Introduction

The most obvious outcome of human conversion of vegetative regions to built-up is fostered by urban growth. This is usually associated with population growth, alterations in land use, area expansion, and some anthropogenic activities (Halder et al. 2021). By projection, there will be over 70% increase in the number of urban dwellers by 2050 due to the present growth rate in urbanisation (Huang et al. 2019). This has culminated in increasing pressure on urban areas and their ecosystem services, e.g. vegetation, water and soil. These components have been substituted with manmade components. These components also include buildings, road and tarmac (Shahmohamadi et al. 2011), often exhibiting stronger thermal conductivities and higher radiation absorption rates in urban environments. The urban heat island phenomenon is a result of differences in land surface temperatures between cities and the rural areas that surround them around the world. This menace has been noted as one of the aftermaths of urban expansion (Ayanlade 2017).

The land surface temperature is a major factor in the energy balance in urban settings. At the lowest points of the atmosphere, it modifies the air’s temperature (Weng et al. 2014). Due to changes in biophysical elements like vegetation, impervious surfaces, and albedo, the shift of LST may also have an impact on the local weather and climate, as well as the net radiation, heat, and water balance (Mahmood et al. 2014). The majority of variations in urban land surface temperatures that have been recorded in recent years can be attributed to the change in urban land use that is observed in many cities, not only in industrialized countries but also in cities in Africa (Mukherjee & Singh 2020). A great deal of research on cities has been done in order to thoroughly evaluate the change in the land surface temperature (LST) produced by different urban land cover patterns. Some of the findings of the previous studies revealed that decadal reduction in vegetation is as result of changes urban land cover in and which subsequent led to an overall rise in the LST (Igun & Williams 2018). The understanding of the intensity of urban heat island (UHI) and its spatial differences in urban areas are essential to the comprehension of the human-environmental relation (Bizcarguenaga Doucet 2021). The combination of remote sensing with Geographic Information Systems (GIS), a widely used and recognized method, has made it possible to detect changes in urban land use and land cover (Aboelnour & Engel 2018). Satellite remote sensing gathers data that is multispectral, multiresolution, and multi-temporal in order to provide datasets for the coverage of urban areas.

In Oyo town, the rate of urban growth is being matched by the emergence of unplanned neighbourhoods with poor infrastructure and services as well as environmental sanitation issues (Idowu 2013). Large areas of cropland and virgin woodland are disappearing and being used for urban or related purposes in Oyo. The Oyo town's population has been rising throughout the years, which has resulted in more business activity. The reflective impervious structures like roads and buildings are now covering land that was formerly covered in greenery (Faragallah & Ragheb 2022). Along with the aforementioned problems, extensive land use changes brought on by urban expansion have been negatively influencing the environment, raising temperatures and changing precipitation. Given the increasing demand on the population and the anthropogenic activities that have contributed to significant environmental stress and ecological instability, concern over urbanization has become important. Land use and land cover in the region are dominated by built-up areas and agriculture (Jimoh et al. 2018). Most developed regions are made up of settlements, farmlands and small/medium scale businesses. This study is essential because Oyo’s temperature is fluctuating, and there hasn't been a thorough examination of urban land surface temperature (ULST) utilising local satellite imagery. Recent years have seen a frequent occurrence of enhanced high-temperature weather. The effects of this temperature fluctuation can be seen on the ecosystem's productivity and stability. To reduce the harmful effects of UHI, it is crucial to carry out this study to look at how urban growth affects temporal changes.

Materials and methods

Study area

Oyo town, the study area, is one of the State's larger towns and, in the hierarchy of Oyo State's settlements, is placed below Ibadan, the capital city. As the seat of the Oyo kingdom's government in the twelfth century, Oyo has a long history (Adeyeri 2019). The town is situated in the southwest of Nigeria, generally between latitudes 7° 47'N and 7° 55'N and longitudes 3° 54'E and 3° 59'E. Atiba, Oyo West, and Oyo East are Oyo Town's three local government areas. Oyo East Local Government has a population of 124,095, Oyo West Local Government has a population of 136,457, and Atiba Local Government has a population of 168,246. According to the National Population Census, which was carried out in 2006, the total population of Oyo town is 428,498 (Balogun 1986).

The region experiences a tropical savanna climate, which is known for its rainy and dry seasons as well as its high humidity. The usual range of annual rainfall is between 1000 and 1400 mm. The wet season typically lasts from April until October. The rainy season typically starts in the second or third week of March, finishes around mid-October, and is followed by the dry season, which lasts from November to March. The vegetation is typical of the Southern Guinea Savanna, frequently distinguished by a blend of grass and trees. It gets between 1000 and 1500 mm of rain every year, falling over six to eight months (Mueller-Dombois & Fosberg 1998). The economy of the area is mainly based on agriculture and craftwork. The town is famously known for its carved calabashes, leatherwork and mat-making. The people mainly trade farm produce such as maize, cassava and yams (Morton 2018).

Data acquisition and analysis

This study employed primary and secondary data. The primary data obtained for this study consisted of geographic coordinates of relevant features of the study area. The geographic coordinates were collected through a handheld Global Positioning System (GPS) receiver. The secondary data that was used in this study include both temperature and urban expansion related data which were sourced from Landsat. Landsat surface temperature data covering 31 years period were collected (1990 to 2021) to measure the urban expansion variability between 1990 and 2021. Besides, four multi-date images of the study area which include Landsat 7 ETM + , 1990, 2000, 2012 and Landsat 8 OLI/TIR, 2021 (Table 1) were selected. These images were collected within December and February periods of their respective years as these periods are dry season period and the percentage of cloud cover are relatively low during the periods (Egeru 2012). Different land use changes in the study area over the previous three decades were identified using the multi spectral images.

Table 1 Sources and characteristics of landsat dataset for Oyo Town
Full size table

Image classification

The classification of the multispectral satellite image was done using a modified version of Anderson’s (1976) Land Use/Land Cover (LULC) classifications by Tewabe and Fentahun (2020). Hence, the Landsat images were divided into five modified categories which comprise bare soil, water body, vegetation and built-up area were employed for the study. To determine which LULC courses are pertinent to the investigation, multiple LULC classes were classified. The images were also categorized using composite bands made up of bands 5, 4, and 3. The LULC layers from the previous were digitized for change detection. The ERDAS Imagine statistical tool was used to quantify this raster layer. Additionally, a maximum likelihood supervised classification algorithm was used to aid in choosing the pixels that best represent the LULC pertinent to the study. In order to determine urban expansion, a statistical analysis was conducted on the four years of satellite photos of the study area that were chosen at random. This analysis looked at how land use and land cover had changed over the previous 31 years.

Derivation of NDVI

Assessment of vegetation health was carried out with the aid of Normalized Difference Vegetation Index (NDVI). This analysis is noted as one of the vegetation indices usually employed for change detection on vegetation. The purpose of using NDVI is to determine the level of vulnerability and state of vegetation to anthropogenic activities (Li et al. 2020). Thus, NDVI was done using ERDAS Imagine, a raster based software. The electromagnetic spectrum's near infrared area is where healthy vegetation may be observed, which is why NDVI is related to vegetation. To account for variations in light and surface slope, the value is transformed to -1 = NDVI = 1.

The equation for the index is as follows:

$$\mathrm{NDVI }=\mathrm{ NIR }-\mathrm{ RED }/\mathrm{ NIR }+\mathrm{ RED}$$
(1)

Where, NIR and RED stand for the electromagnetic spectrum's reflectance in the near-infrared and red regions, respectively.

For Landsat 4–7,

$$\mathrm{NDVI }= (\mathrm{Band }4 -\mathrm{ Band }3) / (\mathrm{Band }4 +\mathrm{ Band }3)$$
(2)

Where: The reflectance bands for the near-infrared and red sections of the electromagnetic spectrum are Band4 and Band3, respectively. For Landsat 8,

$$\mathrm{NDVI }= (\mathrm{Band }5 -\mathrm{ Band }4) / (\mathrm{Band }5 +\mathrm{ Band }4)$$
(3)

Where: The reflectance bands for the near-infrared and red sections of the electromagnetic spectrum are Band5 and Band4, respectively.

Determination of land surface temperature

Thermal bands 6 (10.40–12.50 m), 6H (10.4–12.5 m), and 10 (10.60–11.19 m) of the TM, ETM + , and OLI–TIRS, with spatial resolutions of 120 m, 60 m, and 100 m, respectively, can be used to measure urban surface temperature changes. These thermal bands were utilized to reconstruct LST over the research area after resampling the spatial resolutions to a fixed resolution of 30 m for four separate periods (1990, 2000, 2012, and 2021). The following are the procedures used to determine LST:

  • Step 1: Using the following equation, Spectral Radiance (SR) for Digital Numbers (DNs) was converted.

    $${\mathrm{L}}_{\uplambda }= ({\mathrm{M}}_{\mathrm{L}}\times {\mathrm{Q}}_{\mathrm{cal}}) + {\mathrm{A}}_{\mathrm{L}}$$
    (4)

    Where Lλ denotes the spectral radiance at the sensor's aperture in Wm-2sr-1 m-1, QCal denotes the quantized calibrated pixel values in DNs, and ML and AL denote the band-specific rescaling factors identified as Radiance Multi Band X and Radiance Add Band X, respectively, in the Metadata file (approach provided by Chander and Markhan 2003).

  • Step 2: The effective at-sensor brightness temperature, often known as black body temperature (TB), was calculated from the spectral radiance using Plank’s inverse function.

    $$\mathrm{Tk}=\mathrm{K}2 /\mathrm{ ln}(\mathrm{K}1/{\mathrm{L}}_{\uplambda }+1)$$
    (5)

    where; Tk is the at-satellite brightness temperature (in Kelvin); K2 is the Wm-2sr-1 m-1 calibration constant2; K1 is the Wm-2sr-1 m-1 calibration constant1; and L is the Wm-2sr-1 m-1 spectral radiance at sensor.

  • Step 3: Land Surface emissivity estimation

Equation (v) is used to calculate temperature values connected to a blackbody. Due to the kind of land cover, spectral emissivity (ε) changes are therefore necessary. Each group of land use and land cover receives an emissivity rating according to the Snyder et al. (1998) emissivity categorization technique. This study takes into account a number of surfaces, such as bare soil, water surfaces, agriculture, and vegetation. In order to account for the quantity of vegetation present in each pixel (fv) as well as the cavity effect brought on by surface roughness (c), equation vi was used to calculate the emissivity of a heterogeneous surface (mixed pixels). Equation 6 was used to calculate the emissivity of a heterogeneous surface (mixed pixels), which took into account both the cavity effect brought on by surface roughness and the amount of vegetation present in each pixel (fv) (Sobrino et al. 2004). Equation 7 was used to determine the parameter fv (C).

$$\upvarepsilon ={\upvarepsilon }_{\mathrm{V}}\times {\mathrm{f}}_{\mathrm{V}}+{\upvarepsilon }_{\mathrm{s}}\times (1-{\mathrm{f}}_{\mathrm{V}})+{\mathrm{C}}_{\uplambda }$$
(6)
$${\mathrm{f}}_{\mathrm{V}}=[\mathrm{NDVI}-{\mathrm{NDVI}}_{\mathrm{s}} /\mathrm{ NDV}{\mathrm{I}}_{\mathrm{v}}-{\mathrm{NDVI}}_{\mathrm{s}}{]}^{2}$$
(7)
$${\mathrm{C}}_{\uplambda }=(1-{\upvarepsilon }_{\mathrm{s}})\times {\upvarepsilon }_{\mathrm{V}}\times \mathrm{F{\prime}}\times (1-{\mathrm{f}}_{\mathrm{V}})$$
(8)

Where:

NDVIs and NDVIv are the Normalized Difference Vegetation Index Threshold values for soil pixels (NDVIs = 0.2) and pixels of maximum vegetation (NDVIv = 0.5), respectively, according to Sobrino et al. (1990) and Sobrino et al. (2008); εs and εv are the emissivity and full vegetation pixels with mean values of 0.97 and 0.99, respectively (Sobrino et al. 2004), as computed in Eqs. (2).

The final expression for land surface emissivity, after the 3 steps, is given by:

$$\upvarepsilon =0.004{\mathrm{f}}_{\mathrm{V}}+0.986$$
(9)

  • Step 4: Land surface temperature was extracted using the following equation:

$$\mathrm{LST}={\mathrm{T}}_{\mathrm{B}} / 1+(\uplambda \times {\mathrm{T}}_{\mathrm{B}}/\uprho )\times \mathrm{ln\varepsilon }$$
(10)

Where:

LST is land surface temperature (in Kelvin);

TB is at-satellite brightness temperature (in Kelvin);

λ is the central wavelength of emitted radiance (11.45 μm for TM and ETM + ; 10.90 μm for OLI-TIRS);

ρ is h × c/σ (1.438 × 10−2mK or 14380 µmK);

h is Planck’s constant (6.26 × 10−34Js);

c is the velocity of light (2.998 × 108 m/sec);

σ is Stefan Boltzmann’s constant (1.38 × 10−23JK−1); and.

ε is the land surface emissivity.

Step 5: LST was converted into degrees Celsius (Tc) using the equation

$${\mathrm{T}}_{\mathrm{C}}=\mathrm{ LST }- 273.15$$
(11)

where: Tc is LST in degrees Celsius.

Determination of Urban Heat Island (UHI)

Maps of the land surface temperature in the research area was overlaid with specific positions acquired using the Global Positioning System (GPS). Using ArcGIS 10.3, the temperature data for each of the sites (both urban and rural and suburban) were then extracted. To compute UHI, the formula shown below was utilized.:

$$\mathrm{UHI }= {\mathrm{T}}_{\mathrm{u}} - {\mathrm{T}}_{\mathrm{r}}$$
(12)

where: Tu and Tr are land surface temperature of urban and rural locations respectively.

Accuracy assessment and image validation

The accuracy assessment and image validation were done by conducting validation exercises on the study. To begin with the validation, errors of commission or false-positives, ground truthing (visual inspection) of the forest reserve for the entire periods of 1990 to 2021 were identified. All pixels that represent errors of commission were summed to estimate the errors within each change periods (1990–2000, 2000–2012, 2012–2021). Also, supervised classification of the study area was carried out through the employment of visual interpretation of the multispectral images. The sections of the imageries that represent various defined classes of LULC were subjected to an on-screen digitizing of the homogenous pixels in order to create the vector layer for a particular LULC class. These sampled sites were then utilized to produce signatures for further processes. Hence, each of the classified Landsat data were disaggregated into different LULC using attribute tool in ArcGIS 10.8, and maps representing dominant LULC class per period were derived. Finally, overall accuracy of the classified Landsat data for 1990, 2000, 2012 and 2021 were estimated.

Results and discussions

landuse/landcover change pattern in the study area between 1990 and 2021

The total land area for Oyo is about 2,414.25km2. From the results in Table 2, in 1990, the built-up was barely 14.07km2 which increase to 19.83km2 in 2000, 35.67 km2 in 2012 and 44.84km2 in 2021. Natural vegetation in Oyo decreased from 1,753.42km2 in 1990 to 1,483.42km2 in 2021. Cultivation also increased from 589.12km2 to 870.62km2 in 2021. The summary of the landuse/land cover is as shown in Table 2 while the landuse/land cover classification for the four epochs is shown in Fig. 1. It could be observed as at 2021 that some of the land area that were occupied by natural vegetation have been replaced mostly by built up areas and cultivated farm lands due to urban expansion in Oyo. The bare surface area observed to be 56.42 km2 in 1990 was almost lost in 2021 to be 0.003 km2. The water body was 1.22km2 in 1990 and maintained the same size as 17.38 km2 in year 2000 and 2012 but decreased to 15.37 km2in year 2021.

Table 2 Landuse/land cover change in Oyo from 1990 to 2021
Full size table
Fig. 1
figure 1

LULC classification of Oyo town for 1990, 2000, 2012 and 2021

Full size image

Change detection in landuse between 1990 and 2021

A total of 2,342.54 km2 of the study area’s land is covered by vegetation and agriculture, or 97.03% of the total land area. Waterbodies make up the least amount of land cover; they occupy 1.22 km2, or 0.05% of the study region’s total land area, whereas built-up areas occupy 14.07 km2, or 0.58% of the total land area, and bare ground occupies 56.42 km2, or 2.33% (Jimoh et al. 2018). It was observed that the percentages of change in the various classes of the LULC is as follows; in year 2000, an increase in the built areas from 19.83 km2 to 35.67 km2 in 2012, and 44.84 km2 in 2021, which implies an increase from 14.93% to 79.87%, and 25.71% Whereas, vegetation land cover decreases from -0.11% to -3.01.48% and-12.75% in the period of study. It was also observed that cultivation increased from 4.79% to 6.69% and 24.52% from 2000 to 2021, whereas bare surface was constantly on the decrease from -82.93% to -64.79 to -41.89% over the year of study. Finally, the water body was at a point increased and then decreased. The percentage of change was calculated using: [ (Previous year – Initial year)/ Initial year] × 100.

The urban growth in the area is represented in Fig. 2. The growth in built up as witnessed over the period of study from 1990 (14.07 km2) to 2021 (44.84 km2) was 30.77km2. The south, north-east, and north-western regions of the research area all experienced growth. The changes in urban heat that were seen during the research period were caused by this urban expansion. The built-up area doubled in size during a 12-year period, or 11.36%, during the third stage of urbanization (2002 to 2014), which is when the highest rate of urban growth was observed. The following two urbanization periods, from 1990 to 2002 and 1984 to 1990, respectively, are followed by 5.06% and 3.92%. According to this, the research area’s urbanization rate increased more quickly between 2002 and 2014 than it did during the preceding two time periods. Due to wage increases from the federal and state governments to civil servants during this time, who essentially dominated Oyo Town, workers' take-home pay and personal savings increased. The economy may have improved as a result of this, or because Oyo Town was largely populated by civil servants during this time.

Fig. 2
figure 2

Change detection in Oyo (1990–2021)

Full size image

Estimated Land Surface Temperature (LST) pattern of Oyo between 1990 and 2021

The extent, temporal pattern, and distribution of temperature in the research area were determined by examining the spatial distribution of Land surface temperature (LST) (Fig. 3). The LST distribution was created using the thermal bands from Landsat TM, ETM + , and OLI, as was already indicated. Attempt was made to use imageries of same season for temperature mapping, since temperature of the area is a function of season. Each Landsat image's LST spatial pattern was plotted and observed. As discovered in the study, there were changes in the spatial and temporal distribution of LST in the study area between 1990 and 2021, as shown in Fig. 3. The map legends indicate the minimum and maximum LST of the areas of study as well as the visible changes observed in terms of increase in the spread of the UHI as a result of constant increase in the built up areas and cultivation which implies that there is constant reduction in the vegetal cover. High-density urban areas were where the hot spots were concentrated. By visually examining the LST maps, we found that the area in the study area with the highest temperature had a smaller spatial extent in 1990 compared to that of 2000, 2012 and was greater in 2021.

Fig. 3
figure 3

Distribution of land surface temperature variation in Oyo from 1990 – 2021

Full size image

In 1990, it was generally noted that the central region of Oyo was substantially warmer than the northern and southern regions of the city. It indicates that during this time, anthropogenic influences like urbanization and agricultural development greatly changed Oyo's natural surface conditions and atmospheric qualities, resulting in various heat patterns inside the city. Temperatures as low as approximately 13.64 °C were observed in some locations around the north western parts of the city in 1990 with significant land surface temperature changes as high as 36.85 °C in 2000. The maximum temperature value of about 37.18 °C was observed in 2021 compared to the maximum temperature values in 2000 and 2012, with an average temperature of 30.94 °C. The temperature as observed in the study area in the years under study was constantly increasing. The variable thermal characteristics of the various land cover types prevalent in the city and their evolution over time may be to blame for the differences in surface temperature. The changes in Oyo land cover are most likely explained by increased human activity as a result of the city's rapid population expansion as reported by Jimoh et al. (2018). The result of this study is to the study by Fashae et al (2020) that found surface temperature trends to be upward both in the high and low density residential areas. Similarly, Stemn (2013) reported a high change in surface temperature that is associated with urban expansion while vegetated areas was found to have was found to have a little change in surface temperature.

Relationship between land surface temperature and landuse/landcover

The Pearson correlation between LST and land use and land cover was calculated using the zonal statistics of the images captured in Oyo during the course of the different years and the results shown in Table 3. The four different LULC types and five classes in Oyo have varying LST during the course of the study years. The study area’s bare and built-up surfaces had temperatures between 23.63 °C and 24.180C when the LST map was overlaid on the LULC, but the built-up surfaces had mean temperatures between 35.05 °C and 36.16 °C between 1990 and 2021, which implied a high temperature according to the zonal statistics interpretation. The mean temperature for cultivation ranges from 32.260C to 36.04 °C, with a maximum temperature as high as 37.18 °C in 2021. However, in 2000, plant and aquatic bodies had similar temperature changes over the research period, ranging from 11.21 °C to 26.05 °C. The land surface temperature (LST) has been determined to have a very strong positive link with built up by relating the land surface temperature and land use parameter using zonal statistics. The lowest temperatures are seen near water bodies and in vegetated areas, according to LST pictures, while the greatest temperatures are found in urban areas.

Table 3 Relationship between LULC and estimated average LST in Oyo (1990–2021)
Full size table

This result is consistent with those of Muthamilselvan (2018) that in which areas with settlements and sand were equally reported to have higher temperatures, whereas areas with water bodies and vegetation had lower temperatures. Within a 13-year period, the temporal data have demonstrated considerable variances in temperature changes of roughly 2.32° C. Fashae et al (2020) also noted that the spatio-temporal variation of temperature result indicated that temperature had risen over time, with the Urban Heat Intensity (UHI) appearing to be at its peak in high density settlement areas with intensity values of about 4.9 °C and the least values of intensity at the vegetated and waterbody areas being observed to be around 3.0 °C.

Relationship between temperature and change in landuse

The results of the correlation analysis performed are shown in Table 4. The result showed that vegetation has a very strong significant impact on LST, with a relationship value of -0.99 for the average temperature and -0.90 and -0.66 for the minimum and maximum temperatures, respectively. This effect is considered negative since it clearly demonstrates how flora is constantly declining, which raises temperature as a result. Furthermore, there is a significant association between temperature and cultivation, with relationship values for the minimum, maximum, and average temperatures being 0.82, 0.86, and 0.10, respectively. This connection can be explained by the increase in cultivation, which causes an increase in temperature. With relationship values for the minimum, maximum, and average temperatures of 0.97, 0.10, and 0.77, respectively, the built-up area also exhibits a strong correlation with temperature. This relationship can be attributed to the continuous growth of built-up areas, which in turn leads to an increase in temperature.

Table 4 Relationship between temperature and landuse change
Full size table

Normalized difference vegetation index result

The quantity of biomass or vegetation present in the environment is measured by NDVI. The NDVI scale measures how green and healthy the vegetation is, hence a greater NDVI indicates more greenery (Boyd et al. 1996). NDVI is a reliable gauge of surface temperature, hence it has a negative correlation with surface temperature, according to Xian and Crane (2006).

Four separate image data sets from the study area, taken between 1990 and 2021, are used in the research's NDVI analysis, allowing us to see how the vegetation in the area varies over time. Given that the vegetation in the area depends on the season, the research attempted to collect and use images at the same time for NDVI analysis. Figure 4, which display images that have been mapped to show the spatial and temporal distribution of vegetation cover in the research region, are shown. The range of NDVI is -1.0 to 0.71 in 1990 and –0.06 to 0.04 in 2000,-0.09 to 0.067 in 2012 and finally -0.08 to 0.016 in 2021. It was observed that, vegetation cover is higher in 1990 with max NDVI value of 0.71. However, in 2021, less NDVI values were recorded. The growth of the built-up areas was the only factor contributing to lower NDVI values in 2021. Furthermore, it was found that places with lower NDVI values correspond to developed areas and water bodies, whereas areas with higher NDVI values correspond to vegetated areas and less developed natural surfaces. The lowest NDVI value was recorded in 2021, while the greatest NDVI value was recorded in 1990, showing that the NDVI drops off as LST rises. The shift in land use has a significant impact on the region's temperature regime, as shown by (Fig. 4). In contrast to other places, such as populated areas, the vegetative body has a lower temperature. In comparison to urban green land like agricultural areas, LST values are comparatively greater in built-up and bare soil areas with no vegetation cover.

Fig. 4
figure 4

NDVI in Oyo from 1990 – 2021

Full size image

Lower LSTs are usually seen in areas with high NDVI, and upper LSTs typically indicate the opposite of this situation, according to scientific literature; this could be plainly shown in Lagos City through the urban–rural gradient analysis (Obiefuna et al. 2018). This substantiates the well-known prognosis that built-up areas play a significant role in the spread of LST. According to Obiefuna et al. (2021), the built-up area or urban expansion, which has expanded by more than 770% since 1984 and caused an increase in the mean LST over Lagos from 28.60 °C in 1984 to 30.76 °C in 2019, is the main source of land cover change.

Relationship between LULC, LST and NDVI

Zonal statistics were employed in this study to establish the relationship between LST and vegetation availability. Because most of the research region is in the moderate to low criticality range, there is only a very moderate association between vegetation availability and LST. The NDVI has been found to have a negative relationship with surface temperature. This is a clear illustration of the high and unfavorable association between the LST and NDVI results juxtaposed in (Fig. 5a and b). As a result, locations with less vegetation have higher land surface temperatures. Figure 5 shows that, throughout all years, urban regions had the lowest NDVI values whereas non-urban (vegetated/rural) areas had the greatest NDVI values. Again, in this study, it was discovered that NDVI and LST were highly associated across all types of land cover, particularly in the study area’s urban and vegetated/rural areas. Putting the images side by side could make it easier to see how these variables are related, allowing one to examine how LST, NDVI, and LULC in a particular area are correlated as well as how they differ.

Fig. 5
figure 5

a Relationship between landuse, LST and Vegetation for 1990 and 2000. b Relationship between landuse, LST and Vegetation for 2012 and 2021

Full size image

This was supported by Stemn (2013), whose research showed a negative correlation between NDVI and surface temperature for all three types of land cover taken into account. Consequently, a land cover's radiative surface temperature would decrease as its biomass content increased. Changes in land use and cover also indirectly affect surface temperature due to the relationship between NDVI and surface temperature.

Conclusion

The patterns and distributions of land surface temperature over time in Oyo have revealed that the city's land surface temperature is rising and that the expanding built-up area's density is having a significant impact on it. The study also found that most places with lower temperatures in 1990 were transformed into hot spots in 2000, 2012, and 2021 as a result of an increase in built-up areas and a corresponding decrease in vegetation over time. An area's land surface temperature may alter due to changes in land use and land cover. It also showed an association between LULC and land surface temperature as well as an association between NDVI and land surface temperature in urban areas. In addition, the study recommends using GIS techniques and remote sensing data to map and simulate variations in land surface temperature in metropolitan areas in relation to various forms of land cover and a vegetation index (Shatnawi & Abu Qdais 2019). It was found that there is only a very weak negative association between land surface temperature and vegetation cover using zonal statistics to relate these two variables. Due to the built-up region's high percentage of low-to-non-criticality areas, the association is likewise not significant. On the basis of this information, it has been found that the amount of vegetation has little effect on the temperature of the metropolis. Although the bulk of the built-up area in the study area is within the upper range of LST, there is a positive correlation between LST and land use and land cover.

The study demonstrated the adaptability of remote sensing data when used in conjunction with GIS approaches for mapping and modelling LSTs of various land use types in Oyo from 1990 to 2021 through the use of remotely sensed data to investigate the regional distribution of urban surface temperature in relation to various land-cover types and NDVI. This study revealed that the radiant surface temperature was significantly affected by urban expansion. The study also found that the study area's surface radiation temperature had increased as a result of urbanization. From the findings of this study, the LST recorded for the study area is yet to reach a critical level. However, efforts should be taken to maintain effective land use management and to promote the practice of "urban greenness," especially in metropolitan areas, in order to mitigate the effects of increase in LST and its effects on urban residents.