1 Introduction

Urbanization accelerates the ecological stress by warming the local or global cities for a large extent [1,2,3,4,5,6]. Presently, many urban areas are suffering with a huge land conversion and resultant new heat zones [7,8,9]. Remote sensing techniques are significantly effective in detecting the land use/land cover (LULC) change and its consequences [10]. Several satellite sensors are capable to identify these change zones by using their visible and near-infrared (VNIR) and shortwave infrared (SWIR) bands [11]. Apart from the conventional LULC classification algorithms, some spectral indices are used in detecting specific land features. Normalized difference vegetation index (NDVI) can be considered as the most applied spectral index in this scenario [12]. Recently, thermal infrared (TIR) bands are also used by generating some indices for different types of LULC extraction [13,14,15]. These remote sensing indices are used significantly in several application fields like rocks and mineral mapping, forest mapping, agricultural monitoring, LULC mapping, hazard mapping, urban heat island mapping and monitoring, among others [14, 16,17,18,19].

Land surface temperature (LST) retrieved from several remotely sensed data is widely used in the detection of urban heat island and ecological comfort zone [20,21,22,23]. LST can change significantly in a vast homogeneous land surface or even inside a relatively small heterogeneous urban area [14, 24, 25]. Different types of LULC response differently in TIR band and consequently LST largely varies in an urban environment [26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44]. The LULC types are mainly changed by land conversion process [10]. Thus, time is an important factor in LST monitoring. These spatial and temporal data of LST is also varied with the seasonal changes as sun elevation and sun azimuth are changed with seasons. Hence, the seasonal variation of LST is quite important in any LULC related study.

NDVI is a dominant factor in LST derivation processes and is used invariably in any LST related study [45,46,47,48,49]. NDVI is directly used in the determination of land surface emissivity and thus is a significant factor for LST estimation [50, 51]. It also determines the LULC categories by its optimum threshold limits in different physical environment [14]. Being a vegetation index, NDVI depends largely on seasonal variation [12]. Hence, LST is also regulated by the change of seasons. Thus, seasonal evaluation of LST and NDVI is an important task in LST mapping and monitoring, especially in an urban landscape.

The relationship of LST with NDVI is quite interesting and it attracts the remote sensing scientists from several directions [52,53,54,55]. The nature and strength of this relationship heavily depend on space and time. Generally, in the tropical environment the LST-NDVI relationship is negative [56,57,58]. The negativity of the relationship is determined by the changing type of LULC over time. Thus, spatial and temporal changes in this relationship are observed on different types of LULC. Apart from the spatial and temporal changes, seasonal variation of LST-NDVI relationship is a very important study in any mixed urban land surface.

Several studies are available on the seasonal analysis of LST-NDVI relationship. Many tropical cities are a part of these studies. Many valuable research articles found on LST-NDVI relationships in the Chinese landscape [59,60,61,62,63,64,65,66,67]. Some studies were also performed in Indian urban landscape [35, 36, 68,69,70,71].These studies found that LST builds a negative relationship with NDVI and this negativity can change with season. Wet season reflects a stronger negative correlation than dry season as the moisture content is more in the wet season [72]. This relationship can also change with the change of land surface types. Vegetation surface builds a strong correlation and the strength is reduced on bare land surface, built-up surface, and water surface.

The present study calculates the LST and NDVI from Landsat datasets of four different seasons (winter, pre-monsoon, monsoon, and post-monsoon) in Raipur City of India using a total of 12 Landsat satellite images for 1991, 2004, and 2018. Meanwhile, the LULC map has been obtained by suitable threshold values of NDVI. The main aims of the study were (1) to analyze the seasonal variation of spatial distribution pattern of the LST in the study area, (2) to determine the seasonal variation of LST-NDVI relationship for whole of the city, and (3) to explore the seasonal variation of LST-NDVI relationship on different LULC types.

2 Study area and data

Figure 1 shows the study area (Raipur City of India) of the present research work including the false colour composite (FCC) image of and digital elevation model (DEM) image. Raipur is one of the fastest-growing smart cities in India. The latitudinal and longitudinal extent of the city is from 21°11′22″ N to 21°20′02″ N and from 81°32′20″ E to 81°41′50″ E. The total area of the city is approximately 164.23 km2. The only big river is Mahanadi which flows along the eastern boundary of Raipur. The southern part of the city is covered by dense forests. Geologically the city is very stable. According to India Meteorological Department (IMD) (https://mausam.imd.gov.in) [75], the study area is under a tropical wet and dry climate with four typical seasons (pre-monsoon, monsoon, post-monsoon, and winter). May is the hottest month followed by April, June, and March. July is the rainiest month followed by August, June, and September. October and November are the post-monsoon months experience a pleasant weather condition. December (the coldest month), January, and February are the winter months. The pre-monsoon and winter months (including November) remain dry compared to the monsoon and post-monsoon months.

Fig. 1
figure 1

Location of the study area: a India b Chhattisgarh c FCC image of Raipur City d DEM image of Raipur City (Source: https://earthexplorer.usgs.gov [73] and http://www.surveyofindia.gov.in [74])

Full size image

Four Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensors (TIRS) data of 2018; four Landsat 5 Thematic Mapper (TM) data of 2004; and four Landsat 5 TM data of 1991 have been freely downloaded from the United States Geological Survey (USGS) (https://earthexplorer.usgs.gov) Data Centre (Table 1). Landsat 8 TIRS dataset has two TIR bands (bands 10 and 11) in which band 11 has uncertainty in calibration. Thus, only TIR band 10 (100 m resolution) has been recommended for the present study. The TIR band 10 has been resampled to 30 m × 30 m pixel size with cubic convolution method by the USGS. Landsat 5 TM data has only one TIR band (band 6) of 120 m resolution which has also been resampled to 30 m × 30 m pixel size with cubic convolution method by the USGS. The spatial resolution of visible to near infrared (VNIR) bands of the two types of Landsat datasets is 30 m.

Table 1 Specification of Landsat data sets (Source: USGS)
Full size table

3 Methodology

3.1 Retrieving LST from Landsat data

In this study, the mono-window algorithm was applied to retrieve LST from multi-temporal Landsat satellite sensors [1, 76,77,78,79] where three necessary parameters are ground emissivity, atmospheric transmittance, and effective mean atmospheric temperature. At first, the original TIR bands (100 m resolution for Landsat 8 OLI/TIRS data, 120 m resolution for Landsat 5 TM data) were resampled into 30 m by USGS data centre for further application.

The TIR pixel values are firstly converted into radiance from digital number (DN) values. Radiance for TIR band of Landsat 5 TM data is obtained using Eq. (1) (USGS):

$$L_{\lambda } = \left[ {\frac{{L_{MAX\lambda } - L_{MIN\lambda } }}{{QCAL_{MAX} - QCAL_{MIN} }}} \right]*\left[ {Q_{CAL} - QCAL_{MIN} } \right] + L_{MIN\lambda }$$
(1)

where \(L_{\lambda }\) is Top of Atmosphere (TOA) spectral radiance (Wm−2 sr−1 mm−1), \(Q_{CAL}\) is the quantized calibrated pixel value in DN, \(L_{MIN\lambda }\) (Wm−2 sr−1 mm−1) is the spectral radiance scaled to \(QCAL_{MIN}\), \(L_{MAX\lambda }\) (Wm−2 sr−1 mm−1) is the spectral radiance scaled to \(QCAL_{MAX}\), \(QCAL_{MIN}\) is the minimum quantized calibrated pixel value in DN and \(QCAL_{MAX}\) is the maximum quantized calibrated pixel value in DN. \(L_{MIN\lambda }\), \(L_{MAX\lambda }\), \(QCAL_{MIN}\), and \(QCAL_{MAX}\) values are obtained from the metadata file of Landsat TM data. Radiance for Landsat 8 TIR band is obtained from Eq. (2) [80]:

$$L_{\lambda } = M_{L} \cdot Q_{CAL} + A_{L}$$
(2)

where \(L_{\lambda }\) is the TOA spectral radiance (Wm−2 sr−1 mm−1), \(M_{L}\) is the band-specific multiplicative rescaling factor from the metadata, \(A_{L}\) is the band-specific additive rescaling factor from the metadata, \(Q_{CAL}\) is the quantized and calibrated standard product pixel values (DN). All of these variables can be retrieved from the metadata file of Landsat 8 data.

For Landsat 5 data, the reflectance value is obtained from radiances using Eq. (3) (USGS):

$$\rho_{\lambda } = \frac{{\pi \cdot L_{\lambda } \cdot d^{2} }}{{ESUN_{\lambda } \cdot \cos \theta_{s} }}$$
(3)

where \(\rho_{\lambda }\) is unitless planetary reflectance, \(L_{\lambda }\) is the TOA spectral radiance (Wm−2 sr−1 µm−1), \(d\) is Earth-Sun distance in astronomical units, \(ESUN_{\lambda }\) is the mean solar exo-atmospheric spectral irradiances (Wm−2 µm−1) and \(\theta_{s}\) is the solar zenith angle in degrees. \(ESUN_{\lambda }\) values for each band of Landsat 5 can be obtained from the handbooks of the related mission. \(\theta_{s}\) and \(d\) values can be attained from the metadata file.

For Landsat 8 data, reflectance conversion can be applied to DN values directly as in Eq. (4) [80]:

$$\rho_{\lambda } = \frac{{M_{\rho } \cdot Q_{CAL} + A_{\rho } }}{{\sin \theta_{SE} }}$$
(4)

where \(M_{\rho }\) is the band-specific multiplicative rescaling factor from the metadata, \(A_{\rho }\) is the band-specific additive rescaling factor from the metadata, \(Q_{CAL}\) is the quantized and calibrated standard product pixel values (DN) and \(\theta_{SE}\) is the local sun elevation angle from metadata file.

Equation (5) is used to convert the spectral radiance to at-sensor brightness temperature [81, 14]:

$$T_{b} = \frac{{K_{2} }}{{\ln \left( {\frac{{K_{1} }}{{L_{\lambda } }} + 1} \right)}}$$
(5)

where \(T_{b}\) is the brightness temperature in Kelvin (K), \(L_{\lambda }\) is the spectral radiance in Wm−2 sr−1 mm−1; \(K_{2}\) and \(K_{1}\) are calibration constants. For Landsat 8 data, \(K_{1}\) is 774.89, \(K_{2}\) is 1321.08 (Wm−2 sr−1 mm−1). For Landsat 5 data, \(K_{1}\) is 607.76, \(K_{2}\) is 1260.56 (Wm−2 sr−1 mm−1).

The land surface emissivity \(\varepsilon\), is estimated using the NDVI Thresholds Method [51, 82].

In NDVI Threshold Method, there are three following equations:

  1. a.

    \(NDV\text{I} < 0.2\) for bare soil;

  2. b.

    \(NDV\text{I} > 0.5\) for vegetation;

  3. c.

    \(0.2 < = NDV\text{I} < = 0.5\) for mixed land with bare soil and vegetation.

In the last case, \(\varepsilon\) is estimated from Eq. (6):

$$\varepsilon = \varepsilon_{v} F_{v} + \varepsilon_{s} (1 - F_{v} ) + d\varepsilon$$
(6)

where \(\varepsilon\) is land surface emissivity, \(\varepsilon_{v}\) is vegetation emissivity, \(\varepsilon_{s}\) is soil emissivity, \(Fv\) is fractional vegetation, \(d\varepsilon\) is the effect of the geometrical distribution of the natural surfaces and internal reflections that can be expressed by Eq. (7):

$$d\varepsilon = (1 - \varepsilon_{s} )(1 - F_{v} )F\varepsilon_{v}$$
(7)

where \(\varepsilon_{v}\) is vegetation emissivity, \(\varepsilon_{s}\) is soil emissivity, \(Fv\) is fractional vegetation, \(F\) is a shape factor whose mean is 0.55, the value of \(d\varepsilon\) may be 2% for mixed land surfaces [51].

The fractional vegetation \(F_{v}\), of each pixel, is determined from the NDVI using Eq. (8) [50]:

$$F_{v} = \left( {\frac{{NDVI - NDVI_{\text{min} } }}{{NDVI_{\hbox{max} } - NDVI_{\text{min} } }}} \right)^{2}$$
(8)

where \(NDVI_{\text{min} }\) = 0.2 and \(NDVI_{\text{max} }\) = 0.5.

Finally, the land surface emissivity \(\varepsilon\) can be expressed by Eq. (9):

$$\varepsilon = 0.004*F_{v} + 0.986$$
(9)

where \(\varepsilon\) is land surface emissivity, \(Fv\) is fractional vegetation.

Water vapour content is estimated by Eq. (10) [29, 83]:

$$w = 0.0981*\left[ {10*0.6108*\exp \left( {\frac{{17.27*(T_{0} - 273.15)}}{{237.3 + (T_{0} - 273.15)}}} \right)*RH} \right] + 0.1697$$
(10)

where \(w\) is the water vapour content (g/cm2), \(T_{0}\) is the near-surface air temperature in Kelvin (K), \(RH\) is the relative humidity (%). These parameters of atmospheric profile are the average values of 14 stations around Raipur which are obtained from the Meteorological Centre, Raipur (http://www.imdraipur.gov.in) [84] and the Regional Meteorological Centre, Nagpur (http://www.imdnagpur.gov.in [85]). Atmospheric transmittance is determined for Raipur City using Eq. (11) [76, 86]:

$$\tau = 1.031412 - 0.11536w$$
(11)

where \(\tau\) is the total atmospheric transmittance, \(\varepsilon\) is the land surface emissivity.

Raipur City is located in the tropical region. Thus, Eq. (12) is applied to compute the effective mean atmospheric transmittance of Raipur [76, 86]:

$$T_{a} = 17.9769 + 0.91715T_{0}$$
(12)

LST is retrieved from Landsat 5 TM and Landsat 8 OLI/TIRS satellite data by using Eqs. (13–15) [76]:

$$T_{s} = \frac{{\left[ {a\left( {1 - C - D} \right) + \left( {b\left( {1 - C - D} \right) + C + D} \right)T_{b} - DT_{a} } \right]}}{C}$$
(13)
$$C = \varepsilon \tau$$
(14)
$$D = \left( {1 - \tau } \right)\left[ {1 + \left( {1 - \varepsilon } \right)\tau } \right]$$
(15)

where \(\varepsilon\) is the land surface emissivity, \(\tau\) is the total atmospheric transmittance, \(C\) and \(D\) are internal parameters based on atmospheric transmittance and land surface emissivity, \(T_{b}\) is the at-sensor brightness temperature, \(T_{a}\) is the mean atmospheric temperature, \(T_{0}\) is the near-surface air temperature, \(T_{s}\) is the land surface temperature, \(a = - 6 7. 3 5 5 3 5 1\), \(b = 0. 4 5 8 6 0 6\).

3.2 Extraction of different types of LULC by using the threshold limits of NDVI

NDVI can extract different types of LULC by using the optimum threshold values [14, 87,88,89]. This threshold values can differ with respect to the differences in physical environment. The NDVI threshold limits were applied on the post-monsoon images to extract the different LULC types accurately. Table 2 presents the suitable threshold limits of NDVI used for extracting the vegetation (> 0.2), water bodies (< 0), built-up area/bare land (0–0.2) in the study area.

Table 2 NDVI used for extracting different types of LULC [14]
Full size table

4 Results and discussion

4.1 Accuracy assessment for LULC classification

The maximum likelihood classification method was applied to validate NDVI threshold based LULC classification. The overall accuracy values of the LULC classification were 95.00%, 85.00%, and 87.50% in 1991, 2004, and 2018, respectively. The kappa coefficients for the LULC classification were 0.91, 0.76, and 0.78 in 1991, 2004, and 2018, respectively. The kappa coefficient value of > 0.75 reflects the compatibility of the classification method [90]. In the present study, the average overall accuracy and average kappa coefficient were 89.17% and 0.82, respectively. Thus, the NDVI threshold method based LULC classification was significantly validated.

4.2 Extraction of LULC types using NDVI

Figure 2 shows the FCC images and LULC maps of the post-monsoon Landsat images of 1991, 2004, and 2018. Generally, the post-monsoon images reduce the level of air pollution due to the presence of high moisture content in the air and these images also enhance the greenness of an area. Thus, the post-monsoon images are generally considered for the generation of LULC maps. LULC maps were generated using the threshold limits of NDVI for different types of LULC by using ArcGIS software (https://www.esri.com/) [91]. In 1991, built-up area and bare land are mainly found in the northwest and middle portions of the city. Land conversion accelerates the decrease of vegetal covered area especially during 2004-2018. The major segment of vegetation is mainly found in the east and southwest parts of the city.

Fig. 2
figure 2

FCC satellite images of post-monsoon season: a 14-OCT-1991 b 15-OCT-2004 c 22-OCT-2018; LULC maps in post-monsoon season: d 12-OCT-1991 e 15-OCT-2004 f 22-OCT-2018

Full size image

4.3 Characteristics of the spatial distribution of LST and NDVI

Table 3 shows the LST and NDVI values for different multi-date satellite data. The pre-monsoon image (Fig. 3) has the maximum values of mean LST (29.50 °C in 1991, 36.80 °C in 2004, and 37.90 °C in 2018) followed by monsoon (Fig. 4) image (25.74 °C in 1991, 29.23 °C in 2004, and 31.08 °C in 2018), post-monsoon (Fig. 5) image (24.99 °C in 1991, 27.56 °C in 2004, and 29.01 °C in 2018), and winter (Fig. 6) image (23.99 °C in 1991, 25.17 °C in 2004, and 26.91 °C in 2018). The mean LST of the city is increased by 8.40 °C in the pre-monsoon season, 5.34 °C in the monsoon season, 4.02 °C in the post-monsoon season, and 2.92 °C in the winter season during the whole time span (1991–2018). In the case of NDVI, the maximum value is decreased gradually with time (Figs. 3, 4, 5 and 6). Seasonally, the highest values of NDVI are observed in the post-monsoon images followed by the monsoon, pre-monsoon, and winter images. The figures show that the proportion of vegetation is gradually reduced with time and NDVI is inversely related to LST.

Table 3 Temporal and seasonal variation of LST, NDVI, and LST-NDVI relationship (1991–2018)
Full size table
Fig. 3
figure 3

Spatial distribution of LST in pre-monsoon season: a 18-MAR-1991 b 22-APR-2004 c 28-MAR-2018; spatial distribution of NDVI in pre-monsoon season: d 18-MAR-1991 e 22-APR-2004 f 28-MAR-2018

Full size image
Fig. 4
figure 4

Spatial distribution of LST in monsoon season: a 26-SEP-1991 b 09-JUN-2004 c 16-JUN-2018; spatial distribution of NDVI in monsoon season: d 26-SEP-1991 e 09-JUN-2004 f 16-JUN-2018

Full size image
Fig. 5
figure 5

Spatial distribution of LST in post-monsoon season: a 12-OCT-1991 b 15-OCT-2004 c 22-OCT-2018; spatial distribution of NDVI in post-monsoon season: d 12-OCT-1991 e 15-OCT-2004 f 22-OCT-2018

Full size image
Fig. 6
figure 6

Spatial distribution of LST in winter season: a 14-FEB-1991 b 02-DEC-2004 c 08-FEB-2018; spatial distribution of NDVI in winter season: d 14-FEB-1991 e 02-DEC-2004 f 08-FEB-2018

Full size image

4.4 Relationship between LST and LULC

The LST of the study area is significantly dependent upon the LULC types. Actually, this NDVI-threshold based emissivity method is not suitable for LST extraction of water bodies. However, the present results show that the area with green vegetation has low LST value, whereas the built-up areas and bare lands have moderate to high LST value. In pre-monsoon season, built-up area and bare land has comparatively high LST than the other LULC types. But in the winter season, these areas have comparatively low to moderate LST due to low emissivity. The green areas and water areas are characterized by a relatively stable range of LST.

4.5 LST variation with the change in LULC types

Table 4 presents the temporal changes in LST with the changes in LULC types. Only the post-monsoon images of 1991, 2004, and 2018 were considered for this analysis. Generally, the land is converted into the built-up area or bare land from the other types of LULC, e.g., vegetation or water bodies. No such exceptional cases are seen in Raipur City during the entire period. Built-up area and bare land are increased while vegetation and water bodies are decreased significantly. The mean LST of the built-up area and bare land is increased from 1991 to 2004 (4.23 °C in post-monsoon season) and from 2004 to 2018 (1.49 °C in post-monsoon season), irrespective of any season. The green area gains 4.66 °C mean LST when it is converted into the built-up area and bare land between 1991 and 2018, and gained 2.20 °C mean LST between 2004 and 2018. The converted land from water bodies to the built-up area and bare land gains 2.49 °C during 1991–2018 and 0.96 °C mean LST during 2004–2018, respectively. Furthermore, the unchanged built-up area and bare land have also witnessed an increase in LST during the entire time span (2.79 °C mean LST from 1991 to 2018 and 0.83 °C mean LST from 2004 to 2018). Hence, the results indicate significantly to the trend of climate change.

Table 4 Change in mean LST (°C) with the conversion of different types of LULC into the built-up area/bare land
Full size table

4.6 Seasonal variation on LST-NDVI relationship

Table 3 presented the seasonal variation of mean LST values. Winter images indicate the lowest mean LST for 1991, 2004, and 2018. The highest mean LST values are found in the pre-monsoon images for all the three years. From 1991 to 2004, mean LST has been increased in every season. From 2004 to 2018, mean LST has been increased again for all the seasons. Post-monsoon images have mean LST value nearer to winter images while monsoon images have a slightly high value of mean LST than the post-monsoon images. Figure 7a–d show the seasonal variation of LST-NDVI relationships on different LULC types in pre-monsoon, monsoon, post-monsoon, and winter season, respectively. Here, only three types of LULC were considered, i.e., (1) vegetation, (2) water bodies, and (3) built-up area and bare land. On vegetation, the LST-NDVI relationships were negative, irrespective of any season. Only winter season (Fig. 7d) has weak negative regression, while the other three seasons (Fig. 7a–c) have moderate to strong negative regression. Since, NDVI is a vegetation index the LST-NDVI relationship is strongly effective on vegetation. On water bodies, the relationship is positive (weak to moderate). In the post-monsoon season (Fig. 7c), the relationship is weak to moderate. In the rest of the three seasons (Fig. 7a–d), the relationship is moderate. On the built-up area and bare land, the relationship is not so much significant. All four seasons (Fig. 7a–d) indicate weak regression as the surface materials become more heterogeneous in nature. Figure 7e represents a generalized view of the overall seasonal variation of LST-NDVI relationships. The relationship is negative, irrespective of any season. In winter, the relationship was weak negative (− 0.12 in 1991, − 0.2 in 2004, and − 0.20 in 2018). The pre-monsoon and monsoon season built a moderately negative LST-NDVI relationship. In the pre-monsoon season, the correlation coefficient values of LST-NDVI relationship were − 0.40 (1991), − 0.51 (2004), and − 0.45 (2018). In post-monsoon season, these correlation coefficient values were − 0.48 in 1991, − 0.41 in 2004, and − 0.47 in 2018. The post-monsoon season built a stable and strong negative correlation. The correlation coefficient values of LST-NDVI relationship were − 0.63 for all the 3 years. Hence, the post-monsoon season reveals the best correlation among the four seasons. It was mainly due to the high intensity of moisture and chlorophyll content in green vegetation. Dry atmosphere reduces the strength of correlation, whereas the wet seasons (post-monsoon and monsoon) enhance the strength of correlation.

Fig. 7
figure 7

Seasonal variation of LST-NDVI relationship on different types of LULC: a pre-monsoon b monsoon c post-monsoon d winter e overall

Full size image

Liang et al. [92], Ghobadi et al. [27], and Guha et al. [24] observed a negative LST-NDVI relationship in their study. A significant positive linear LST-NDVI relationship was observed between LST and NDVI [93]. In Shanghai City, Yue et al. [56] showed a negative LST-NDVI relationship and it varied on different LULC types. Sun and Kafatos [94] stated that LST-NDVI correlation was positive in winter season, while it was negative in summer season. This relationship was also negative in Mashhad, Iran [57]. The relationship was strong negative in Berlin City for any season [95]. This correlation tends to be more negative with the increase of surface moisture [96,97,98,99]. The present study also found that the LST-NDVI correlation is negative, irrespective of any season. The value of correlation coefficient is inversely related to the surface moisture content, i.e., negativity of the relationship increases with the increase of surface moisture content.

5 Conclusion

The present study analyzes the spatial, temporal, and seasonal relationship of LST and NDVI in a tropical city of India using 12 Landsat data sets of four different seasons (winter, pre-monsoon, monsoon, and post-monsoon) for 1991, 2004, and 2018. The mono-window algorithm was applied in deriving LST. In general, the results showed that LST is inversely related to NDVI, irrespective of any season. In the post-monsoon season, the relationship was strong negative (− 0.63), while it was found weak negative (− 0.17) in winter. A moderate range of negativity (− 0.45) was noticed in pre-monsoon and monsoon season. The presence of healthy green plants and high moisture content in the air are the main responsible factors for high negativity. The LST-NDVI relationship varies for specific LULC types. The green area presents a strong negative (− 0.51) regression, while the built-up area and bare land presents a weak positive regression (0.14). The relationship is moderately positive (0.45) on water bodies. On vegetation, the LST-NDVI relationship was highly negative in the pre-monsoon (− 0.65), monsoon (− 0.52), and post-monsoon (− 0.58) seasons, while it was weak negative (− 0.28) in the winter season. The mean LST of the study area was increased by 5.16 °C during 1991–2018. The conversion of other lands into the built-up area and bare land influences a lot on the mean LST of the city. Both the changed and unchanged built-up area and bare land suffer from the increasing trend of LST. This result significantly presents the influence of climate change in Raipur City.