An Assessment of the Impact of Land Use and Land Cover Change on the Degradation of Ecosystem Service Values in Kathmandu Valley Using Remote Sensing and GIS

Abstract

Land use and land cover (LULC) robustly influence the delivery of the ecosystem services that humans rely on. This study used Kathmandu Valley as a study area which is a fast-growing and most vulnerable city to climate change. Remote sensing and GIS methods are the most significant methods for measuring the impact of LULC on the ecosystem service value (ESV). The satellite-based dataset was used for quantitative assessment of the LULC and ecosystem service value for 10-year intervals from the year 1989 to 2019. The result revealed that the area of forest cover, cropland, and waterbodies decreased by 28.33%, 4.35%, and 91.5%, respectively, whereas human settlement and shrubland increased by more than a hundred times and barren land by 21.14% at the end of the study period. This study found that Kathmandu valley lost 20.60% ESV over 30 years which dropped from USD 122.84 million to USD 97.54 million. The urban growth and extension of agricultural land to forest cover areas were found to be contributing factors for the reduction in ESV of Kathmandu valley. Cropland transformed into shrubland, bringing about an increase in ESV of some areas of the study region. In conclusion, the aggressive increase in population growth with inadequate urban planning and fragmentation of farmlands influenced the ESV of Kathmandu valley.

1. Introduction

Changes in land use and land cover have been accompanied by a perceived rise in the availability of a diverse variety of ecosystem services (ES) for both local and global consumers. Recreation and ecotourism, biodiversity habitat provision, soil conservation, carbon storage, resource availability, and water quality are all examples of ecosystem services [1]. ES is particularly necessary for those who rely on subsistence livelihoods, which comprise about one-fifth of the world’s population [2]. Human survival is entangled with a variety of natural resources and ecosystems, including agriculture, forestry, and water, which provide a wide range of valuable services to human society at all times [2]. Human interference has had a significant impact on land cover types for many years [3,4,5], and this has resulted in significant changes in the amount and capacity of ecosystem services provided [6,7,8]. The large-scale alteration in LULC due to significant human actions and urbanization directly impacts socio-economic development and environmental sustainability [9,10,11,12,13]. As a city grows, it may improve the livelihood and wages of people but also has substantial environmental consequences. On local, regional, and global scales, land conversion for urbanization is one of the most prevalent drivers of ecosystem damage, altering the earth’s landscape and potentially affecting vital ecosystem services in the long term [14]. The transformation of numerous land covers into built-up areas is the specific factor for the alteration of surface temperature, trace gases, aerosols, and changes between the ground surface and the atmosphere [15]. As more than half of the world’s population is already living in cities, and is expected to continue [16], these environmental effects are predicted to worsen in the mid-twenty-first century.

The pattern of converting natural landscapes into urban settlements has changed the confined environment, which is hotter than its surroundings [17,18,19,20]. The major drivers for the conversion of arable land to urban sprawl [21] are rapid population growth, urban expansion, and socio-economic development despite provisional services being incredibly important for the environment and human well-being [22,23,24]. Kathmandu valley’s haphazard and unplanned urbanization has degraded the urban environment, increased urban poverty, and exposed the valley’s rising urban population to multiple hazards such as flooding, drought, air pollution, etc. Despite rapid urbanization in Kathmandu valley, the Government of Nepal (GoN) has implemented limited measures to improve the ecological environment. There are only a few specific projects based on water-related services rather than conserving forest areas and preserving fertile lands. Though the GoN has recognized and integrated ecosystem services into national development plans, there are still major losses in ESV due to a lack of proper planning and monitoring for natural capital management. A deeper understanding of LULC was analyzed with their drivers and effects on ecosystem services, as well as to comprehend landscape patterns and changes. The quantification of ecosystem services into monetary value would provide evidence to decision-makers to understand the potential cost of ecosystem restoration. The remote sensing and Geographical Information System (GIS) tools practiced in measuring land use cover change provide a new way for ecosystem service studies [25,26]. Satellite images have been used as the primary data sources for detecting urban expansion due to long-term data availability at the required spatial and temporal distribution. The study of land cover change and its effects on the environment of urban areas is critical for long-term development and planning to achieve an efficient balance between society, national economics, and the environment [27,28,29]. While it is demonstrated that the measurement of ESV with scientific studies using high-resolution images and GIS tools can reveal some remarkable pieces of information on the degradation of the ecosystem and environment, no such studies have been conducted in recent years for Kathmandu valley. There is a clear need for a detailed spatiotemporal analysis of land cover and its impact on ESV for a rapidly urbanizing area such as Kathmandu valley. The main objective of this study is to (a) analyze the changes in the LULC pattern and its impact on the degradation of the ESV of Kathmandu valley and (b) provide guidance and information for the management of future expansion of urban settlement and sustainable use of land resources.

2. Materials and Methods

2.1. Study Area

Kathmandu valley lies in the central part of Nepal. It covers three districts Bhaktapur, Kathmandu, and Lalitpur. Kathmandu valley lies between latitudes 27°30′ N to 27°50′ and longitudes 85°10′ E to 85°31′ E, with an area covering 933.73 sq km (Figure 1). It is a bowl-shaped area surrounded by a mountain with an elevation ranging between 1900–2800 m above sea level (masl). The average height of Kathmandu valley is 1300 masl Kathmandu valley lies in the sub-tropical region experiencing average temperature and annual humidity of 24 °C [30] and 75%, respectively, due to changes in physiology and climatic conditions [25]. The mean annual rainfall of Kathmandu valley is 1600 mm, with nearly 80% occurring during the monsoon season [31]. It is an important site for a study of the impact of LULC on ecosystem services due to urbanization, air pollution, and environmental degradation. The key problem for the study area’s future clash between economic development and environmental protection as Kathmandu has been named one of the world’s 15 most vulnerable cities by the International Institute for Environment and Development (IIED) [32].

2.2. Data Sources

In this study, 1989 is chosen as the earliest year for the data analysis based on the low image resolution and data availability of satellite images and the quality of the existing reference dataset. Data are available through the year 2019. The geometrically corrected data of Landsat Level 1 products were downloaded from the United States Geographical Survey (USGS) [33]. The study followed 10-year intervals to visualize the difference in LULC dynamics.

Spatial data were acquired at a resolution of 30 m from the Landsat-5 Thematic Mapper (TM) for the years 1989, 1999, and 2009, and Landsat-8 Operational Land Imager (OLI) images were used for the year 2019. More details of the acquired data are shown in

Table 1. Landsat Data Used for the Study.

Landsat	Path/Row	Spatial Resolution	Acquired Year
Landsat 8 OLI/TIRS	141/41	30m	18 December 2019
Landsat 5 TM	141/41	30m	28 December 2009
Landsat 5 TM	141/41	30m	27 October 1999
Landsat 5 TM	141/41	30m	2 December 1989

. The acquired data period was selected for the post-monsoon period (October to December), which mostly has vegetation green to semi-evergreen [34] and free from cloud cover [35]. After downloading data from the USGS source was pre-processed by extracting the study area, radiometric and atmospheric correction.

2.3. Image Pre-Processing and Classification

To avoid data error or tampering and to establish a direct link between data and biophysical processes, satellite images have to be preprocessed. Each Landsat image is an L1T product, and geometric corrections were performed during data preprocessing. This study used atmospheric and radiometric correction procedures in ArcGIS 10.3 for removing atmospheric noises and surface reflectance caused by the earth’s rotation [36]. All of the images were processed for radiometric calibration (digital number (DN) to top-of-atmosphere (TOA)) and atmospheric corrections [37,38]. The atmospheric correction eliminated the effect of atmospheric absorption and scattering, which gives true surface reflectance data [39,40]. As the band composition is very important for LULC classification and every Landsat sensor has a different band, those bands were carefully composited into one layer. The framework of this study is shown in Figure 2.

In this study, the LULC classification uses a supervised classification method. In supervised classification [41], Maximum Likelihood Classification(MLC) is the most widely used [42,43,44] parametric algorithm, and it gives better classification results for quantitative analysis [45]. At first, a number of signatures of each land class were generated for data training. These signatures were grouped manually into predefined land classes using simple random sampling, high-resolution Google Earth, and subject-matter expertise and familiarity with the study area. For RS image classification, the quality and quantity of selection of training samples are largely related to the accuracy of overall classification [46]. In order to get a better classification result, more than 15 samples of each land cover type were selected for training data [47]. Then supervised classification was carried out of all Landsat image using MLC. Based on the degree of similarity and properties of pixels, MLC is categorized into one of the land class groups. In this study, six land cover types were selected based on the land use types and intensity of land use changes.

Table 2. Major Land Use and Land Cover types.

Land Cover Types	Description
Forest Cover	A land dominated by trees, including natural woodlands and community plantations.
Shrubland	Bushes, grasslands, shrub cover, and degraded forests
Barren land	Areas of silt and sand with very little or no vegetation, such as the shores of rivers.
Cropland	Farmlands and cultivable lands, including seasonal croplands
Built-up Area	Residential, commercial, industrial, roads, and construction site
Waterbody	All types of water bodies such as rivers, ponds, and lakes

shows the major land cover types and their description of their characteristics in the study area. The descriptions of land cover types in the table below describe the nature and characteristics of land classes in the study area.

2.4. Accuracy Assessment

The final and most important step during the land cover type classification in the remote sensing data process is accuracy assessment or validation [48]. Accuracy assessment is an important activity due to the possibility of inaccuracy in the Landsat data. It compares the produced data by the user with the real ground truth data. The accuracy of classified land cover classes was measured using an error matrix [49], user accuracy, producer’s accuracy, overall accuracy [48], and Kappa coefficient [10] to verify the precision and error of the images by comparing them to actual ground truth points [50]. The accuracy assessment of land cover data was carried out using the generally accepted stratified random sampling method. Google Earth provides high-resolution satellite photos at no cost, which are critical references for validating the LULC classification [42,51,52,53,54]. In order to understand the accuracy of classified land classes, nearly 900 samples of actual ground truth data of each class were taken randomly and compared to classified land classes. The number of samples of each class sample depended on the area of the land class, varying between 15 to 900. First of all, an accuracy assessment point was created in classified data in GIS which was exported to Google Earth (Google Earth Pro version 7.3). Each random point was verified by comparing it with the ground truth for accuracy assessment. For this study, Google Earth images from December 2009 and 2019 were used as the reference for classified land classes of 2009 and 2019, respectively, and a 1:25,000 scale topographic map as a reference for classified land classes of 1989 and 1999. An error matrix was generated using reference points.

The overall accuracy of the classified image is determined by comparing how each satellite image cell is classified to the definite land cover conditions acquired from the actual land cover. The likelihood of a categorized pixel matching the land cover type of its corresponding real-world place is measured by the user’s accuracy [41,55]. Producer accuracy is a measure of how well real-world land cover types can be classified [56]. The kappa coefficient is a frequently used method for agreement with classified and actual land cover and is calculated using equation 1 shown below for each land cover classification [49]. The kappa coefficient value ranges from-1 to +1; the higher the value, the greater the agreement [57].

$$\mathrm{Kappa} \mathrm{Coefficient} =\frac{{{\displaystyle \sum }}_{i=1}^m{x}_{ii}-{{\displaystyle \sum }}_{i=1}^m{x}_{ii}\left(A_i{C}_i\right)}{n^2-{{\displaystyle \sum }}_{i=1}^m{x}_{ii}\left(A_i{C}_i\right)}$$

(1)

where n is the total number of stratified random samples taken, x_ii gives the number of samples in actual class i, A_i is the total number of actual ground truth samples associated with class i, C_i represents the total number of the classified sample belonging to class i, m is numbers of rows and columns in the error matrix.

2.5. Ecosystem Service Value Assessment

The various direct and indirect methods can be used for estimating the value of ecosystem services. Using a supply and demand analysis of the sum of consumer and producer surplus (or the price times quantity for certain ecological services), Costanza et al. calculated the value per unit area of each ecosystem service for diverse ecosystems [58]. To estimate the global ESV, 17 ecosystem services from 16 biomes were categorized [32,58]. Xie et al. [34] estimated the ESV per unit area of the different land types in the Tibetan Plateau based on Costanza’s global ESV [35]. For the analysis of ESV in the Koshi river basin, Zhao et al. [59] used the modified value calculated by Xie et al. As the Kathmandu valley is in the same river basin with similar climatic and topographical features, this study followed the adjusted price value coefficient of the ecosystem by Xie et al. [50], which is shown in

Table 3. ESV coefficient for different land cover types given by Xie et al [34].

Land Cover Types	ESV (USD/ha/year)
Forest Cover	2168.84
Shrubland	1089.19
Barren land	0.00
Cropland	699.37
Built-up Area	0.00
Waterbodies	6552.97

.

The ecosystem service value is calculated for six different land cover types, which are represented as the forest cover for dense forests, shrubland for light forest cover shrubs and grasslands, cropland for cultivated areas, barren land for dry and unused land, built-up area for urban settlements and water body for water resources. The ESV coefficient for the built-up area and the barren area was considered zero as the urban area is responsible for heat generation and erosion generating large amounts of heat undesirable for the residents and environment and causing loss of habitats [23,60]. The total ecosystem service value of each classified land cover was obtained by multiplying the value coefficient with the area of each land type [59,60].

$$ESV=\sum (A_i\times V{C}_i)$$

(2)

where ESV is the total estimated ecosystem service value of each land type, A_i denotes the area of each land cover type in ha, and VC_i represents the ecosystem value coefficient in USD/ha/year.

3. Results

3.1. Land Use and Land Lover Accuracy Assessment

According to the results of the accuracy assessment for the land cover classification (

Table 4. Confusion matrix of land classification in 2019.

LULC Class	Forest Cover	Shrubland	Barren Land	Cropland	Built-Up Area	Water Bodies	Total
Forest Cover	255	2	0	20	1	0	278
Shrubland	3	67	0	22	1	0	93
Barren land	0	0	13	3	3	0	19
Cropland	0	1	2	312	13	1	329
Built-up Area	0	0	0	1	147	3	151
Waterbody	0	0	0	0	3	13	16
Total	258	70	15	358	168	17	886

), the user’s accuracy and producer’s accuracy were found to be 94.78 and 96.19 for 1989, 95.93 and 94.44 for 1999, 94.44 and 94.72 for 2009 and 84.27 and 88.72 for 2019 respectively. The overall accuracy was 94.86 for 1989, 93.61 in 1999, 95.18 for 2009, and 91.08 for 2019. The kappa coefficients were found within the almost perfect agreement [61] of 0.93, 0.91,0.93,0.87 for 1989, 1999, 2009, and 2019 respectively. The UA and PA for each land class, OA, and kappa coefficient are shown in

Table 5. Assessment Accuracy on LULC Classification for 1989, 1999, 2009, and 2019.

LULC Class	1989		1999		2009		2019
	User’s	Producer’s	User’s	Producer’s	User’s	Producer’s	User’s	Producer’s
Forest Cover	98.00	98.49	96.46	96.95	98.38	98.70	91.73	98.84
Shrubland	86.96	95.24	95.95	98.61	84.62	100.00	72.04	95.71
Barren land	100.00	100.00	100.00	78.95	100.00	96.00	68.42	86.67
Cropland	92.20	93.10	89.45	94.25	93.10	94.08	94.83	87.15
Built-up Area	97.22	90.32	93.7	91.39	96.97	88.89	97.35	87.50
Waterbody	94.29	100.00	100.00	90.91	93.55	90.63	81.25	76.47
Overall Accuracy (%)	94.86		93.61		95.18		91.08
Kappa Coefficient (%)	93.26		91.51		93.31		87.57

All Users and Producers indicate accuracy is in percentage.

.

3.2. Analysis of Land Cover Change in Kathmandu Valley

In this study, Kathmandu valley was divided into six different land cover classes for 1989, 1999, 2009, and 2019. In 1989, the land cover of Kathmandu valley was dominated by forest area at 43.83% of the total area. There was less barren or unused land (0.1%) than in other land cover types. The built-up area covered 9.54% of the total area of Kathmandu valley, which is the last area of built-up during the study period. Cropland, shrubland, and waterbody were 43.67%, 2.3%, and 0.5% of the total land cover area of Kathmandu valley, respectively. In 1999, 2009, and 2019 forest cover shrunk to 43.77%, 40.97%, and 31.41% of the total land cover of Kathmandu valley, respectively. On the contrary, cropland was nearly constant at 41%, 40.97%, and 41.77% in 1999, 2009, and 2019 respectively. The built-up area is responsible for the largest transfer in the land cover by the increasing area of 2148.21 ha, 3,517.11 ha, and 10,014.39 ha in 1999, 2009, and 2019, respectively.

Comparing the land cover of 1989 and 2019, one can see significant variation in the land cover classes with a drastic reduction in the water bodies (91.56%) and cropland (4.35%). There has been a rapid growth in the rate of barren land, built-up area, and shrubland, amounting to 21.14%, 112.45%, and 173.68%, respectively. Urban (built-up) areas have increased exponentially during the study period. This has occurred at the expense of forest cover (decreased by 28.3%) and, to a lesser extent, cropland (decreased by 4.4%). Figure 3 shows the spatial distribution of the losses and gains of the classified land cover types in Kathmandu valley. These expansions are associated with the conversion of cropland into the urban settlement and the reduction in the forest cover for further expansion of cropland. Urban expansion has been accelerated in the nearby area of cities and built-ups along the main roads forming the new settlements Figure 4 and Figure 5 illustrated the transition of land cover changes between the years 1989 and 2019 whereas the status of land cover for the different years is shown in

Table 6. Landcover Status in 1989, 1999, 2009, and 2019.

LULC Class	1989		1999		2009		2019
	ha	%	ha	%	ha	%	ha	%
Forest Cover	40,907.34	43.83	40,854.60	43.77	38,238.57	40.97	29,319.93	31.41
Shrubland	2149.38	2.30	2704.32	2.90	3160.08	3.39	5882.49	6.30
Barren land	112.41	0.12	256.68	0.27	1203.12	1.29	136.17	0.15
Cropland	40,763.97	43.67	38,271.15	41.00	38,242.17	40.97	38,990.61	41.77
Built-up Area	8905.77	9.54	11,053.98	11.84	12,422.88	13.31	18,920.16	20.32
Waterbodies	499.32	0.53	197.46	0.21	71.37	0.08	42.12	0.05
Total	93,338.19	100.00	93,338.19	100.00	93,338.19	100.00	93,338.19	100.00

and

Table 7. Change in Land Cover during the Study Period.

LULC Class	1989–1999		1999–2009		2009–2019		1989–2019
	ha	%	ha	%	ha	%	ha	%
Forest Cover	−52.74	−0.13	−2616.03	−6.40	−8918.64	−23.32	−11587.41	−28.33
Shrubland	554.94	25.82	455.76	16.85	2722.41	86.15	3733.11	173.68
Barren land	144.27	128.34	946.44	368.72	−1066.95	−88.68	23.76	21.14
Cropland	−2492.82	−6.12	−28.98	−0.08	748.44	1.96	−1773.36	−4.35
Built-up Area	2148.21	24.12	1368.9	12.38	6497.28	52.30	10014.39	112.45
Waterbodies	−301.86	−60.45	−126.09	−63.86	−29.25	−40.98	−457.20	−91.56

shows the change in land cover between study periods.

3.3. Assessment of Ecosystem Service Value Changes

In this study, the ESV for the Kathmandu valley was calculated using values derived by Costanza et al. and modified by Xie et al. for the Tibetan Plateau. Based on the ESV coefficient modified by Xie et al. this study estimated the value for 1989, 1999, 2009, and 2019. Over the study period of 30 years, it was discovered that variation in the ecosystem service differs in gain and losses across space and time. It shows there is a significant decrease in the overall ecosystem service value and an increase in the built-up and barren land area, which is the major factor for the reduction in ESV. From the estimation, it was found Kathmandu valley stored USD 122.84 million value of ecosystem services in 1989, an amount that declined with each passing year. By 2019, the total ESV had fallen to USD 97.54 million. The ESV for the classified shrubland, cropland, and water bodies was USD 2.95, USD 26.76, and USD 1.29 million, respectively, for 1999. Between 1999 to 2009, the ESV decreased by 5.03%, which was near USD 4.2 million. The ESV for the categorized land was USD 82.93, USD 3.44, USD 26.74, and USD 0.467 million for forest cover, shrubland, cropland, and waterbodies, respectively. From 2009 to 2019, the ESV value dropped by USD 16 million (14.12%). Among the different LULCs, forest cover comprised higher ecosystem service values as compared to other land cover types due to larger area coverage. The forest cover was measured at USD 63.59 million per year in 2019, cropland at USD 27.27 million per year, waterbodies at USD 0.27 million per year, and shrubland at USD 6.4 million per year, respectively 2019. The dense forest cover is responsible for the ecosystem services to improve habitat quality, soil erosion, and other complements [60]. Though there is less area covered by the water bodies than shrubland, due to the higher value of the ESV coefficient, it contributes significantly higher. However, the total estimated ESV decreased by 2.63%, 7.53%, and 20.60% in 1989–1999, 1989–2009, and 1989–2019 respectively (

Table 8. Ecosystem Service Values for Kathmandu valley in 1989, 1999, 2009, and 2019.

LULC Class	1989	1999	2009	2019
Forest Cover ($)	88,721,475.29	88,607,090.66	82,933,340.16	63,590,236.98
Shrubland ($)	2,341,083.20	2,945,518.30	3,441,927.54	6,407,149.28
Cropland ($)	28,509,097.70	26,765,694.18	26,745,426.43	27,268,862.92
Waterbody ($)	3,272,028.98	1,293,949.46	467,685.47	276,011.10
Total ($)	122,843,685.17	119,612,252.60	113,588,379.60	97,542,260.28

). From 1989 to 2019, the largest decrease in the ecosystem service values is in the forest cover class, with USD 25.13 million (28.33%). During 1989–1999 and 1999–2009, there was a slight reduction in the forest cover, but there was a larger value drop during 2009–2019. The study found cropland slightly decreased during 1989-1999 and 1999–2009 but increased in 2009–2019 (Figure 6 and Table 8).

The variation in cropland is affected by urbanization and deforestation. The rapid rural-urban migration, increase in the densely built-up area, and degradation of forest cover into shrubland is the main reason for decreasing ecosystem value in forest land cover type [62]. The summary of temporal changes in ESV value for the different land cover of the study area is presented in

Table 9. Change in ESV during the study period in USD and percentage.

LULC Class	1989–1999		1999–2009		2009–2019		1989–2019
	$	%	$	%	$	%	$	%
Forest Cover	−114,384.62	−0.13	−5,673,750.51	−6.40	−19,343,103.18	−23.32	−25,131,238.30	−28.33
Shrubland	604,435.10	25.82	496,409.23	16.85	2,965,221.75	86.15	4,066,066.08	173.68
Cropland	−1,743,403.52	−6.12	−20,267.74	−0.08	523,436.48	1.96	−1,240,234.78	−4.35
Waterbody	−1,978,079.52	−60.45	−826,263.99	−63.86	−191,674.37	−40.98	−2,996,017.88	−91.56
Total	−3,231,142.57	−2.63	−6,023,873.00	−5.03	16,046,119.32	−14.12	−25,301,424.89	−20.60

. According to the result of the study, there is no doubt that the conversion of the forest cover into other land cover reduces the total ESV of Kathmandu valley (Figure 7 and Table 9).

4. Discussion

Compared to other estimation methods, LULC analysis is a widely used method for measuring the spatial and temporal variation of the ecosystem service values at a regional and global scale. Kathmandu valley is one of the most rapidly urbanizing cities in South Asia, which has been experiencing rapid land cover changes for the last 30 years. In this study, the ecosystem services values of Kathmandu valley were measured using remote sensing and geographical information. As per the quantitative evidence of this study, Kathmandu valley has witnessed considerable land cover changes since 1989. Previous studies have also found that the Kathmandu valley was dominated by the rapid change in urbanization, deforestation, and variation in the agricultural land area between 1989 to 2019 [59,63,64]. Wang et al. studies on LULC of Kathmandu district found the built-up area is increased by 52.33% [63], which justifies the surges of population density in Kathmandu rather than in Lalitpur and Bhaktapur districts. The rate of change in land cover in Kathmandu valley was found to be higher than in other parts of the country [54,56,63,65]. A study of LULC on a major built-up area of Kathmandu valley provides a detailed expansion of human settlement, increasing by 4.96 % and cropland reduced by 6.51% between the years 2010 to 2018, excluding suburban areas [66]. These changes were more noticeable latter part of the study period (2009-2019). The growth of the built-up area was at the expense of forest cover and, to a lesser extent, cropland.

The result shows built-up areas expanded into nearby cropland, dense forest areas changed into agricultural land, and waterbodies converted into barren land. Some parts of the remaining forest cover are converted into shrubland, degrading its quality and reducing the area for carbon sequestration, similar to the result found by Khanal et al. [64]. The increase in the dense urban settlement can be related to the growing urban area of Kathmandu valley, spreading the settlement outward as well as becoming denser due to conversion from other land covers into built-up land. During the study period, the Kathmandu valley experienced a significant loss of forest cover, with around 11,600 ha converted for the urban (built-up) purpose. Farmers also were driven by the rapid growth of the built-up area to remove the forest and increase agricultural land on the slopes. The cropland area has migrated to the outer area of urban settlement, delineated by the forest covers and shrubland. However, the rapid expansion of built-up regions over the study period may be an increase by rural-urban migration, political unrest, economic centrality, and a boom in real estate business in the Kathmandu valley, as found by Ishtiaque et al. [67] and Khanal et al. [64] which demands sustainable management of urban areas including nature-based solution pathways. Rural-to-urban migration accounts for the majority of this expansion, which is fueled by the economic opportunities accessible in the capital city compared to rural areas [20]. Throughout the 1990s, urban in-migration accounted for up to 40% of population growth [67], and as per the CBS report in the 2000s, the net inflow of rural-urban migrants was 36% [68]. Land abandonment in the hills and conversion of agricultural lands to urban areas could have resulted from the loss of highly fertile agricultural lands and, as a result, a reduction in food production [69]. This might be an issue for a food-security-sensitive country such as Nepal. Barren land was increased slightly in 1999–2009 and then decreased dramatically in 2009-2019 which is a quite common pattern for the conversion of other land types to urban settlements. Government and policy changes in managing land and land-related resources may be responsible for the differences in land cover between the years [70]. Public policies are significant drivers of LULC transformation and play an equal role in encouraging long-term land use [71]. Government policies on land use and management continue to be technical, with little regard for the role of land users, their experiences, knowledge, and adaptability.

The loss of ecosystem services as a result of urbanization is not only a national or regional problem but a worldwide one [72,73,74]. This study estimated the total annual ecosystem service values in Kathmandu valley decreased from USD 122.84 to USD 97.54 million between 1989 and 2019. It was found that ESV dropped by 7.53% between 1989 to 2009 whereas Shrestha & Acharya also found a similar rate of reduction in ESV value in Kathmandu valley [73]. Similarly, the capital city of the neighboring country India observed dropped in ESV by USD 56 ha⁻¹ year⁻¹ [74]. These changes will affect ecosystem architecture, functions, species geographic distributions, and ecological resilience, all affecting urban ecosystem services [73]. Changes in LULC are likely to have similar effects in cities around the world, such as increasing urban heat island effects, flood security hazards, air quality degradation, and difficulty caused by different species [75]. The decline in the forest area cover in the Kathmandu valley has a significant impact on the forest ecosystem services. Due to forest land degradation, the economic loss associated with the dense forest was estimated at USD 25.13 million per year (−28.33%). Ecosystem services are harmed by haphazard urbanization and the loss of forest and agricultural land [32,76].

Global studies conducted around the world found that the significant causes of loss of ecosystem services in urban areas are severe consequences of overpopulation [77,78,79]. Over a thirty-year study period, the Kathmandu valley has experienced major changes in the land cover, which caused the loss of 20.60% of ecosystem service value. From the conversion pattern, it was found that the extension of the urban areas into cropland increases the deforestation rate for expanding agricultural land, reducing the share of forest cover in ecosystem service value. The healthy ecosystem of Kathmandu valley was found to be altered during the study period by providing goods and services through human activities. This study found the water bodies of the study area had dropped from USD 3.27 million in 1989 to USD 0.27 million in 2019. The riverside became a trash dump due to uncontrolled urbanization and rapid urban growth [80]. When comparing the result of LULC for this study with other studies conducted in the different river basins of Nepal, the Koshi River basin was the most stringent [62], followed by the Gandaki River basin, with the least change in Karnali River basin [32]. Sharma et al. concluded that ESV is decreasing trend in the Terai Arc Landscape in lowland Nepal [26]. One of the significant observations of the LULC effect on the ESV of Kathmandu valley is also an increase in the land surface temperature creating an urban heat island. Recreational ecosystem services, such as those supplied by urban nature, are a significant aspect of a high-quality living environment and are beneficial to public health [76]. Urban vegetation has the potential to help with carbon sequestration and, consequently, climate change mitigation. The findings of this research can be used as a theoretical foundation for environmental policy formation in Kathmandu valley and implementation based on the studied area’s features. The government of Nepal has adopted to follow the environmental protection act 2019 [81] and the National Climate Change Policy 2019 [82]. This action envisions future ecological priority and green development resulting in a clean and livable environment. In Kathmandu valley, this requires water resource regulation, water quality assurance, and environmental protection [83]. Ecological security programs in the study area could be used to accomplish the dual effect of environmental protection and poverty reduction as applied in Xiang city of China [84,85]. Climate change and biological diversity are the most serious risks to ecosystem services protection. Land use planning is a solution for integrating urban community structures in cities and urban regions to reduce CO₂ emissions, air pollution [86], and solid waste management [87]. The difficulty is to build well-managed settlements preserving environments and critical ecosystem services including recreational services, stormwater absorption, and carbon sinks [88]. Effective planning can mitigate the adverse effects of urban growth and increase ecological services.

Though this methodology is easy to adopt and cost-effective, there are a few limitations. Firstly, there may be uncertainty due to the quality of satellite images. The image selection from a suitable season is significant. In this study, Landsat image was acquired only post-monsoon season with minimum cloud cover to increase the accuracy between the databases. Utilizing high-resolution satellite data can improve classification outcomes and ESV estimation over a large area. Secondly, the ESV coefficient can affect the precision of input data. This study uses the local scale value coefficient derived by Xie. et al. [34] in the Tibetan plateau, which provides the actual value of ecosystem services.

5. Conclusions

This study investigated the trend of the LULC and variation of ESV in Kathmandu valley using remote sensing data for thirty years, from 1989 to 2019. According to the above analysis, the total value of ecosystem services was USD 97.54 million per year in 2019, where the most valuable land cover type was found to be forest area covers, having USD 63.59 million per year ecosystem service value. As for individual land cover types, all land cover varied from 1989 to 2019 and affected the total ecosystem service value of Kathmandu valley. The land cover transitions show that built-up areas and water bodies account for most of the loss in ESV. The increase in the ESV only occurs in the shrubland land-use class, and the degradation of dense forest cover in shrubland is the main reason for the increment in ESV in the shrubland land class. LULC change is being driven by a combination of factors, including growing urban populations and their livelihoods, unplanned urban settlement, transportation congestion, air pollution, unmanaged solid waste disposal, and global climate change. The global price structure would be substantially different if ecosystem services were truly compensated for in terms of their valuable contribution to the global economy. The cost of goods that rely on ecosystem services, whether directly or indirectly, would be significantly higher. If the value of ecosystem services were properly accounted for world’s gross national product would be considerably different in terms of both volume and composition. However, the findings of this study suggest that the current value of ecosystem services is, at best, a static snapshot of a biosphere that is a complex and dynamic system. Nevertheless, this study has provided new insight into variation in ESV in the region over the past 30 years of the study period. The results can be used by policymakers for urban planning, conservation of natural ecosystems, climate change mitigation and adaptation plans, and maintenance of biodiversity conservation. This study recommends integrating nature-based solutions in urban development plans, policies, and financial support for implementing smart interventions. The findings also suggest that policymakers should take into account the regional heterogeneity of ES supply and the gradient analysis for a more accurate definition of ES supply. An effective decision and plan can be prepared to deal with the growth of urban settlements, the depletion of forest cover, the reduction in open space, the variation of farm spaces, and the reduction in small to medium size water bodies. Some recommended plans are green roof space, rainwater harvesting, sufficient use of clean and green energy, and plantation in available spaces at large scales with the active participation of communities and coordination with governmental bodies to enhance the ecosystem services by increasing LULC dynamics. This study clearly states the importance of remote sensing and satellite images in quantifying land cover changes and ecosystem conservation. The result of the study is useful in land use, and land cover model analysis tests alternate approaches for determining how they affect the ecosystem. ESV calculation is a conclusive and suitable method for valuing the ecosystem in terms of money, giving the scientific foundation for directing the policies. It can be used to compare the accuracy and classification of the land using various techniques and models. The most critically affected ecosystem service function in Kathmandu valley provides a case study for research. Additionally, by creating future scenarios that take into account the urbanization pattern and demographic expansion in the landscape and evaluate their effects on ESV, the findings could be expanded. Since this study compared data from several other ecosystems, the information it contains will be crucial for Nepal’s future research and policy development.