Effect of landscape pattern changes and environmental indices on land surface temperature in a fragile ecosystem in southeastern Iran

Climate change and urbanization along with uncontrolled development in less developed countries have led to an increased ecosystems’ thermal environment. Some factors such as environmental indices and landscape pattern changes can alter Land Surface Temperature (LST). Thus, the accurate evaluation of the relationship between these factors and LST is considered important for managing ecosystems, especially fragile ones under high stress. The southeast of Iran has witnessed many destructions in the environmental dimension in the past years. Moreover, this region has a low socio-economic situation, which increases the need to study in this region. In the present study, we used Landsat TM5 satellite images (1989), Landsat 8 OLI/TIRS ones (2019), and Google Earth Engine (GEE) system to prepare the maps of temporal-spatial LST changes, Land Use/Land Cover (LULC), and selected environmental indices including Normalized Difference Vegetation (NDVI), Built-up (NDBI), Water Indices (NDWI), Land Surface Moisture (LSM) and albedo. Then, the correlation levels of LST with the aforementioned indices were assessed by using Geographically Weighted Regression (GWR), as well as assessing LST variation following LULC change. In addition, the Moran index was used to analyze global and local spatial autocorrelation. The results represented an 8.67-degree increase in the mean LST during 1989–2019. Urban and built-up areas had a significant effect on increasing the temperature of the region. Additionally, water bodies and vegetation cover in the region were the most crucial parameters in LST reduction. All of the applied indices were strongly related to LST (>0.70), while some exhibited more correlation in each year. Further, the highest correlation of LST was observed with LSM and NDBI in 1989, as well as with NDVI and NDWI during 2019. In addition, the Moran index value reduced from 1989 to 2019 (from 0.93 to 0.89). Finally, the region rehabilitation based on sustainable development principles played an important role in the direct and indirect decrease in LST.


Introduction
Human activities are considered as one of the most important drivers of ecosystem degradation, which disrupt critical ecosystem functions. The changes in landscape composition (ratio, size, and type of each Land Use/Land Cover (LULC)) and configuration (distribution and spatial characteristics of landscape points) reduce the capacity of ecosystem to provide goods and services (Yohannes et al. 2021).
Along with global warming, human activities such as urbanization, unmanaged agricultural practices, and industrial projects have degraded habitat and disturbed ecological processes. In addition, these activities have led to an ever-increasing rise in the earth's temperature. This elevation has become one of the most important challenges ahead in the 21 st century (Guo et al. 2020). Further, Land Surface Temperature (LST) is one of the crucial environmental parameters, which is extracted from ground measurements and satellite data related to thermal infrared waves. Therefore, it is used as an important variable to assess various factors such as different evapotranspiration, thermal inertia, surface albedo, daily temperature change, and urban thermal island effect at global, regional, and local levels (Govil et al. 2019). The LST is affected by and has an effect on physical, chemical, and biological processes (dos Santos et al. 2017).
The LULC variations influence natural ecosystems worldwide. Wetlands, one of the most important natural ecosystems, play a key role in supplying water to humans and providing ecosystem goods and services in various regions in spite of covering only about 4% of the land surface (2-6% depending on the definition) (Cai et al. 2016;Ciężkowski et al. 2020). Furthermore, the types of LULC have a significant role in the LST changes due to their solar reflectance (albedo), thermal conductivity, light spectrum, surface roughness, and moisture (Chen and Zhang 2017;Reisi et al. 2019;Tan et al. 2020).
Remote sensing and satellite image-based analysis allow the simultaneous observation of temporal and spatial changes in land surface, and its continuous and repeated cover, as well as high resolution and cost savings in the studies. During the recent years, many studies have focused on analyzing LST and evaluated the effect of diverse factors such as LULC (Govil et al. 2019;Govind and Ramesh 2020;Macarof and Statescu 2017;Masoud et al. 2019;Mushore et al. 2017;Nadizadeh Shorabeh et al. 2020;Phan et al. 2018;Tan et al. 2020), as well as different environmental components like Normalized Difference Vegetation (NDVI), Built-up (NDBI), and Water Indices (NDWI), and Land Surface Moisture (LSM) on the LST. In this regard, satellite images such as Landsat ones (Masoud et al. 2019;Mushore et al. 2017) were used in monthly (Phan et al. 2018), seasonal (Macarof and Statescu 2017; Govil et al. 2019), yearly (Jamei et al. 2019), and long-term (Nadizadeh Shorabeh et al. 2020) periods. In general, green vegetation cover and water bodies have low LST, while high LST is observed in build-up zones, bare rock, or dry soil (rock out crop and/or bare land) (Guha et al. 2020;Li et al. 2017).
In a more detailed look, we review some studies in this context. For instance, Lu et al. (2022) investigated the spatial pattern of LST in built-up areas. They indicated that the urban centrality is related to the LST. Furthermore, they founded out that green space can reduce the LST in an urban area. Therefore, it is better to plan vegetation cover at the edge of an urban area to increase the energy exchange surface. Das et al. (2022) studied the spatiotemporal of LULC pattern and its impact on LST in eastern India. The results of their research stated that LST in the mentioned area is grown dramatically over time that is due to urbanization and losing green spaces. Wu et al. (2021) investigated how environmental factors affect the spatial and temporal heterogeneity of the LST. Therefore, the researchers found that the LST in the mentioned study is related to the environmental indices and landscape patterns. In addition, they indicated that the impact of water bodies and vegetation covers in hot seasons is more than other land uses. Jamei et al. (2019) analyzed the spatial variations of surface urban heat islands and their relationship with vegetation and built-up areas in Melbourne. They stated that the LST has changed in different parts of the city with various landscape patterns during different years. As seen above, most of the studies worldwide issued in the intended context are carried out in humid regions or urban areas. Therefore, there are not enough studies in arid regions with fragile and vulnerable habitats, especially in low socioeconomic regions with fast unsustainable development.
Iran with an area of 1.648 million km 2 is located in the center of the Middle East in the southwest Asia, most regions of which have an arid and semi-arid climate. This country has faced several environmental problems like flood, drought, and soil salinity, as well as declining the number of lakes, wetlands, and rivers since the recent decades. The Hamoun wetlands, a group of endangered international wetlands in the southeastern Iran and western Afghanistan, are at risk of disappearing. These wetlands have witnessed widespread drought due to reduced rainfall, climate change, and water mismanagement in the Sistan region during the recent decades, especially since 1997. Additionally, the mean temperature and precipitation increase by 2.6 °C and decrease by 35% over the future decades (Emadodin et al. 2019;Vaghefi et al. 2019). However, the eastern and southeastern areas of Iran, which are in a low position economically and socially, have been more damaged in terms of environment and sustainable development compared to others (Amoushahi et al. 2022).
In this regard, factors such as the drying up of Hamoun and Jazmurian international wetlands caused by the water-political problems between Iran and Afghanistan, as well as the inadequate and improper management of water resources, and long severe periods of drought have led to problems. These consequences include more environmental crises, intensified dust storms, sand dune movement and expansion, farm salinization, less product yield, and migration in the Sistan and Baluchistan province, Iran (Eskandari Damaneh et al. 2018;Mahmoudi and Rigi Chahi 2019;Shakeryari et al. 2016). It is worth noting that Sistan and Balochistan province is considered as the main source of dust in southwest Asia due to the existence of numerous dust sources like the Hamoun wetlands which completely dry up in the summer due to LULC over time (Rashki et al. 2012).
Our study area, located in the Sistan plain, is one of the zones having an international border with Afghanistan. This region is fragile with respect to climate and environmental conditions, as well as important in terms of natural and human ecology due to its location in the end of Helmand catchment and its proximity to Hamoun international wetlands (Shahriar et al. 2018). The drying up of wetlands and rivers, as well as excessive evaporation, less precipitation, and successive droughts has led to the salinization of water and soil, which has become a serious problem in this region (Hosseini and Delbari 2015).
Various studies have recently highlighted climatic conditions, vegetation cover, drought, and desertification in the southeastern Iran. In some of the above-mentioned studies, remote sensing technology was employed to assess the effect of drought and climate on vegetation indices ) and LULC changes Maleki et al. 2019). However, no research has analyzed the effective parameters on LST such as LULC, and diverse environmental indices like vegetation cover, water bodies, built-up areas, soil moisture, and albedo. Thus, due to lack of enough information and research in this ecological and anthropogenic sensitive region, this study aimed to (1) evaluate the annual changes in LULC during 1989 and 2019 in the region, (2) examine the alteration in the selected environmental indices playing a determinant role in the LST of the zone, and (3) unfold the spatiotemporal variations of the LST in the region due to anthropogenic and natural changes. The study utilized Google Earth Engine (GEE) system and remote sensing technology. The region was selected as a representative of the arid and fragile climate of the southeastern Iran, and the two intended years reflected the years before and after the severe drought and widespread land-use change in the region. The other objectives included determining the correlation level of LST with the aforementioned indices through using Geographically Weighted Regression (GWR), as well as assessing LST variation following LULC change.
The results of this study can be valuable for better environmental and socioeconomic planning in this region, grown under human pressures without considering the principles of sustainable development.

Area under study
The region under study is located between 61˚ 25′ 35"-61˚ 83′ 84" E longitudes and 30˚ 59′ 15"-31˚ 35′ 61" N latitudes in the easternmost border region of Iran with Afghanistan in the Sistan and Balochistan province (Fig. 1). This zone is a part of Sistan plain, the area of which is equal to 296514 km 2 . In addition, the minimum and maximum elevation of the region is 426 and 522 m, respectively (Rashki et al. 2012). Based on the De Martonne classification, the zone has arid climate (Maleki et al. 2019) and high mean annual evaporation (approximately 4800 mm) , as well as the mean annual precipitation of 50-55 mm ).

Data sources and preprocessing
In this study, Landsat TM5 satellite images (1989) and Landsat 8 OLI/TIRS ones (2019) were extracted by using GEE system to prepare LST and LULC maps (Table 1). In addition, this system (Shiflett et al. 2017) was applied to calculate the indices of land surface cover such as NDVI, NDWI, NDBI, LSM, and albedo.
The correlation between the intended parameters with LST was analyzed in QGIS software through employing the GWR. Further, the relationship between land use map and LST was obtained. Figure 2 displays a flowchart of stages in the present study.

Selection of appropriate environmental indices
Primary indices were selected according to the previous studies, some of which were eliminated by considering factors like the conditions of the intended area to reduce the number of the indices and simplify process. For example, most parts of the area are plain and possess low elevation. Accordingly, no significant difference was found between their aspect, slope, and elevation, leading to the removal of these variables. To make better decision on the primary parameters, the correlation between different indices was obtained through using GWR based on the recommendation of the various studies (Alibakhshi et al. 2020;Kashki et al. 2021;Xu et al. 2021;Liu et al. 2022) (Table S1). Then, the factors with high correlation were eliminated so that NDMI was removed due to its high correlation with NDWI and NDVI. Finally, NDVI, NDWI, NDBI, LSM, and albedo were selected to evaluate the relationship between LST and environmental parameters.

LST calculation
The mean annual LST (MALST) was computed through converting spectral radiance, recovering brightness temperature, and calculating ratio vegetation index and surface emissivity (Eq. (6)).
The digital values are converted to spectral radiance based on the spectrum radiation reference as follows (Eq. (1)).
Where L indicates spectral radiance and Gain demonstrates band specific multiplicative rescaling factor. Additionally, DN reflects digital numbers for each pixel and Bias reveals add-on rescaling factor (Yue et al. 2019).
where NDVI min and NDVI max are respectively considered as the least NDVI in bare soil pixel and highest value in the vegetation-containing pixel.
Equation (4) is applied for the conditional estimation of surface emissivity.
in this respect, C demonstrates surface roughness factor, and E v and E s are the emissivity of vegetation cover and soil, respectively (Artis and Carnahan 1982).
Equation (5) reveals the conditions for emissivity determination by using NDVI (Griend and Owe 1993).
where LST i and NLST i refer to LST and its normalized value in the pixel i, respectively. The subscripts of min and max represent the minimum and maximum LST in each image, respectively. To examine the trend of LST changes in the period under study, normalized land surface temperature was classified according to the mean and standard deviation. As outlined in Table 2, LST mean and LST STD are the mean and standard deviation of LST in a normalized land surface temperature image (Firozjaei et al. 2018).

Environmental index calculation
The five environmental indices influencing LST were utilized in this study.
NDVI This index reflects the trend of vegetation cover variations. Given the dimensionlessness of this index, its values range -1 to +1. Practically, the amounts below 0.1 are related to land without water, while the higher levels are associated with agricultural activities and forest regions (Eq. 8). where red and near-infrared bands are denoted as R and NIR, respectively (Rouse et al. 1974).
NDWI It can be applied as a strong normalized difference spectral index to specify the correlation with LST, which is determined by green and near-infrared (NIR) bands. As shown in Table 1, the second and fourth bands are respectively considered as green and NIR bands in both TM and ETM, while bands 3 and 5 are respectively used in this regard for OLI/TIRS data. This parameter varies between −1 and 1 so that the negative values reveal built-up and bare lands without water, while the positive NDWI indicates water bodies and vegetation cover (Eq. LSM This variable is among the most important environmental indices. Soil moisture at 1-2 m above land surface has been widely identified as a key factor in the numerous environmental studies such as those in the field of meteorology, hydrology, agriculture, and climate changes. Thus, it should be estimated in the climate, hydrological, and agricultural research, the monitoring of which acts a crucial role in supervising and forecasting flood, drought, and other climatic phenomena (Hu and Xu 2018;Sun and Pinker 2004).
NDBI It is one of the indices for obtaining human land use intensity, human effect, and built-up uses (Eq. (12)) (Hu and Xu 2018), in which near-and middle-infrared bands are demonstrated as NIR and SWIR1, respectively.
Albedo Surface albedo is defined as the ratio of the reflected light to the total energy incident on a surface in a hemisphere space. It plays a fundamental role in global climate changes, which influences land energy level. The variations in the physicochemical properties and spatial structure of local objects in turn alter surface albedo (Alibakhshi et al. 2020). Currently, this index is extensively applied in modeling for radiation-energy balance, numerical weather prediction, atmospheric circulation, and land surface processes (Eq. (13)) (Bonafoni et al. 2017;Trlica et al. 2017). Regarding the factor, blue, green, red, and near-, middle-, and far-infrared bands are respectively denoted as Blue, Green, Red, NIR, SWIR1, and SWIR2.

Classification of LULC map
Yousefi et al. (2015) reported non-parametric Support Vector Machine (SVM) method as the best approach to produce land use map in the arid regions. This method was introduced by Vapnik and Chervonenkis (1971) as a linear classifier for the first time. In the present study, it was used to prepare the LULC map related to the lands of the region in the GEE system in the five categories of irrigated and rainfed agriculture, bare land, water body, and built-up areas for 1989 and 2019. The accuracy of land use classification in the region under study was evaluated by using Overall Accuracy (OA) and Kappa Coefficient (KC) as assessment criteria based on the widely applicable approach of error matrix calculation. The mathematical model of OA and KC can be expressed as follows (Eqs. (14) and (15)).
where q reflects the number of classes, n refers to the total number of intended pixels, and ∑ nii indicates the sum of main diagonal elements in error matrix. Further, n+ reveals the margin of rows, and n+i demonstrates the marginal sum of the columns in the matrix (Rousta et al. 2018

Correlation between LST and environmental indices
During the recent years, an extensive range of various regression techniques has been utilized to predict and model environmental indices. Ordinary Least Squares (OLS) and GWR are the two most used technics by researchers in recent years (Taghadosi et al. 2019). Some researchers have compared these methods to detect the relationship between LST and LULC indices such as bare soil index (SI), index-based built-up index (IBI), NDBI, NDVI, NDWI, and albedo. Their results revealed that the GWR is more accurate and efficient for modeling spatial correlation and spatial heterogeneity in the interaction between the dependent and the independent factors (Hu and Xu 2018;Kashki et al., 2021;Mirchooli et al. 2020).The GWR, proposed by Brunsdon et al. (2010), is known as a potent method for modeling the location-dependent data at local level. Furthermore, this approach locally acts in the region, in which each observation point is weighted by its distance from reference point and the coefficients are not considered in the different fixed or same locations (Wheeler and Páez 2010). Therefore, considering its consistency with the realities of the region, (Kashki et al. 2021), GWR method was applied to assess the relationship between LST and intended indices. The regression is determined by using Eq. 16.
where n represents the number of independent variables, and y and x i illustrate dependent and independent parameters, respectively. Additionally, 0 , 1 , and denote intercept, coefficient, and error, respectively. Equation (17) can be used for the local estimation of this regression.
in which u i , v i specifies the location of data i, 0 u j , v j is intercept, and y i demonstrates independent variable. The j u i , v i indicates the value of parameter j and i is random in the i th location.
Further, the data weight is obtained according to the distance from the location i. The parameter estimation matrix for i is as Eq. (18). where W u i , v i represents a spatial weight matrix, and X and Y reflect dependent and independent variables, respectively.
Gaussian function is applied to compute weighted operator based on Eq. (19), in which W ij demonstrates the data weight observed in the location j for determining dependent parameter in the i th location, and h is considered as bandwidth. In this respect, the weight of location j declines by distancing from location i (Mirchooli et al. 2020).
Furthermore, the validity or efficiency of multivariate regression models was compared by using the coefficient of determination (R 2 ), as well as Akaike information criterion (AIC). The R 2 indicates the percentage of the dependent variable variance explained by an independent one. The numerical amount of this coefficient ranges between 0 and 1 so that zero reflects that an independent parameter has no role in estimating the dependent variable, while one reveals the estimation of 100% of dependent parameter variance by the independent one. The AIC is used to measure the relative efficiency of model, balancing the accuracy and complexity of the model. A small amount of this criterion exhibits the closeness of the level estimated by the model to the observation value or ground truth (Fotheringham et al. 2002).

Spatial autocorrelation
Among spatial autocorrelation coefficients, Moran Index is a widely used coefficient to examine significant correlation between an individual observation and its surrounding space (Fan and Wang 2020;Lee and Li 2017;Lu et al. 2022). This index was introduced by Moran in 1948, which is divided into global and local spatial autocorrelation. The global Moran's I (Eq. (20)) which captures the spatial pattern according to the location of phenomena can denote whether geographic features have clustered, random, or dispersed pattern. Another one, local Moran's I, (Eq. (21)) is utilized for presenting the Local Indicator of Spatial Analysis (LISA). This index uses Geoda software to assess the spatial autocorrelation of LST in the present study.
where x i indicates the amount of feature in location i and n is the total number of grades in the area under study. In addition, W ij denotes matrix weight (i= 1, 2, 3,… and j= 1, 2, 3,…), which equals zero if i and j are adjacent. Moran's I is equal to +1 and -1 in the cases of positive and negative spatial autocorrelation, respectively. However, zero shows the lack of spatial autocorrelation.
LISA can reflect the continuity of local spatial relations by analyzing Moran's I value in each spatial unit (Anselin 2010).
Further, local Moran's I is applied to identify hot and cold spots based on comparing with neighbor samples, the negative levels of which refer to data dispersion. The positive local Moran's I exhibits spatial autocorrelation and clustered pattern in the data distribution. In this respect, high-high and low-low clusters demonstrate a high and a low value in a neighborhood of high (hot spots) and low values (cold points), respectively. Furthermore, high-low clusters represent the non-clustered distribution of data so that a high value is surrounded by low values, while low-high ones are related to a low value surrounded by high value feature (Wang et al. 2020). Figure 3 and Table 3 present the temporal-spatial distribution of LST in the region, as well as the area and percentage related to each of the five LST classes during 1989 and 2019, respectively. The percentage of alteration in various land uses over the intended periods is provided in Fig. 4. Based on the results, LST was in the range of 16.16-39.03 °C in 1989, which reached 23.24-50.02 °C with the minimum and maximum rise of 7.08  (Fig. 3). As shown in Fig. 4 and  were mainly detected around Chah Nimehs, especially the Chah Nimeh Four which was not in 1989. Figure 5 displays the environmental indices under study. As depicted, a rise is observed in the area of agricultural lands, and consequently that of vegetation cover in the region based on the results of NDVI. During 1989, NBDI level is more related to the existence of bare lands, while cities play a greater role in NDBI calculation in 2019 because of replacing a part of bare lands with cities and residential zones. Further, NDWI is low in most areas with the land use of bare land, agricultural, and built-up regions during both years. The results of LSM determination reveal higher moisture in 2019 than 1989. Among the land uses, the zones with rainfed agriculture exhibit more albedo amount, while this parameter is minimized in the water bodies during the periods.

Classification of LULC map
The results introduced rain fed agriculture, irrigated one, bare land, water body, and built-up as the maximum percentage of land uses in the region. As shown in Table 4, KC and OA are acceptable for both years, reflecting the high accuracy of land use classification.
In this study, the highest change was equal to 309.68% in built-up land use during the periods (Table 5), which can be primarily attributed to a higher area of built-up regions because of expanding Zabol and Zahak cities in 2019 compared to 1989. In addition, rainfed and irrigated agriculture increased by 7.80% and reduced by 8.32%, respectively. In terms of water land use, an insignificant decline (−5.17%) was found in 2019 in comparison with 1989 in spite of drying up the large portions of natural water resources in the region during 1989-2019 (4502 ha). Of course, most of their drying up was compensated by constructing Chah Nimeh Four (area of 7538.22 ha and capacity of 800 million m 3 ). The Chah Nimeh Four, possesses the greatest area among the other Chah Nimehs (a total area of 4506.39 ha with a capacity of 700 million m 3 ), was added to the water resources of the zone after construction. The construction of this Chah Nimeh resulted in improving the area of irrigated agricultural lands in the north of its location.
Further, the area of bare lands diminished by 8.16% over the two years. The conversion of bare lands to agricultural ones in the northern parts of the Chah Nimeh Four, which supplies a large portion of irrigated agriculture in this region, indicated the land use alteration. During 2019, rainfed agricultural zone slightly enhanced, revealing the variation in the type of cultivation. Figure 7 illustrates the percentage of various land use changes during the intended periods.

Correlation between LST with environmental indices based on the GWR model
The results of analyzing LST spatial variations with environmental indices (Table 6) represented the maximum correlation of LST with LSM (R 2 = 0.87; Adj-R 2 = 0.83)

Relationship between LST and LULC
The mean LST was maximized as 33. 58, 30.65, 30.07, 26.67, and 22.75 °C for bare land, built-up, irrigated agriculture, rainfed agriculture, and water land uses in 1989, respectively. Bare land land use experienced the highest mean LST (42.49 °C) in 2019, and built-up category with above 8-degree increase was ranked as the second (39.20 °C) (Fig. 8).

Spatial autocorrelation analysis
Local spatial autocorrelation analysis maps were classified into high-high, low-low, low-high, high-low, and   not significant categories. Based on the results of both years, the highest area of high-high class was related to the southern regions of study area and a large part of its eastern zones, which were mostly covered by bare lands (Fig. 9). During 1989, the low-low areas were more observed in the portions of the north, center, and east, which decreased in the northern and northwestern. Furthermore, wider high-high areas were found in the parts in 2019. The results of Moran's I are depicted in Fig. 9.
In general, Moran's I close to +1 indicates the spatial autocorrelation and clustered pattern of data, while the data are fragmented and dispersed if the index is near −1. The results suggested the high cluster pattern of LST with the value of 0.93 in 1989, which continued in 2019 with a lower value (0.89). Finally, the high temperature in 2019 was more dispersed than 1989.

Correlation between LST and environmental indices (NDVI, NDBI, NDWI, LSM, and albedo)
The previous studies have revealed an alteration in LST following the variation in environmental indices. An inverse relationship is always observed between LST with NDVI, NDWI, LSM, and albedo (except aquatic environments), while LST is directly related to NDBI (Traore et al. 2021). The mean LST is maximized in bare lands, and minimized in water bodies and dense vegetation covers (Xu et al. 2013;Chen and Zhang 2017;Pal and Ziaul 2017;Phan et al. 2018). Regarding the present study, a higher LST was obtained in the regions with large NDBI such as urban areas and bare lands. The presence of asphalt and high-absorption materials leads to an enhancement in solar energy absorption, and consequently greater LST in urban and built-up zones. Human activities like the heat emitted from homes, vehicles, and industries elevate temperature in urban regions. The results are consistent with those of some other studies (Benz et al., 2015;Balew and Korme 2020;Chaka and Oda 2021). Further, a rise in water and vegetation cover reduced LST, and vice versa. This issue can be ascribed to the higher moisture caused by improving water resources and vegetation cover. Therefore, LST declines by promoting LSM, NDWI, and NDVI. These results are confirmed by the results of some studies (Chen and Zhang 2017;Traore et al. 2021). Furthermore, albedo was inversely related to LST so that LST diminished by increasing albedo in land. Of course, this issue was not true for water bodies due to low albedo in the areas. The results are in line with those of other studies (Trlica et al. 2017;Saher et al. 2021). In the present study, all of the used parameters exhibited high correlation with LST (>0.70), while some were more correlated in each year. The results reflected the highest relationship of LST was related to LSM and NDBI in 1989, as well as NDVI and NDWI in 2019. Accordingly, the effectiveness of each index on LST determination slightly changes by varying the environmental conditions of the region, which is in line with those of other studies (Sekertekin and Zadbagher 2021;Traore et al. 2021). Moreover, the results of the applying GWR method showed that this approach is accurate enough in giving insight into the spatial relationship between LST and intended indices. This results is in agreement with some previous studies such as Hu and Xu (2019), Kashki et al. (2021), andMirchooli et al. (2020). The results of Moran I analysis in different parts of the study area are meaningful based on ground realities and land uses. Zhang et al. (2021) stated this fact in their research too.

Relationship between LST and LULC
As already mentioned, LST was in the range of 16. 16-39.03 °C in 1989, and 23.24-50.02 °C during 2019. In addition, the mean levels of LST were respectively equal to 30.99 and 39.66 °C in 1989 and 2019, reflecting 8.67 °C elevation. Given a rise in the population of the intended area during the recent years, and consequently a significant growth in the built-up zones, urban and built-up regions had a strong effect on enhancing the temperature of this region. Further, water bodies and vegetation cover were the most important factor in decreasing LST. Similar results have been reported in other studies (Mu et al. 2019;Traore et al. 2021;Wu et al. 2021;Yang et al. 2021;Yuan et al. 2022).
The results are related to Chah-nimeh 1 four, constructed in 2009 and operated, representing its significant effect on agriculture, LST, and salinity in the region. The mean LST in the region declined by more than 7 °C on average 1 Chah-nimeh is four large natural wells in the Sistan plain, to which the excess water of the Helmand River is directed by a channel after constructing Chah-nimeh Four (32.30 °C in 1989 to 25 °C in 2019). Along with affecting LST directly, water level in the region highly influenced the amount of the agricultural land use which is the only land use with vegetation cover in the region. Thus, natural and nonnatural water resources can be considered as a crucial parameter in diminishing LST in the area, leading to more balance in environmental conditions. The results are consistent with those of Traore et al. (2021) , and Chen and Zhang (2017). Based on the results of the present study, LULC was significantly correlated to LST changes, which is in line with those of some other studies (e.g., Edan et al. 2021;Gohain et al. 2021;Kafy et al. 2021). LULC can occur in the different parts of the world over time, create many variations in ecosystems, and damage to the environment and residents of the region. Furthermore, the incompatibility of LULC with sustainable development can strongly affect LST, and subsequently cause ecosystem changes. The alteration and destruction, especially in fragile ecosystems, can be accompanied by more effects. During 1989-2019, construction land use was rapidly expanded due to the overgrowth of population. The result led to excessive natural resource consumption and an increase in impermeable areas with high energy absorption and less energy reflectance. Therefore, a rise in this land use promoted LST. Additionally, water resources play a key role in the area under study because of locating in arid region with low water. Due to climate changes in the recent years, excessive natural resource consumption, and internal management problems, as well as constructing a dam on the Helmand River in Afghanistan which flows into Hamoun wetlands, a large percentage of natural water resources, particularly those in the northwestern portion, were dried up in the intended period. Accordingly, many environmental, economic, and social problems were produced since a majority of residents were farmer and made their living from this way. The dried-up water resources became a source of dust rise, especially when blowing 120-day winds 2 . However, the construction of Chah-nimeh four in 2009 compensated a part of this damage and greatly helped the ecosystem of the region. This Chah-nimeh is considered as an effective factor in rehabilitating the region. Despite the removal of some agricultural lands, especially irrigated agriculture around the dried-up water resources, the Chah-nimeh resulted in forming agricultural lands in other portions, particularly its surrounding regions. Further, it significantly improved the environmental quality of the study area. It is worth noting that the restoration of natural and nonnatural water resources in the zone directly and indirectly caused the ecosystem rehabilitation through restoring agricultural lands and vegetation cover. Furthermore, this issue was associated with better environmental and socio-economic status.
Around the region under study, most of the water resources inside and outside the borders of Iran, especially on the borders such as Hamoun wetlands, were drying up (Fig. 10). This issue led to groundwater overuse, and consequently land subsidence along with elevating LST. It seems that the proper planning for the rehabilitation of the zone planners and authorities based on the sustainable development can have a high effect on enhancing the region situation.

Conclusion
In this study, we explored the variations of LST and the accompanying relationship with Landscape patterns and environmental indices (NDVI, NDBI, NDWI, LSM, and albedo). Then we presented the temporal-spatial distribution of LST in the region, as well as classified it into five classes (very low, low, moderate, high, and very high) between 1989 and 2019. In addition, the researchers investigated the amount of change in LST about various land uses and different environmental indices over the intended periods.
The literature review revealed some gaps in previous studies related to this sensitive region. First, there is a need for a comprehensive study of environmental indices and land use and land cover changes in the region. Second, understanding LST variations over time in this particular region with severe climatic conditions and poor socio-economic status can help to plan better for the future. Third, this study guides officials and planners to understand the impact of climate change and the reduction of water resources on dry and vulnerable environments in similar international environments.
Based on the study results, it is recommended to plant trees suitable for this climate and resistant to water shortage, as well as prevent the degradation of the vegetation cover in the zone and agricultural lands. Moreover, preventing the conversion of green land uses into urban areas, which can strongly increase LST, is really suggested. In addition, the construction of Chah-nimehs with appropriate spatial distribution in the various parts of the region along with restoring the dried-up water resources is suggested as an effective rehabilitation approach to prevent high soil salinity, more LST, and great dust in the area. Further studies are recommended to be conducted on predicting the future changes in temperature and land use and presenting a codified plan to reduce the effects of temperature rise in the region. Future studies can focus on assessing the socioeconomic condition in the region and its effectiveness on LST and forecasting LST in the next years according to the diverse climate scenarios. Of course, the implementation Fig. 10 Image of the region under study and its surrounding areas 2 The 120-day wind of Sistan is a type of wind that blows from the end of May to the end of September in the Sistan region. The duration of this wind is usually 120 to 130 days and sometimes even up to 170 days. This wind causes soil erosion in the Sistan region of a study in a poor area in terms of environmental and socio-economic conditions in a developing country can be accompanied by limitations. The lack of adequate synoptic stations in the zone and the insufficiency of appropriate classified information about the region were among the limitations of this study.