in hydrological modelling, agriculture, climate, biogeochemistry, hydrology, and ecology (Islam et al., 2021; Xu et al., 2016). Differences in NDMI produce important changes in regional runoff and surface energy balance (Mohammed et al., 2019; Olmanson et al., 2016). The NDMI plays a significant part in crop production, including vegetation, climate, drought, and agriculture (Kazmi et al., 2023; Zahoor et al., 2019). By NDMI and crops’ water requirements in various phenological stages, one can reveal whether the crops are exposed to water stress and predict crop yields (Karuppasamy et al., 2022; A. Naz & Rasheed, 2017). The NDMI is an important parameter that directly and directly affects the water cycle (Das & Sarkar, 2019; Hu et al., 2023; Hussain, Mubeen, Nasim, et al., 2023; Sabagh et al., 2020; Sinha et al., 2012).

The RS is a significant technique that detects data about the object placed remotely without being in contact with it (Kharazmi et al., 2018; Sahana et al., 2016). Over the last few years, RS has become a useful tool for supporting humans in their progress in resolving related ecological tasks on a worldwide, regional, and local scale (Kidane et al., 2019; Romaguera et al., 2018). The RS technique has been applied to the classification mapping of LULC changes with various methods and databases (Pal & Ziaul, 2017; Sadiq Khan et al., 2020). The method of post-classification assessment of LULC changes was frequently used to define areas of vegetation changes (Khan et al., 2019; Tariq et al., 2023). Pakistan is one of the countries most affected due to climate change between the climate-affected countries (Hussain et al., 2022; Ullah et al., 2019). Most farmers in Punjab, Pakistan, are affected by the rapid increase in urban areas, climate change, change in cropping patterns, and decreasing irrigation water and rainfall because they depend on agriculture production to secure their livelihood (S. Naz et al., 2022). Southern Punjab is well known for crop production in Pakistan, but climate change and decreasing irrigation water affect various crop productions in Southern Punjab (Yang et al., 2023). Due to climate change, Southern Punjab has mostly been affected by water bodies and vegetation cover in the past few years (Nasim et al., 2018; Rizvi et al., 2020; Hussain, Amin, et al., 2022, p. c). This article reviews various LULC systems: forests, vegetation, build-up areas, bare soil, and water bodies. Therefore, the current study was executed to estimate LULC, NDVI, and NDMI changes in seasonal intervals from 21 years (2000 to 2021) to generate an accurate database on LULC for the said period in Southern Punjab (Pakistan). The main objectives of our research are:

1. To derive change detection of LULC from optical satellite images using a classification process in Southern Punjab, Pakistan.

2. To calculate the change analysis of NDVI and NDMI from 2000 to 2021 in the study area.

3. To identify the relationship between NDVI and NDMI between 2000 and 2021 in Southern Punjab, Pakistan.

2. Materials and method

2.1. Study site and survey

The research site is bounded by Multan, Vehari, and Lodhran districts of Southern Punjab, Pakistan (Figure 1). The study area lies between latitude 29°19’ 11^’’ N to 30 °28 ^’16^’’ N and longitude 70° 58’ 34^’’ E to 71°43’ 25^’’ E. District Khanewala and Sahiwal bound the study area on the north, the district Bahawalpur on the south, the district Pakpattan on the east, and the district Dera Ghazi Khan on the west. The district Vehari has an overall area of 4,364 km² divided into three tehsils Vehari, Burewala, and Mailsi. The district Multan has an overall area of 3,720 km² divided into 4 tehsils Multan Sadder, Multan City, Shujabad, and Jalalpur Pirwala. The district Lodhran has an overall area of 2,778 square kilometers divided into three tehsils Lodhran, Dunyapur, and Kahror Pacca (Census Report 2017). The area of districts Multan and Vehari is very productive and plain, with the River of Chenab passing on the western side and the River of Sutlej passing on the southern side, respectively. The total area with mild winters and sweltering summer’s features an arid climate [76]. The study area witnessed some of the most extreme weather, with the highest recorded temperature of approximately 52°C (126 °F), and the lowest recorded temperature of approximately − 1°C (30 °F).

Figure 1. Study area and survey points of districts Lodhran, Vehari and Multan.

The survey was conducted in 30 union councils for each district (Multan, Lodhran, and Vehari) to collect local persons’ LULC and climate change data. And in each union council, three villages were selected for the survey, and three or four people were interviewed in each village. A simple random technique (SRT) was used for the selection of survey locations. Software GPS (Geographic Position System) Essential was used to collect survey information from the farmers. It is a mobile-based application that allows collecting georeferenced as well as digital survey information. An example of the survey form used during the survey can be found in Figure 1. Survey Points of district Vehari are Sheikh Fazil, Chak 215 WB, Haji Sher, Chak 1 WB, Luddan, Karam Pur, Chak 32 WB, Arain Wahan, Chak Rasool Pura, and Tiba Sultan Pur. Survey Points of district Lodhran are Qatal Pur, Chak 5 M, Waris Shahwala, LailPur, Kahror Pacca, Sumra, JallaArain, Basti Rasool Pur, Ali Wala, and Abdullah Wala. Survey Points of district Multan are Muhammad Wala, Abadi Adda Bosan, Rafiq Abad, MauzaAli Wala, Larr, Tibba Salab Abad, Adda Pul Khara, Manik Wali, Basti Sadiq Wali, and Bahadar Pur.

2.2. Data set and image classification

The information about LULC is mostly used to describe the land properties like vegetation area, bare soil, forest, water bodies’ and build-up area etc., shown in Table 1. The Landsat data were downloaded freely from the official NASA (National Aeronautics and Space Administration) Earth Explorer United States Geological Survey (USGS) website (earthexplorer.usgs.gov), shown in Table 2. For identification of LULC and NDMI changes, the Landsat image of Landsat 5 Thematic Mapper (TM) for the year 2000, Enhanced TM Plus (ETM+), Landsat 7 for the year 2007, and Operational Land Imager (OLI), Landsat 8, for the years 2014 and 2021 were used.

Table 1. LULC classification pattern [64].

LULC types	LULC Description
Forest area	Land which is covered by tree-crown areal density with trees capable of producing timber as well as other wood products.
Vegetation area	Natural vegetation area, vegetated land, crop fields, parks, and agricultural regions, etc.
build-up	residential and Commercial, buildings, and road systems.
Bare soil	Open land, empty and Unused areas, bare soil, and fallow areas etc.
Water-bodies	low lying lands, canals, lakes, marshy lands, swamps, River, and ponds.

Table 2. Details of Landsat images with their specification.

No.	Satellite/Sensor	Pixel size	Spectral resolution	Band used	Path/Row	Date
1	LANDSAT 5	30 m	Multispectral (6 bands)	1,2,3,4,5,7	150/039	March, 2000
					150/040	March, 2000
2	LANDSAT 7	30 m	Multispectral (8 bands)	1,2,3,4,5,7	150/039	March, 2007
					150/040	March, 2007
3	LANDSAT 8	30 m	Multispectral (11 bands)	1,2,3,4,5,6,7,9	150/039	March, 2014
					150/040	March, 2014
4	LANDSAT 8	30 m	Multispectral (11 bands)	1,2,3,4,5,6,7,9	150/039	March, 2021
					150/040	March, 2021

Landsat images were pre-processed in the software ERDAS Imagine 15 for layer stacking, mosaicking, and sub-setting of the image according to the study area (Zaidi et al., 2017). We pre-process the Landsat images to correct for atmospheric effects, radiometric calibration, and geometric alignment (Zhang et al., 2016). After pre-processing, the supervised classification scheme was applied to temporal satellite datasets for the years 2000, 2007, 2014, and 2021, explaining the maximum likelihood algorithm to identify LULC changes observed in the study area. A supervised classification method was applied to 2000, 2007, 2014, and 2021 Landsat images to generate classified LULC maps. For each LULC type, training samples were selected by delimiting polygons around different demonstrative sites using the software Arc GIS 10.4 (Hussain, Lu, et al., 2022; B. Zoungrana et al., 2015).

2.3. Estimation of NDVI and NDMI

The NDVI was used to identify the different regions’ healthy and weak vegetation areas. The NDMI was used to extract the moisture area in the different areas. Software Arc GIS 10.4 was applied to calculate NDVI and NDMI (Majeed et al., 2021; B. J. Zoungrana et al., 2018).

(1) $\mathit{NDVI}=\frac{\mathit{NIR}-\mathit{RED}}{\mathit{NIR}+\mathit{RED}}\dots \dots \dots \dots \dots \dots \dots \dots \dots$(1)

(2) $\mathit{NDMI}=\frac{\mathit{NIR}-\mathit{MIR}}{\mathit{NIR}+\mathit{MIR}}\dots \dots \dots \dots \dots \dots \dots \dots \dots ..$(2)

Where NIR is a near-infrared band (TM and ETM+ band 4, OLI band 5), RED is the red band (TM and ETM+ band 3, OLI band 4), and MIR is a middle infrared band (TM and ETM+ band 5, OLI band 6). These indices, like NDMI and NDVI, were estimated using Landsat images and have values between −1 and 1.

2.4. Assessing accuracy assessment

Accuracy assessment is an important stage in evaluating different image processing procedures in image classification (Das & Angadi, 2020). For suitable research results, there should be a great level of sureness in the results that is what accuracy assessment. An error matrix is applied to evaluate the accuracy of classified images (Hc et al., 2020). For the accuracy assessment of LULC maps extracted from Landsat images, the simple random technique (SRT) was used to denote various LULC classes in the study area (Hussain, Mubeen, et al., 2022).

(3) $\mathrm{Overall}\mathrm{accuracy}=\frac{\mathrm{number}\mathrm{of}\mathrm{sampling}\mathrm{classes}\mathrm{classified}\mathrm{correctly}}{\mathrm{number}\mathrm{of}\mathrm{reference}\mathrm{sampling}\mathrm{classes}}$(3)

The KHAT (k) values are an estimate of how well RS classification is correct and agrees with reference data. Conceptually, k can be defined as (Mukherjee & Singh, 2020) [57]:

(4) $\mathrm{k}=\frac{\mathrm{Observed}\mathrm{Accuracy}-\mathrm{Chance}\mathrm{Assessment}}{1-\mathrm{Chance}\mathrm{Agreement}}\dots \dots \dots \dots$(4)

This statistic indicates the extent to which the percentage correct values of an error matrix are due to the “true” vs. “chance” agreement.

The details of the methodological and analytical steps are described in Figure 2.

Figure 2. Details of classification and methodological steps.

3. Results

3.1. Areal distribution of LULC classes

The supervised classification analysis for the years 2000, 2007, 2014, and 2021 and the area were covered with different LULC classes in Southern Punjab (Figure 3). A supervised classification method of LULC classes was used along with a reconnaissance survey, GIS data, and additional information for the study area. In the year 2000, water bodies were (3.2%) followed by bare soil at about (4%); the area covered by vegetation area was (88.3%), while the build-up area was (2.8%), and forest area was (1.7%). The 2007 image analysis of the study area showed that the vegetation area accounts for 87.4%, while the water bodies’ build-up area and bare soil covered 2%, 6.2%, and 3.2%, respectively. For the 2014 image, the vegetation area accounts for 86.7% of the overall area, followed by the build-up area with a value of 7.6%. Water bodies and bare soil had an almost equal coverage of 1.7% and 2.9%, respectively. Similarly, in the year 2021, water bodies were (1%) followed by bare soil of about (2.9%); vegetation area was (84.9%); however, build-up area was (10.3%), and forest area was (0.9%) as shown in Figure 4. It is denoted that during the last 18 years, almost 4.6% of vegetation area has been converted to roads and build-up areas. It was observed that there had been fast changes in LULC in the study, particularly in vegetation areas and forest areas. The vegetation area decreased by 0.58% during 2000 to 2014 as well as 2.7% during 2000 to 2021 as vegetation area was converted to roads and build-up areas. The decrease in water resources has also been a major reason for reducing vegetation areas.

Figure 3. The LULC maps of study area for (1) = 2000 (2) = 2007 (3) = 2014 (4) = 2021.

Figure 4. Area calculation of various LULC classes during 2000 to 2021.

3.2. Change detection

Using Landsat images and output LULC maps (after using the supervised classification method), the observed changes are listed in Table 3. In 2000, the “vegetation area” was occupied the greatest class among all the classes in the study area. The LULC change indicates that the “Vegetation” area reduced from 2000 to 2014 in different regions, for example, decreased “Vegetation” between 2000 and 2021. Table 3 shows that the “bare soil” area has been continually reduced during the last few years. In 2000, 4% of “bare soil” area was reduced to 2.9% in 2021. Cities of Multan, Vehari, and Lodhran are increasing continually; however, also for a large number of people from the rural areas and further small urban areas, leaving a little land available to be called bare soil.

Table 3. LULC changes 2000 to 2021 of the study area.

LULC categories	2000–2007		2007–2014		2014–2021		2000–2021
	Ha	%	Ha	%		Ha	%	Ha
Forest	−3051.43	−0.28	−4213.87	−0.38	−1397.8	−0.13	−8663.2	−0.79
Vegetation	−7354.58	−0.67	−10156.3	−0.93	−18911.3	−1.72	−36422.2	−3.32
Water	−6935.84	−0.63	−9578.06	−0.87	−7845.7	−0.71	−24359.7	−2.22
build-up	22242.95	2.02	30716.45	2.79	27604.3	2.52	80563.7	7.34
bare soil	−4901.06	−0.44	−6768.14	−0.61	550.5	0.05	−11118.7	−1.01

Vegetation areas and water bodies decreased slightly to 3.32% and 2.22% from 2000 to 2021, respectively. Finally, the build-up area increased in coverage from 4.82% in 2000 to 2014 and 7.34% from 2000 to 2021. In 2000, forest area accounted for about 1.7%, which decreased 0.79% from 2000 to 2021. However, bare soil has slightly decreased to 1.01% from 2000 to 2021 (Table 3). Change detection was assisted in improving the recognized factors that are drivers of various changes like social, political, geographical, economic, as well as environmental and their input to the LULC changes.

3.3. Assessing accuracy Assessment

Accuracy assessment is a key component of different LULC classes. Error matrix is a quite widespread and general means to show accuracy assessment. Different statistical processes of thematic accuracy assessment were drawn from the error matrix, for example, % of producers as well as user’s accuracy, overall accuracy, and kappa coefficient (K). Details of producer’s and user’s accuracy for various LULC classes for the years (2000, 2007, 2014, and 2021) are shown in Table 4. The average producer and user accuracy of vegetation area were 88.2% and 82.8%, respectively. Similarly, the average producer and user accuracy of water-bodies were 84.6% and 87.2%, respectively. The build-up area’s average producer and user accuracy were 83.4% and 79%, respectively. Average producer’s and user’s accuracies are 78.6% and 79% for 2000, 84.2% and 83.3% for 2007, 89.9% and 88.4% for 2014, and 82.3% and 82.2% for 2021, respectively. Overall accuracy is 82% for 2000, 84% for 2007, 79% for 2014, and 83% for 2021 (Table 5). Kappa values were attained at 73%, 76%, 68%, and 70% for the years 2000, 2007, 2014, and 2021, respectively. In addition, Usman et al. (2015) showed that the overall accuracy ranged from 78.2% to 82.8% and depended on the satellite data’s spatial analysis. The lower accuracy values in this comparison stem from a minor plot and a combination of pixel sizes due to a mixed cropping pattern during the kharif season. The current study also had a reasonable overall accuracy (82 to 83%) and K value (73 to 70%). These values are compatible with already conducted studies, showing our results’ reliability.

Table 4. Details of producer’s and user’s accuracy for various classes.

LULC Classes		Class and season
		Producer’s accuracy (%)					User’s accuracy (%)
		2000	2007	2014	2021	Avg	2000	2007	2014	2021	Avg
LULC classes	1	76.7	87.1	86.7	86.7	84.3	85.2	90	86.2	83.9	86.3
	2	88	86.4	93.3	85	88.2	79.3	76	93.5	82.6	82.8
	3	78.3	83.3	92	84.7	84.6	87.1	86.2	87.7	87.7	87.2
	4	80	86.7	87.5	79.2	83.4	73.3	78.7	88	76	79
	5	70	77.4	90	75.9	78.3	70	85.7	86.7	80.8	80.8

Where: 1 = Forest, 2 = Vegetation, 3 = Water, 4 = Build-up, and 5 = Bare soil.

Table 5. Summary of (kappa coefficients) as well as average accuracies.

Sr. No.	Year	Avg. Producer’s accuracy	Avg. User’s accuracy	Overall accuracy	K
1	2000	78.6	79	0.82	0.73
2	2007	84.2	83.3	0.84	0.76
3	2014	89.9	88.4	0.79	0.68
4	2021	82.3	82.2	0.83	0.70
Average		83.8	83.2	0.82	0.72

3.4. Normalized Difference Vegetation Index (NDVI)

Landsat images have been used for the assessment of the vegetation measurement on the LULC based on NDVI. The NDVI model derived through Landsat images of 2000, 2007, 2014, and 2021 is shown in Figure 5. The NDVI values in the range 2000 were −0.32 to + 0.77 in the study area, while in 2021, NDVI values changed (minimum −0.27 and maximum + 0.54). Average NDVI values were calculated at 0.23, 0.17, 0.19, and 0.14 for the years 2000, 2007, 2014, and 2021, respectively (Table 6). It was noted that there was a deceased (NDVI values) change in 2021 as compared to 2000 in the study area. Higher NDVI values showed the most productive regions, such as vegetation areas and forests. Conversely, lower values of NDVI showed that there are least and less productive regions, for example, bare soil and build-up areas. So, the NDVI model showed the main changes in the study area’s most productive and productive regions. Generally, the NDVI values are higher in forest and vegetation areas than for bare soil and build-up areas, and therefore, the enlarged forest and vegetation area may contribute to satellite-observed vegetation greenness throughout southern Punjab during the last few years.

Figure 5. The NDVI maps of study area for (1) = 2000 (2) = 2007 (3) = 2014 (4) = 2021.

Table 6. Minimum and maximum values of NDVI and NDMI of the study area.

Years	NDVI Max	NDVI Min	NDVI Avg.	NDMI Max	NDMI Min	NDMI Avg.
2000	0.77	−0.32	0.225	0.78	−0.22	0.28
2007	0.72	−0.38	0.17	0.74	−0.25	0.245
2014	0.68	−0.3	0.19	0.67	−0.27	0.2
2021	0.54	−0.27	0.135	0.62	−0.33	0.145

3.5. Normalized Difference Moisture Index (NDMI)

The NDMI was calculated by using mean yearly precipitation and evapotranspiration (ETo). The NDMI values ranged from − 0.22 to + 0.78 in 2000, and the NDMI value ranged − 0.25 to + 0.74 in 2007 in the study area. In 2014, it decreased from − 0.27 to + 0.67; in 2021, the NDMI value showed the minimum value (− 0.33) and maximum (+0.62). Average NDMI values were calculated at 0.28, 0.25, 0.2, and 0.15 for 2000, 2007, 2014, and 2021, respectively (Table 6). Moisture index values showed a gradual decrease from 2000 to 2021 in Southern Punjab. The NDMI showed that maximum values were noted in vegetation and forest areas. In the study area, loss of NDMI is the main cause of climate change, which strongly affects vegetation areas converted into build-up areas in the study area. The final outcome is attainable with the NDMI values range ranging from −1 to 1, where values between 0 and 1 are the areas with a high amount of greenness regions like forest and vegetation. The values between −1 and 0 are the regions with low greenness areas, like bare soil and build-up areas in the study area (Figure 6).

Figure 6. The NDMI maps of study area for (1) = 2000 (2) = 2007 (3) = 2014 (4) = 2021.

3.6. Relationship between NDVI and NDMI

Statistical linear regression analysis shows that NDVI coupled with NDMI is the best technique for vegetation monitoring when compared to other indices [92, 93]. However, a positive, strong linear relationship between NDMI and NDVI existed, with a correlation coefficient of R2 = 0.92 for 2000, 0.87 for 2007, 0.82 for 2014, and 0.73 for 2021, represented in Figure 7. Regression analysis indicates that in such areas where NDMI values were most significant, NDVI values were also greatest, but where NDMI was less, NDVI was also less. The outcome of the study indicates that lesser NDMI were observed in areas characterized by NDVI negative indicates NDVI and NDMI values at the same time as negative values denote another LULC types.

Figure 7. Relationship between NDVI and NDMI in the study area.

4. Discussion

The relation of LULC data with NDVI and NDMI values offers a comprehensive view of how human activities and natural processes influence vegetation health and moisture content. For example, urban expansion often leads to reduced vegetation cover and increased impervious surfaces, resulting in lower NDVI values and potentially altered hydrological cycles, as indicated by NDMI changes. Similarly, deforestation can be tracked through declining NDVI values and subsequent impacts on moisture availability indicated by NDMI shifts. According to Hussain, Ahmad, et al. (2020), vegetation areas were changed to residential, commercial areas, and roads from 1977 to 2017. By A. Ali et al. (2018), the urban area has increased; however, the rate of increase for the populated region was a bit less during the last few years in the Multan region. This directly found that a rapid increase in urban areas is estimated, which would eventually cause a reduction in forest and vegetation areas in the study area. According to Hussain, Mubeen, et al. (2022), change detection showed the spatial distribution of LULC variations in the Multan region from 1988 to 2017. Change detection was assisted in improving recognized factors that drivers of various changes like social, political, geographical, economic, as well as environmental and their input to the LULC changes. According to our study, various new colonies have been built in the study area, mainly Multan and Vehari districts, which mostly contributed to the expansion of the urban area during the last few years. In the future, the expansion of urban areas is projected to be administered by building different roads, mainly the ones, e.g., Vehari road and M4 motorway.

Waleed et al. (2022) found that one of the greatest effective methods is based on following the temporal variation in vegetation cover, for example, NDVI in Multan and Faisalabad districts of Pakistan. According to Rani et al. (2018) rapid expansion in urban areas is often perceived as an indicator of commercial growth; together, it impacts environmental services and resources. Ahmad (2012) presented that mean NDVI values ranged between 0.05 and 0.25 from 2002 to 2010 in the Cholistan Desert, Pakistan. It showed that vegetated areas have maximum NDVI values, and non-vegetated areas have less NDVI values. Our results also represented that average NDVI values were calculated in the range 0.14 to 0.23 from 2000 to 2021 in the Multan region. Our study shows a net decrease in NDVI during the last few years in the Multan district. The decrease in water resources has also been a major reason for reducing vegetation areas. These effects may consist of a loss of biodiversity, decrease in vegetation area, effect of urban heat islands, emissions of greenhouse gases, and spatial separation of natural environments, soil, light, water, and noise pollution. According to S. Rahman and Mesev (2019), NDMI values near one indicate maximum moisture and water-like forest and vegetation area, and 0 represents less moisture content, like bare soil and build-up areas. The major outcomes of NDMI showed that almost 50% of the area was close to 0, which shows a moisture shortage in the off-season in Southern Punjab.

According to Saha et al. (2018), the soil moisture index ranged between 0 and 1 during 2017 in the district Kanker of Chhattisgarh, India. According to Karan et al. (2016), the moisture index is extremely correlated with the canopy water content of the canopy, and therefore, it can track the changes more closely in plant biomass and water stress than NDVI (Hussain, Raza, et al., 2023; Saleem et al., 2020). According to our study, in such areas where NDMI values were most significant, NDVI values were also greatest, but where NDMI was less, NDVI was also less in the study area. In most areas of Multan regions, NDMI values have closed to zero, which was extremely affected by water shortage (Hussain, Qin, et al., 2022; Tan et al., 2020). Observing NDMI in different regions is a significant task for predicting crop drought and efficient agricultural water management. Researchers and decision-makers gain insights into environmental health and resilience by analyzing the relationships between LULC, NDVI, and NDMI. These indices aid in identifying areas vulnerable to environmental degradation, assisting in land management strategies, and informing policies to mitigate negative impacts on ecosystems and water resources. Integrating these tools into environmental monitoring systems empowers us to make informed choices that promote sustainable development, protect biodiversity, and address the challenges a changing climate poses. LULC changes have far-reaching consequences for ecosystems, biodiversity, hydrology, and climate. Urbanization, deforestation, agriculture expansion, and industrialization are some of the key drivers of LULC changes. Monitoring LULC changes helps us comprehend the impacts of human activities on the environment and make informed decisions for sustainable development.

5. Conclusion

The main objective of this study was to calculate the LULC, NDMI, and NDVI change estimates in Southern Punjab (Multan, Vehari, and Lodhran districts) through RS and GIS tools. It was observed that there had been fast changes in LULC in the study area, particularly in the vegetation area and forest area. Vegetation areas decreased by 0.58% from 2000 to 2014 and 2.7% during 2000 to 2021 as vegetation and forest areas were converted to roads and build-up areas. Average NDVI values were calculated at 0.23, 0.17, 0.19, and 0.14 for the years 2000, 2007, 2014, and 2021, respectively. Similarly, average NDMI was calculated at 0.28, 0.25, 0.2, and 0.15 for the years 2000, 2007, 2014, and 2021, respectively. This research shows that NDVI correlated positively with NDMI. The decrease in water resources has also been one of the main reasons for the reduction of forest and vegetation areas. It was noted that there was a decrease (NDVI and NDMI values) in 2021 compared to 2000. Overall, producers and user’s accuracy was 82% for 2000, 84% for 2007, 79% for 2014, and 83% for 2021. Due to these changes, we lost our natural ecology and biodiversity; increasing build-up of regions may also lead to a lot of ecological problems. Therefore, the government should make a policy to provide sufficient irrigation water in Southern Punjab by managing the canal system and some other suitable measures. This research will help enhance the local government’s capacity in the study area to apply sound strategies at the local and national levels for the development of agriculture. The outcomes of this research will help the decision-makers take any measures or decisions for future development. This information will also help decision-making processes to promote sustainable development, balance human needs with environmental preservation, and mitigate the negative consequences of unchecked LULC changes in the future.

Authors’ contributions

Sajjad hussain, Muhammad Mubeen, Wajid Nasim, and Shankar Karuppannan proposed the main concept and were highly involved in write-up. Ashfaq Ahmad assisted in the data analysis and preparation of the spatial map. Muhammad Amjad¹ and Hafiz Mohkum Hammad are involved in the write-up and review. Sajjad Hussain, Shah Fahad, and Shankar Karuppannan were involved in reviewing, editing, and correcting English grammar.

Availability of data and material

The datasets used and/or analyzed during the current study are available in the article/from the corresponding author on request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.