Introduction

The wide availability of the Landsat data has been extensively used in land use/cover classification and change detection at regional scales (Weng 2001, 2002; Mubea and Menz 2012; Turner et al. 1994). Urbanization process brings changes in land use pattern resulting in significant environmental impacts (Xiao et al. 2002). Land use land cover change by urbanization process, especially the transformation of natural land to urban uses, is the most significant form of global environmental change (Briassoulis 2000). One of the major impacts of LULC change driven by urbanization is the change in land surface temperature (LST) which is closely related with different LULC parameters. Increase of LST gives rise to urban heat island which is a worldwide concerning phenomenon in the cities at present. For controlling the land surface temperature, major influencing factors are the characteristics of surface cover, urban morphology, and anthropogenic activity.

The rapid population growth in urban areas creates pressure on the natural environment and on its resources. As a result, the land use pattern and land cover of the surface change rapidly. This type of scenario is mostly being seen in developing countries as well in India (Mohan et al. 2011). The environmental degradation is not only the result of population growth actually; it is the product of population, energy, industrialization, and urbanization (Bowler et al. 2010). Studies have carried out for evaluating the rise in urban temperature and formation of urban heat islands in the cities of world (Liu and Weng 2008; Ayanlade and Howard 2019). Due to increasing population and infrastructure for rapidly growing urban areas, the urban heat islands (UHIs) have been developed (Sobrino et al. 2004). All cities in India have been experiencing this ongoing phenomenon involving large scale land use changes with globalization (Ramachandra et al. 2012; Mallick and Rahman 2012). Recent studies found out the rapid changes in the land use and land cover of a region which has become a major environmental issue in India (Mohan et al. 2011). The study by Ramachandra clearly shows that the rate of increase in the urbanization leads to change in the land surface temperature. The increase in the temperature is in the range of 3 to 4 °C during 1989 to 2010 in India. The studies by Bowler et al. (2010); Afzan Buyadi et al. (2013) point out that temperature increase can lead to unsustainable development with the reduction of green spaces and also changes in local climate and formation of UHIs.

Fast urbanization in Siliguri Jalpaiguri planning area (SJDA) has been noticed from post independence which has played a significant and key role in the urbanization and development process of West Bengal. Majority of studies based on land use parameters and urbanizations have focused on mega cities of India while very less importance has been given to urban entities of north of West Bengal. SJDA has emerged as the third largest urban agglomeration in West Bengal just after Kolkata and Asansol-Durgapur with 83.65% decadal urban growth rate. This population outgrowth and accelerating urban land have triggered the land cover conversion rapidly to accommodate the need for residential area over the time. A rational and comprehensive strategy for the land use planning and urban environment within a regionally and ecologically sustainable framework has become an absolute necessity in developmental process of the region.

The present study attempts to explore three objectives: (I) To assess the changing pattern of land use in the Siliguri Jalpaiguri planning region, focusing on land conversion of urban areas, (II) to show the relationship between land use land cover and various spatial indices, and (III) to assess the environmental impact of urbanization with reference to reduction of green space and increase of land surface temperature. In order to reach the objectives, the present study has employed statistical, geospatial tools, and indices. Studies have been carried out to explore the relationship between different land use indices through RS and GIS technology. For this purpose, Normalized Difference Vegetation Index (NDVI by Carlson and Ripley 1997; Purevdorj et al. 1998), Normalized Difference Built Up index (NDBI by Zha et al. 2003), and Modified Normalized Difference Water Index (MNDWI by Xu 2006), etc were formulated to show spatial distribution of different LULC types and their relationships. Monitoring and management of land use dynamics would help in land use planning by bridging the knowledge gap between present and past and mitigation of environmental impacts.

Study area

The Siliguri Jalpaiguri Development Area (SJDA) is located (Fig. 1) in the northern part of West Bengal comprising components of two districts, Darjeeling and Jalpaiguri. The southern part of Darjeeling district and the adjacent western part of Jalpaiguri district define the formation of the SJDA. The surrounding hinterland of the area comprises the whole of North Bengal including the five district of Darjeeling, Coochbihar, Jalpaiguri, Malda, and Dinajpur. Physiographically, Siliguri Jalpaiguri Development area may be divided into three divisions: hills, piedmont, and plains. The SJDA zone mostly lies under the piedmont zone with the major central southern part being “Terai-Dooars” and the northern part comprising of alluvial fans old terraces. A very small component of “active plain” zone exists at the south-eastern part of SJDA within Jalpaiguri district.

Fig. 1
figure 1

Location of the study area within West Bengal and within Darjeeling, Jalpaiguri District

Full size image

The study area covers four blocks of Darjeeling district namely Matigara, Naxalbari, Phansidewa, and Khoribari and four blocks of Jalpaiguri districts namely Jalpaiguri, Rajganj, Mal, and Mainaguri. The study area consists of one Municipal Corporation and two municipalities which are Jalpaiguri municipality and Mal municipality. The area includes 23 census towns, 14 in Darjeeling district and 9 in Jalpaiguri district and 540 revenue villages. The land area under the jurisdiction of SJDA is 2658 km2 after the recent inclusion of Mal and a portion of Mainaguri CD block consisting 14 villages. In the census year 2011, the total population of SJDA was 2.37 million, out of 1.28 is rural and 1.09 million populations is urban. The region is growing fast characterized by combined effect of demographic and spatial dynamics (Fig. 1).

Data base and methodology

Landsat imagery is particularly helpful in monitoring land use change detection as it provides an efficient support for wide landscape analysis and such data is regularly updated and easily accessible (Nkeki 2016). The data utilized in this study are cloud-free multi-temporal and multi-spectral Landsat TM (Thematic Mapper) data for 1991, 2001, and Landsat 8 OLI (Operational Land Imager) data for 2017. These satellite data have obtained from USGS GLOVIS (United States Geological Survey Global Visualization Viewer) website to analyze the land use land cover pattern. Specification of the satellite data has provided below (Table 1).

Table 1 Characteristic of Landsat satellite images used in this study
Full size table

Image processing and classification

The obtained Landsat data of two path/row series were mosaicked to extract the study area in a single image with the help of Arc GIS 10.3.1 software. Then the images were georeferenced to UTM 45° N WGS-84 datum for supervised classification and necessary output generation. The methodology of the present analysis can be divided into three stages: generation of land use/land covers map, derivation of spatial indices, and assessment and determination of the relationship between LST and other land use parameter. The whole methodological process adopted in the study has been summarized in below-listed diagram (Fig. 2).

Fig. 2
figure 2

Flow diagram of methodological proceeding

Full size image

Using the Landsat images of three temporal periods, a combination of supervised classification was performed in the ArcGIS 10.3 software for deriving land use layers. The land use categories included forest, light vegetation cover, agricultural plantation, construction land, cultivated land, water, and waste land. Information has been extracted with the use of GIS for land use analysis to detect the changes in the vast expanse of land cover of the study region.

Accuracy assessment of the classified images has been made through Kappa coefficient to validate the post classification analysis. More than 300 reference points were generated on 1991 and 2017 land use map for assessing user’s accuracy, producer’s accuracy, overall accuracy, and Kappa coefficient. Selected points were superimposed on the classified images to characterize the underlying pixel or land use types of each point which were compared with the existing ground characteristics using Google archive data. Kappa coefficient is one of the most reliable accuracy indicators (Foody 1992; Ma and Redmond 1995). Kappa coefficient (K) value is computed by the following Cohen’s formula:

$$ K=\frac{\sum \frac{a}{N}-\sum \mathrm{ef}}{1-\sum \mathrm{ef}} $$
(1)

where a is the sum of diagonal frequency, N is the total number of data, and ef is the expected frequency.

Expected frequency (ef) can be calculated using the following formula:

$$ \mathrm{ef}=\frac{\mathrm{Row}\ \mathrm{total}\times \mathrm{Column}\ \mathrm{total}}{N} $$
(2)

Coefficient value closer to 1 produces perfect agreement between the classified image and referenced image which signifies highest number of correctly classified pixels.

Method of derivation of land surface temperature (LST)

Land surface temperature (LST) of SJDA and its urban area is quantitatively derived from Lantsat TM 5 for the year 1991 and 2001 and Landsat 8 OLI (Operational Land Imager) imagery for the year 2017 having 0% cloud cover. These images of different sensors have different thermal bands such as for Landsat TM 5 band-6 is thermal band, and Landsat OLI imagery provides 11 bands in which band 10 and band 11 are thermal bands. Derivation of temperature from TM and OLI imageries follows different steps in converting digital values into spectral radiance which are discussed below:

  • Step 1: Conversion of band 6 digital values to spectral radiances from Landsat TM 5.

To convert the band 6 digital values into radiance values (Lλ), the following formula has been used (Landsat Project Science Office 2002):

$$ \mathrm{L}\lambda =\frac{\mathrm{LMAX}\lambda -\mathrm{LMIN}\lambda }{\mathrm{QCALMAX}-\mathrm{QCALMIN}}\times \left(\mathrm{QCAL}-\mathrm{QCALMIN}\right)+\mathrm{LMIN}\lambda $$
(3)

where Lλ is the atmospherically corrected cell value as radiance, QCAL is the digital value of imagery, LMINλ is the spectral radiance scales to QCALMIN, LMAXλ is the spectral radiance scales to QCALMAX, QCALMIN is the minimum quantized calibrated pixel value (typically 1), and QCALMAX is the maximum quantized calibrated pixel value (typically 255).

Conversion of band 10 and 11 digital values to spectral radiances from Landsat OLI.

To convert the band 10 and 11 digital values of Landsat OLI, the following formula has been used:

$$ \mathrm{L}\lambda =\mathrm{ML}\times \mathrm{QCAL}+\mathrm{AL} $$
(4)

where Lλ is the TOA (top of atmosphere) spectral radiance, ML is the radiance multi band X, AL is the radiance add band X, Qcal is the quantized and calibrated standard product pixel value, and X is the band number.

Here ML and AL are band-specific multiplicative rescaling factor and band-specific additive rescaling factor respectively, acquired from metadata file (MTL file).

  • Step2: Conversion from spectral radiance to at-satellite brightness temperatures

Emissivity corrections have applied to radiant temperature according to the land cover nature following Artis and Carnahan (1982).

$$ T=\frac{K_2}{Ln\left(\frac{K_1}{L_{\rightthreetimes }}+1\right)}-273.15 $$
(5)

where T represents at-satellite brightness temperature in Kelvin (K), Lλ represents at-satellite radiance in W/(m2 sr μm), and K1 and K2 represent thermal calibration constant in W/(m2 sr μm).

(For Landsat-5 TM, K1 = 607.76 K2 = 1260.56 for band 6, and For Landsat 8 OLI value of K1 for band 10 and 11 are 774.8853 and 480.8883 respectively and K2 for band 10 and 11 are 1321.0789 and 1201.1442 respectively. Values of K2 and K1 were obtained from metadata file.

The values of thermal constant for Landsat TM and Landsat OLI are then converted from Kelvin (K) to °C using the relation of 0 °C equals 273.15 K. for better comprehension.

  • Step 3: Evaluation of land surface emissivity (E)

The above-derived temperature values are referenced to a black body. Therefore, spectral emissivity (E) corrections become necessary. These can be done according to the land cover nature (Snyder et al. 1998) and also by determining the corresponding emissivity values from the proportion of vegetation (Pv) values for each pixel.

$$ E=0.004\ast {\mathrm{P}}_{\mathrm{V}}+0.986 $$
(6)

where proportion of vegetation (PV) can be calculated as:

$$ {\mathrm{P}}_{\mathrm{V}}={\left\{\left(\mathrm{NDVI}-{\mathrm{NDVI}}_{\mathrm{min}}\right)/\left({\mathrm{NDVI}}_{\mathrm{max}}-{\mathrm{NDVI}}_{\mathrm{min}}\right)\right\}}^2 $$
(7)

  • Step 4: Land surface temperature (LST)

The following equation is used to derive LST.

$$ \mathrm{BT}/1+\mathrm{W}\ast \left(\mathrm{BT}/\mathrm{P}\right)\ast \mathrm{Ln}\left(\mathrm{E}\right) $$
(8)

where, BT represents the at-satellite brightness temperature, W is the wavelength of emitted radiance, P = h*c/s (1.438 × 10˄-2 m K), h is the Planck’s constant (6.626 × 10˄-34 Js), s is the Boltzmann constant (1.38 × 10˄-23 J/K) and is the velocity of light (2.998 × 10˄8 m/s).

Methodology for calculating spatial indices

Calculation of Normalized Difference Vegetation Index (NDVI)

To establish the correlation between LST with different spatial indices like NDVI (Normalized Difference Vegetation Index), NDWI (Normalized Difference Water Index), and NDBI (Normalized Difference Built Up Index) of the study area has been extracted by the following way.

Townshend and Justice (1986) method has been applied for extracting NDVI value.

$$ \mathrm{NDVI}=\frac{\left(\mathrm{NIR}\mathrm{band}-\mathrm{Rband}\right)}{\left(\mathrm{NIR}\ \mathrm{band}+\mathrm{Rband}\right)} $$
(9)

where, NIR is the DN (digital number) value from the near infrared band and R is the DN value from the red band. The NDVI value ranges between 0 and 1. Values near to 0 indicate low vegetation cover and the value close to 1 indicates high density of vegetation.

Calculation of Normalized Difference Built up Index (NDBI)

NDBI is obtained by using the following formula (Zha et al. 2003) where the value closer to 1 indicates high density of built-up land.

$$ \mathrm{NDBI}=\frac{\mathrm{MIR}\ \mathrm{band}-\mathrm{NIR}\ \mathrm{band}}{\mathrm{MIR}\ \mathrm{band}+\mathrm{NIR}\ \mathrm{band}} $$
(10)

where, MIR is the DN value from the middle infrared band and NIR is the near infrared band.

Calculation of Modified Normalized Difference Water Index (MNDWI)

$$ \mathrm{MNDWI}=\frac{\left(\mathrm{Green}\ \mathrm{band}-\mathrm{NIR}\ \mathrm{band}\right)}{\left(\mathrm{Green}\ \mathrm{band}-\mathrm{NIR}\ \mathrm{band}\right)} $$
(11)

To eliminate the problem of inclusion of built-up area in NIR band-modified NDWI (MNDWI) is being applied (Gu et al. 2008), where, green means green band and NIR is the near-infrared band.

After calculating the values of different spatial indices, output maps have been prepared to show spatial distribution of vegetation cover, built-up areas, water coverage over the time.

Statistical analysis and the relationship between LST and related indices

The distribution pattern of LST among different LULC types such as built-up area, vegetation cover, agricultural land, water bodies, and waste land, has been obtained along the projected profile lines. In order to show the relationship between LST and spatial indices, regression analysis is carried out with the values measured from selected sample points over the region. The correlation is determined from slopes of regression lines by fitting the trend line using Microsoft Excel.

Results and analysis

Accuracy assessment

The accuracies of classified land use land cover maps were assessed by verifying selected samples of each land cover classes with the help of ground truth data from Google archive images. Tables 2 and 3 display the results of accuracy assessment of post classification, i.e., the user’s accuracy, producer’s accuracy, overall accuracy, and Kappa coefficient of classified LULC layers for the years 1991 and 2017. The post land use classification overall accuracy of 1991 and 2017 are 87.20% and 93.50% which indicates good accuracy results and effectiveness of the adopted classification technique (Tables 2 and 3).

Table 2 Accuracy assessment of land use land cover map of 1991 from Landsat TM data
Full size table
Table 3 Accuracy assessment of land use land cover map of 2017 from Landsat OLI data
Full size table

Land use and land cover classification

Satellite images of three different years (1991, 2001, and 2017) have been compared to quantify the magnitude of LULC change in SJDA for last 26 years. Seven land use types were identified for obtaining the information of changes in land use pattern induced by growth of population (Table 4 and Fig. 3).

Table 4 Land use/cover description
Full size table
Fig. 3
figure 3

Percentage change in decadal population in SJDA (1971–2011)

Full size image

The state of land use and land cover in three temporal periods is provided in Table 5 and Fig. 4. In 1991, chief portion of land use has been given over to agricultural practice (44.63%) and intensive tea plantation (13.62%) which constitutes a major part in the region. Forest area is located primarily in the northern and north western hilly and Terai regions of the area. Forest tracts are mutated into new cultivable lands with growing population. The share of residential land use to total geographical land of the region is only 4% in 1991. This indicates that the area has substantial land for accommodating growing population.

Fig. 4
figure 4

Land use/land cover maps of SJDA. a 1991. b 2001. c 2017

Full size image

In 2001, the most prevalent land cover is agriculture though in 2001 agricultural land squeezed from 1187 to 1027 Km2 due to transformation of agricultural land to mainly tea gardens. Tea plantation land in 2001 and 2017 also, had been accelerated from 19% in 2001 to 29.17% in 2017 of total land cover. Forest area reduced in coverage area, and light vegetation also remarkably decreased from 11% in 2001 to 8% in 2017 (Table 5 here).

Table 5 Land use/land cover distribution of SJDA: 1991, 2001, and 2017
Full size table

In 2017 (Fig. 4c), again agricultural land decreased to 885 Km2 sharing 30.74% land cover area, due to the trend associated with intensification of agricultural plantation and shifting occupational pattern of settlements from traditional forest settlement to plantation settlement. So the areas where the forests still survive are now more interspersed by numerous patches of tea gardens as well as cultivation. Due to the formation of reserviors and channels, the water management system of the region was improved which reflected in rise of water coverage area from 1991 to 2017. Accelerating constructional land results in a huge decrease (56%) of waste land in last 26 years. The LULC maps displaying built-up area increased distinctly in coverage as the region has experienced very rapid growth rate (122%) of built-up area from 95 to 211 Km2 in between 1991 and 2001 (Fig. 4a–c).

Land conversion

Transition matrix between 2001 and 2017 has been produced to assess the current pattern of land cover change in the region. In between 2001 and 2017 (ref.Table 6), significant transition of wasteland (72.75%) to agricultural field, water body, and built-up area is found. Sixty-nine percent converted water body area was shared by agricultural field, waste land, and built-up area. Significant change is marked for agricultural land which mainly gained its land from waste land (36.51%), water body (28.86%), and light vegetation (3.22%). Tea garden area expanded at the expenses of agricultural field (30.54%), light vegetation (22.34%), and forest (12%). (Table 6).

Table 6 Land cover transition matrix of SJDA, from 2001 to 2017
Full size table

Percent change of built-up area is 39% within the time span of 2001–2017, and this built-up area has increased significantly at the cost of the decrease of water body, light vegetation, and waste land. As the expansion in built-up land was predominantly in the areas immediately surrounding the core urban areas, 10-Km buffer zone around the center from Siliguri Municipal Corporation (SMC) and 5-Km buffer area from Jalpaiguri municipal (JM) area has taken to visualize land conversion dynamics clearly (Fig. 5). It is noted from the land transform map of SMC and JM that in between 2001 and 2017, agricultural lands (37.35%), sparse vegetation (10.94%), water bodies (3.69%), and vacant land (2.86%) are more affected by residential expansion (Fig. 5).

Fig. 5
figure 5

Land transformation to residential area around the two major urban centers of SJDA from 2001 to 2017

Full size image

From the analysis of 1991, 2001, and 2017 land use maps, it is evident that most of the areas in the region have undergone a transition in status from forest and vegetated area to extensively cultivated, plantation, and expanding residential land.

Impacts of land cover change

Change in land surface temperature distribution (LST)

Land surface temperature distribution maps of Siliguri Jalpaiguri Planning Area in different temporal periods have generated to present the change in LST distribution (Fig. 6 here). It is evident from the figures that LST values increased throughout the years from outskirts towards the inner urban areas. Figure 6 a) shows that in 1991 the lowest radiant temperature is 9.12 °C in dense hilly forest area in the northern part and the highest radiant temperature is 21.84 °C in the built-up area. For the year 2001, the radiant temperature ranges between 19.30 and 26.25 °C; the highest mean temperature is within built-up area, and the lowest temperature is within forest and water body bodies. However, the areas of sand deposits show high temperature because of high temperature reflectance. In 2017 (Fig. 6c), the highest temperature is increased to 31.1 °C clustered in the inner urban area. Table 7 depicts clear impression of the changes of LST over different land use units from 1991 to 2017. LST which is considerably increased over each land cover units specially, built-up land, sand deposition, and water body area has experienced 6.25 °C, 8.3 °C, and 7.99 °C increase of mean LST respectively. On average over 26 years of observation reveal almost 7 °C increase of mean LST in the study area (Tables 7 and 8).

Fig. 6
figure 6

Spatial distribution of land surface temperature in different temporal periods. a 1991. b 2001. c 2017

Full size image
Table 7 Land use units wise changes in LST
Full size table
Table 8 Land surface temperature statistics of different temporal periods
Full size table

Temperature distribution along LST transects profile by LULC types

Two profiles across (A-B) the same area and month (January) for two different years are generated to show the variations of land surface temperature in between 26 years. Basically, the spatial patterns of LST concentration highlight the change of LULC distribution as there is close association between LST and LULC types (Fig. 7).

Fig. 7
figure 7

Comparison of LST profiles over different land cover types

Full size image

Comparison of change in distribution of temperature with different LULC types reveals in Fig. 7. The graph plotted in Fig. 7a shows highest land surface temperature value of 20.4 °C in built-up area in 1991. Temperature cross section profile gradually changes over different land cover areas. The surface temperature is much lower in forested area, and the lowest temperature acquired in the water body which is 11.2 °C. Surface temperature in tea garden area, agricultural field, scattered vegetated area shows moderate temperature ranges from 15 to 17.5 °C.

In 2017, Fig. 7b insights a different temperature statistics across the same profile. Temperature in the built-up surface rises to 28.5 °C though highest temperature acquired in sand deposited area because of high temperature absorption. Lowest temperature found in water body area is 17 °C. The temperature in forest, tea garden, and agricultural land is significantly lowered than built-up land which ranges between 20 and 26 °C. So the transect profiles clearly show that LST rises much high during 1991–2017 in different land cover category.

Urban temperature analysis

Special emphasis of the study has been given over to the analysis of urban land surface temperature profiles of SJDA. The changes of temperature in different temporal periods from core urban to periphery area along the profile covering Siliguri municipal corporation (SMC), its surrounding suburbs, and rural areas have been extracted (Fig. 8). The temperature of the core built-up areas of SMC is much higher than the areas located further away. Throughout the years, urban built-up area exhibits the highest surface temperature of 18.13 °C in1991, 22.5 °C in 2001, and accelerates to 29.4 °C in 2017. This implies rapid development of dense built-up surface such as stone, concrete, and metal constructions by replacing natural features as vegetation water body, which increases the surface radiant temperature. The difference in derived radiant temperature between urban, sub urban, and rural residential areas has been summarized in Table 8. Nevertheless, the temperature difference clearly indicates the formation of Urban Heat Island (Fig. 8 and Table 9).

Fig. 8
figure 8

Urban heat island profile

Full size image
Table 9 Land surface temperature by residential land types
Full size table

Figure 9 highlights the proportion of area within the 10-Km buffer zone around SMC, under different temperature zones and it is shifting from 1991 to 2017. A huge shifting of area under different temperature classes can be visualized within last 26 years. While in 1991, the largest proportion of area (47.96%) lied under the zone of 15.1–16.8 °C; in 2017, it is found that 70% area lie under the temperature of 20.64–24.05 °C, and 27% area lie under 24.5–27.49 °C (Fig. 9).

Fig. 9
figure 9

Proportion of urban area under different temperature classes from 1991 to 2017

Full size image

Analysis of spatial characteristics of NDVI, NDBI, and MNDWI

Vegetation cover maps have been prepared by calculating NDVI for the year 1991, 2001, and 2017. The descriptive statistics of NDVI, NDBI, and NDWI through temporal periods are shown in Table 10. Coefficient of variation (CV) has been put to depict the relative dispersion of measured spatial indices for each study year.

Table 10 Maximum, minimum, mean, S.D., and C.V of NDVI, NDBI, and MNDWI in study years
Full size table

The decrease in the vegetation growth coverage (scatter vegetation and forest) within the urban area can clearly be seen. Lower NDVI value was also located in water body areas. Higher NDVI values are observed in the middle portion of dense forest area of the region. NDVI reflects an opposite picture of land surface temperature as NDVI value increases hence LST value decreases. Vegetation cover highlights lowest mean NDVI value in 2017 with a value of − 0.23 and higher variation (43.48) of vegetation cover in 2017 (Fig. 10).

Fig. 10
figure 10

Spatial distribution of a NDVI, b MNDWI, and c NDBI of different temporal periods

Full size image

MNDWI pattern of the year 1991, 2001, and 2017 has extracted by using MNDWI index (Fig. 10b). There is significant change in the coverage of the water bodies as maximum MNDWI value decreases from 1 in 1991 to 0.19 in 2017. The main water body is river Teesta in the north eastern part of the region where MNDWI concentration is highest at the Teesta barrage. Presence of water bodies helps in reducing its own as well as surrounding temperature at some level (Table 10).

Built-up maps are generated by using NDBI (Normalized Difference Built up Index) to visualize built-up growth (Fig. 10c) of the area. High NDBI values are found in built-up land and vacant land of dense town areas. Maximum NDBI value remarkably increased due to land transformed to constructed land with industrial and commercial building, high rise residential building, facilities building, road, and transport communication. Hence high NDBI value can be visualized in built-up areas and other impervious areas, over the water bodies, and vegetation cover, NDBI values can be observed low.

Correlation between LST and LULC types

To obtain the relationship between LST and other spatial indices, 60 sample points are measured within the planning region during the study years. The extracted values of the selected parameters (i.e., LST, NDVI, NDBI, and NDWI) are put for regression model analysis. Specifically, the relationship between LST and different land use types was carried out by analyzing linear regression model for each land use respectively. Strong negative correlation (ref. Figure 11a) between surface temperature and vegetation cover can be found in the area where the value of R2 (coefficient of determination), derived from the regression model, are 0.533, 0.49, and 0.61 for the year 1991, 2001, and 2017 respectively. The high value of R2 in 2017 established the fact that vegetation plays vital roles to alleviate surface radiant temperature (Fig. 11).

Fig. 11
figure 11

Regression correlation between LST and a NDVI, b MNDWI, and c NDBI

Full size image

Negative correlation has been depicted between LST and NDWI implying lower temperature in water body area and higher temperature in non water body areas. This relationship clearly emerged from the linear regression model where the high R2 value is 0.43 in 2017 implying that water body helps in reducing LST.

Figure 11 c shows perfect positive relation between LST and NDBI. The R2 value generated from the model is higher in 2017 which is 0.68. Such a high value of R2 established the facts that increase in built up or impervious surface trapped the radiation that positively controls LST. Thus, it is emerged from the study that land surface temperature (LST) is sensitive to each land use type, hence can be used to detect land use/land cover changes.

Conclusions

The study arrives at the major findings that there is an evident change in the spatial pattern of land use to accommodate the dynamic growth of the planning region. Due to rapid population growth in the urban areas of SJDA, there are dramatic changes in the land use pattern. A huge amount of light vegetation and agricultural land has been cleared; open spaces and wetlands are transformed to urban construction land cover. The study has confirmed that land cover alteration has great impact on the environment in terms of land surface temperature and green space reduction. The study also puts in evidence the accelerating temperature difference between rural and urban which leads to the formation of urban heat islands. Vegetation cover within the urban environment has significant impacts on the distribution of the UHI as green space regulates the radiant surface temperature. High LST has been found at dense built-up urban land which triggered the UHI phenomenon of the study region.

One of the significant outputs of the study is the assessment and evaluation of different spatial indices based on LULC types over the years and plotting of regression onto spatial indices with respect to LST. The relation emerges as NDBI and LST were positive, and the correlations between MNDWI, NDVI, and LST were negative. This can be used to understand and monitor the spatial dynamics of the region based on LULC changes by urbanization. Therefore, comprehensive analysis of changing land use pattern and its effects driven by urbanization could provide an insight on proper land use planning to mitigate the UHI effects and provide ecological sustainability in the region.