The study area, Abuja Municipal Area Council, is situated between latitudes 8˚37'41'' and 9˚9'15'' north of the equator, and longitudes 7˚3'55'' and 7˚34' east of the Greenwich Meridian. Abuja Municipal, being the Capital City, covers an area of approximately 1456 km². The area contains the following districts and satellite towns; Central Business District, Maitama, Asokoro, Wuse, Kubwa, Lugbe etc. as depicted in Figure 1.

The study area has a projected population of 3,564,126 persons, being the fourth densely inhabited area in Nigeria [16]. Under Koppen climate classification (Aw), the study area has a tropical wet and dry climate, experiences three weather conditions annually, which include a warm, humid rainy season and a scorching dry season with a brief interlude of harmattan associated with dust haze, intensified coldness and dryness in between the two. The study area records an annual rainfall of about 1631.7 mm, highest within the FCC [17]. It records relative humidity in the dry season which goes as high as (20%) in the afternoons at the northern high elevations and about (30%) in the extreme south. During the rainy season, relative humidity rises as much as (50%) [17] The dominant vegetation of the area is classified into these savannah types, park/grassy savannah, savannah woodland, and shrub savannah, with the grassy savannah being the most dominant class. The soil is basically alluvial and luvisols making it a fertile ground for agriculture and vegetation growth. Abuja has witnessed a large influx of people into the city, this unprecedented increase has led to the emergence of satellite towns and smaller settlements to accommodate this increased populace [16]. This increase in the physical boundaries implies a corresponding loss of vegetation and land in the area thereby a direct impact on the micro-climate.

The LANDSAT data were downloaded from USGS Earth Explorer, in 2017. The Thematic Mapper (TM) images were downloaded on 26^th Dec., 1986. The Enhance Thematic Mapper plus (ETM+) images were downloaded on 27^th Dec., 2001 and the Operational Land Imager (OLI) on 28^th Dec., 2016. The intervals of +15 years were deliberately chosen by the researcher to ensure uniformity between the datasets. The Landsat satellite data have 30 m spatial resolutions, the TM/ETM+ images have spectral range of 0.45 - 2.35 micrometer (µm) with bands 1 to 7 and 8 respectively while the Operational Land Imager (OLI) extends to band 12. They were used for image classification and LST extraction. The administrative maps of Nigeria containing states and LGA’s were gotten from the National Space Research and Development Agency (NASRDA). It is a projected vector shape file, that was used to specify the boundary of the study area.

The mono-window algorithm method is adopted to retrieve the LST from the imageries selected for this study. The Landsat-5 TM thermal bands 6 (10.40 - 12.50 μm), ETM+ bands 6L (10.4 - 12.5 μm) and TIRS 10 and 11 (10.60 - 11.19 μm) have a spatial resolution of 30 m respectively which is considered suitable as shown by many literatures for capturing the multifaceted intra-urban temperature differences thus making it effective for urban climate analysis. The Landsat ETM+ sensor, images of the thermal band are taken twice: one in the low-gain mode (band 6L) and the other in the high-gain mode (band 6H). Band 6L is used to image surfaces with high brightness, band 6H is for low brightness. Band 6L was used in this study, due to errors contained in the 6H band. Consequently, the LANDSAT thermal bands were used to retrieve LST over the study area for the three different periods (1986, 2001, and 2016) using various procedures which range from radiometric calibration, conversion of DN to radiance, correction for atmospheric absorption, re-emission and surface emissivity which has been used in [18] as described below:

Conversion of Digital Numbers (DN) of the bands to Spectral Radiance [18]

L λ = ⌊ L MAX − L MIN Q Calmax − Q CALMIN ⌋ × ( DN − 1 ) + L MIN (1)

where:

L_MAX = the spectral radiance that is scaled to Q_CALMAX in W/(m²∙sr∙μm) L_MIN = the spectral radiance that is scaled to Q_CALMIN in W/(m²∙sr∙μm) Q_CALMAX = the maximum quantized calibrated pixel value (corresponding to L_MAX) in DN = 255 Q_CALMIN = the minimum quantized calibrated pixel value (corresponding to L_MIN) in DN = 1.

Conversion from Spectral Radiance to At-Satellite Brightness Temperature [18]

T = K 2 ln ( K 1 L λ + 1 ) − 273.15 (2)

where:

T = At-satellite brightness temperature, L_l = Spectral radiance (gotten from Equations (1) and (2)), K₁ = Band specific thermal conversion constant from the metadata, x is the thermal band number, K₂ = Band specific thermal conversion constant from the metadata, −273.15 = Constant for conversion from Kelvin to Degrees Celsius as shown in [16].

Correcting for Land Surface Emissivity (LSE) [18]

The temperature values obtained using Equation (2) are reference to a blackbody. Therefore, corrections for spectral emissivity (ε) became necessary according to the nature of land cover (Equation (3))

e = 0.004 P V + 0.986 (3)

where, e = Land Surface Emissivity, 0.004 & 0.986 = Constants for emissivity estimation, P_V = Proportion of vegetation [16] given by the equation

P V ( NDVI − NDVI min NDVI max − NDVI min ) (4)

where, NDVI = Normalized Differential Vegetation Index as computed with Equation (1) for each of the years, NDVI_min = Minimum value of NDVI for that year, NDVI_max = Maximum value of NDVI for that year [9].

Estimation of the Land Surface Temperature [18]

LST = B T 1 + W × B T P × ln ( ∑ ) (5)

where: LST = Land Surface Temperature, B_T = At-satellite brightness temperature, W = Wavelength of emitted radiance (µm) [16] given as:

P = h × c s ( 1.438 × 10 − 2 m ⋅ K ) = 14380 (6)

h = Planck’s constant (6.626 × 10⁻³⁴ J∙s), S = Boltzmann constant (1.38 × 10⁻²³ J/K), C = Velocity of light (2.998 × 10⁸ m/s), e = LSE.

Points representing ten rural and ten urban areas in Abuja Municipal were generated, and the temperature readings for each location were extracted using the “Extract Multi Values to Point” tool in ArcGIS 10.4. Then, the urban heat island was assessed using the equation:

UHI = T U − T R (7)

where:

UHI = Urban Heat Intensity;

T_U = Temperature at urban station;

T_R = Temperature at rural station.

Students “t” Test was carried out to determine the significant difference between the urban and rural temperature values and test of hypothesis.

The t-test is used to determine if two sets of data are significantly different from each other. In this study, it is used to test the hypothesis and determine the significance of the difference between urban/rural temperatures. The independent unpaired samples t-test was implemented due to the independent and identically distributed samples of the rural and urban temperature values. The t statistic to test if the means are significantly different can be calculated as follows:

t = X ¯ 1 − X ¯ 2 S 1 2 N 1 + S 2 2 N 2 (8)

where: S₁ and S₂ = Standard deviation for N, N = Number of observations,

X ¯ 1 − X ¯ 2 = Standard error of the difference between the two means.

For significance testing, the degree of freedom for this test is given by:

N₁ + N₂ − 2, where N = Number of observations in each group.

Urban areas tend to have higher air temperatures than surrounding rural areas as a result of vegetation cover being replaced by non-porous, non-evaporating, highly thermal conductive surfaces such as concrete and asphalt [19] [20]. The UHII is determined as the spatially averaged temperature difference between an urban and its surrounding rural area [11]. In order to determine urban heat island intensity in Abuja Municipal, twenty transects representing urban and rural areas located in the area were chosen using satellite derived images of 1986, 2001 and 2016 respectively (Figure 2). These transects pass over various landscapes with different environmental settings, an inquiry into the Urban Heat Island characteristics of the profile will help to understand the factors shaping the city’s thermal landscape. The urban and rural transects selected are displayed in Table 1.

The temperature difference between the ten urban and ten rural stations based on the results of satellite derived imagery of 1986 (Figure 3) is displayed in Table 2.

Tables 2-5 and Figure 3 divulge the temperature characteristics over twenty selected locations in Abuja, FCT as derived from the satellite image of 1986, 2001 and 2016 (Figure 2). As revealed by the satellite derived image (Figure 2), a great variation was observed in the temperature distribution of the selected

locations for the years under study. Table 2 shows the average surface temperature and the differences between the 10 rural and urban points selected for 1986. In 1986, Gwarinpa was the urban area was associated with the highest temperature value of 26.7˚C and Kurunduma the rural area associated with the lowest temperature with a temperature value of 21.5˚C, the highest difference was observed between Maitama urban area and Kurunduma rural area with a temperature difference of 3.9˚C. The findings revealed the average temperature difference (UHII) to be 1.0˚C (Table 2).

For 2001, each of the locations experienced an increase in mean temperature with Gwarinpa and Kurunduma being attributed with the highest and lowest temperature of the year, with a temperature of 32.8˚C and 25.0˚C (Table 3). furthermore, though Gwarinpa urban area and Kurunduma rural area recorded the mean highest and lowest temperature, the area with the highest urban heat island intensity was between Maitama urban area and Kurunduma rural area with a temperature difference of 7.0˚C (Figure 3).

Looking at the temperature characteristics in 2016 (Table 4), there was also a significant increase in mean average temperature compared to 1986 and 2001 respectively. The highest temperature was observed in Area 11 (34.5˚C), an urban area, the lowest temperature was observed in Kurunduma (28.1˚C). The highest temperature difference of 3.7˚C in 2016 was observed between the central business district (CBD) and Wupa rural area (Table 4).

The hypothesis is stated as thus: “There is no significant difference in land surface temperature between the urban and rural stations selected.” The variables considered are the independent and equally the temperature distributed values of each of these locations. The different periods under study (1986, 2001 and 2016) were tested to determine the significant difference in the land surface temperature values to acknowledge the presence of a substantial UHI within the study area (Tables 5-7).

Table 5 represents the student “t” test summary of 1986. From the student “t” distribution table (Table 5), the critical value of “t” at 0.05 is 1.73. Since the calculated “t” value of 2.50 is higher than the critical value, the null hypothesis is rejected for 1986, thus, inferring a significant temperature difference between the urban and rural stations, indicating the presence of an urban heat island. Similarly, a critical value of 1.73 was observed in 2001 (Table 6). Since the calculated “t” value of 3.34 is higher than the critical value, the null hypothesis is rejected for 2001, thus, inferring a significant temperature difference between the urban and rural stations, indicating the presence of an urban heat island. The same trend was observed in 2016 as divulged in Table 7. The result as seen in Table 7 reveals the critical value of “t” at 0.05 to be 1.73. Since the calculated “t” value of 5.57 is higher than the critical value, the null hypothesis is rejected for 2016 as well, thus, indicating a significant temperature difference between the urban and rural stations, indicating the presence of an urban heat island.

The findings of this study reveal an increase in the surface temperature over the study area, which has been attributed to increased impervious surfaces, and loss of vegetative cover. During the study period, the mean land surface temperature increased by 4.9˚C from 23.5˚C to 28.4˚C, an increase that was due to high level of urbanization that went on during that period [15]. In 1986, the average temperature difference of 1.0˚C was observed between the urban areas and rural areas. The temperature differences are a major pointer to the low level of built up areas in the study area. By 2001, the urban areas gained significant development which caused loss of vegetative cover which consequently led to significant surface temperature increase of 2.2˚C in 2001 compared to 1986. Thus, an indicator that by this time, the development phases had hit full stride. It was also observed that some rural areas such as Rubuchi and Karmajiji had higher temperature readings than their counterparts in the urban areas (Asokoro and Garki) with readings of −0.4˚C and −1.3˚C. The highest difference observed between some rural areas can be attributed to the increase in spatial extent due to the influx of migrants [15]. Increase in the average temperature difference of 2.9˚C by 2016 has been pointed to the presence of a very strong urban heat island.