Impact of land cover changes on land surface temperature variations in Adama City, Ethiopia

In recent decades, human activities, particularly urbanization, have significantly altered the Earth's surface (Wang et al. 2022). The rapid growth of urban areas contributes to further urbanization, resulting in continuous and drastic changes in land use and land cover (LULC) that give rise to severe ecological problems (Samie et al. 2020). Industrial sectors' intensive greenhouse gas (GHG) emissions have contributed to the increase in Earth's surface temperature, while urbanization and other human-induced factors further exacerbate GHG emissions (Dey et al. 2023). One crucial factor in urban environmental issues is the land surface temperature (LST), which directly influences the urban heat island (UHI) effect (Valizadeh Kamran et al. 2017). The intensity of the UHI is associated with changes in LULCs. The accelerated urbanization process leads to dynamic changes in land use, land cover, and higher LSTs (Choudhury et al. 2019). Urban areas become hotter than their surrounding areas due to changes in land use, especially the transformation of green areas into buildings. These impermeable materials absorb and slowly release heat, resulting in the warming of the air above them (Board 2019). A study across different geographic regions has consistently demonstrated the impact of land use changes on LSTs. For instance, in Europe, a study focusing on several urban areas revealed that cities such as Madrid and Rome experience significant temperature increases due to urbanization, which reduces green spaces and increases heat retention (Mancini et al. 2017). In Asia, the rapid urbanization of the Yangtze River Delta in China has led to higher LST, particularly in areas where agricultural lands have been converted into urban environments (Zhou et al. 2016).

In developing countries, such as Africa, urbanization and climate change present exceptional challenges. One specific challenge is the UHI effect (Liet et al. 2021). The spatial pattern of LST in urban areas is influenced by landscape composition and configuration (Fu & Weng 2016; Monteiro et al. 2016). For instance, urban expansion in Nairobi, Kenya, has been associated with rising LST, especially in densely developed areas (Muthee et al. 2019). In Ethiopia, studies in cities such as Addis Ababa and Mekelle have also shown significant temperature increases linked to urban sprawl and land use changes, with expanding built-up areas LST (Kebede et al. 2018; Tesfaye et al. 2020). These studies underscore the global nature of the land use changes, emphasizing the need for context-specific strategies to address these impacts. Moreover in Ethiopia, unmanaged LULC change is a significant environmental problem that greatly impact urban development. Many regions in the country are experiencing a substantial increase in LST from year to year (Belete & Suryabhagavan 2019). Uncontrolled urban expansion due to rapid population growth and economic progress is the main cause. As a result of LULC change, urban areas have higher LST compared to rural areas (Kasa et al. 2011; Faqe Ibrahim 2017). In Adama City, urbanization is rapidly increasing, leading to challenges characterized by changes in LULC and subsequent variations in LST. Although previous studies have examined the relationship between land use change and LST in various cities across Ethiopia, they often did not adequately consider the significant role of vegetation cover in mitigating temperature. Studies by Warkaye et al. (2018) focused on the impacts of urbanization on microclimates but did not comprehensively assess how different levels of vegetation cover specifically influence temperature regulation. Other studies, such as that by Debie et al. (2022), demonstrated that urban green spaces can significantly reduce LST in Ethiopian cities such as Addis Ababa and Dire Dawa. However, these studies acknowledged limitations in their methodologies, such as a lack of high-resolution satellite imagery or insufficient temporal data, which affected the robustness of their analyses regarding the role of vegetation. In order to better understand the dynamics of vegetation cover and its critical function in regulating LST Ethiopian urban settings, further research employing advanced remote sensing techniques and spatial analysis is essential. The previous study did not pay enough attention to the significant role of urban green space. The specific degree and intensity to which land use and land cover (LULC) would affect urban LST and the abundance role of vegetation covers are still unidentified, which hinders the capacity to make exact endorsements and recommendations for urban land use optimization and landscape planning. This study aims to fill this gap by investigating the effect of LULC change on UHI and LST using geospatial techniques with the land use transfer matrix method. Since urban LST relies heavily on the physical or thermal properties of an object, it is crucial to examine the size, density, and distribution of green land cover in urban areas to determine their impact on the UHI and establish the relationship with LST. Therefore, this study serves as an indication for the Adama City Administration to plan the provision and management of urban green infrastructure. The objective is to create a favorable urban environment for residents, specifically by exploring the impact of urban land cover change on variations in LST.

To conduct this study, an explanatory research design with the application of quantitative approach was employed. This approach is particularly effective for identifying relationships between variables and understanding the underlying patterns within the data (Table 1).

To achieve the study objectives, diverse datasets were collected from different sources. Some of the data sources utilized during the research period included satellite images of two different years (2002 and 2022). The reason for choosing these years was to analyze the status of land cover with respect to LST distribution in the city. Primary data sources include observations and fieldwork. Observations were made to collect ground truth points for validation of LULC types, for further identification of each LULC in the study area. These were carried out in order to identify the LULC types such as built-up, open spaces, water bodies, and the spatial distribution of vegetation in the study area. Secondary data sources include reviewing different relevant published and unpublished literature related to the specific study in different areas. Necessary data, including physiographic data, were extracted from secondary data sources.

Image prepossessing included layer stacking, false color combinations, and image sub-setting. Image classification was used to convert image data into thematic data for meaningful information. Therefore, the satellite images were categorized into built-up areas, dense vegetation, sparse vegetation, water bodies, and bare land (open spaces).

The determination of the overall accuracy of satellite image categorization in comparison to the actual situation is accomplished by accuracy evaluation, which is a critical validation technique (Congalton & Green 2009). Therefore, in this work, a confusion matrix was carried out utilizing a point-based validation technique to calculate the classification's accuracy. An error matrix was used to evaluate the categorized maps and compare them to the ground truth and referenced data.

In the confusion/error matrix, there are rows and columns. The rows represent the categorization values, while the columns represent the facts from the field. The diagonal line of the error matrix represents the number of pixels successfully identified. The overall accuracy index is determined by dividing the total number of pixels in the matrix by the number of pixels that have been correctly classified (Fung & Le Drew 1988), as shown in Equation (1).

The LST of the study area was calculated using thermal bands from Landsat images of ETM+ 2002 and TIRS 2022. The LST was extracted from Landsat 7 and 8 images using a single channel and the mono-window approach (Qin et al. 2011). Band 6 for Landsat 7 and band 10 for Landsat 8 were employed in the retrieval. The brightness temperature of one band of TIRS as well as the mean and difference in land surface emissivity are used to calculate LST (Abulibdeh 2021).

Step I: conversion of DN into radiance

Step II: conversion to brightness temperature

Step III: land surface emissivity estimation

The main goal of this phase was to validate the LST results from Landsat data by comparing them with ground-based measurements. To do this, three sets of site temperature data from meteorological stations were used. We focused on data from February 2002 and 2022, the same months as the Landsat imagery acquisition. Temperature data from seven meteorological stations near the study area were utilized for this comparison. The close alignment between the satellite-derived LST and ground-based temperatures confirmed the accuracy and reliability of the LST results and validated the methodology employed.

The primary goal of the LULC change analysis in this study is to ascertain how current and historical human activity has changed the landscape's composition and layout. The study discovered that throughout the course of 20 years, Adama City has noticed variations in LULC at a distinct rate. Therefore, built-up, open space, dense vegetation, sparse vegetation, and water bodies were identified as major LULC classes in the study area. According to the LULC results for 2002, the city had a built-up area that was 67.13 km² (48.58%) in size and a dense and sparse vegetation cover that was 66.24 km² (47.95%) in size.

Matiwos (2018) noted that there is a strong propensity for the conversion of LU and LC into built-up regions due to the urban area's quick growth and expansion. Also, Aiymeku et al. (2024) noted that in recent years, cities have faced significant challenges in terms of sustainability as a result of rapid urbanization and indelible worldwide environmental changes. This shows that during the urban development process, built-up areas absorbed a sizeable quantity of land from vegetation cover, which caused a rise and amplification in LST.

The findings of this study are supported by similar observations in other regions. For instance, Kebede et al. (2018) reported a significant reduction in green spaces in Addis Ababa, Ethiopia, where the built-up area expanded considerably at the expense of vegetation cover, leading to higher LSTs. Similarly, Zhou et al. (2016) found that rapid urbanization in China's Yangtze River Delta resulted in the conversion of agricultural and forested lands into urban areas, which significantly increased local temperatures. These studies highlight the consistent pattern of urban expansion leading to a reduction in vegetation and a subsequent rise in LSTs across different geographic locations.

Moreover, Seto et al. (2011) emphasized that urban growth typically leads to the loss of natural landscapes, contributing to the UHI effect. Their studies demonstrated that cities experiencing rapid expansion often see a parallel increase in LSTs due to the replacement of natural vegetation with impervious surfaces. These findings are consistent with the results of the current study, reinforcing the conclusion that the conversion of vegetation cover to built-up areas is a key driver of rising LSTs.

According to the land conversion results, the amount of variation for the specified time periods between 2002 and 2022 indicates that the trend of the built-up area has significantly increased as a result of the suburbanization process. Table 3 demonstrates that with 30.97 km² of built-up areas, sparse vegetation is the land type that has been changed the greatest. According to the findings, one of the LULC kinds that altered positively was built-up area.

Table 3

LULC transfer matrix of 2002 and 2022 study period

View Large

According to Naikoo et al. (2020), significant changes in LULC have occurred as built-up areas have increased. According to Table 3, thick and sparse vegetation cover were extensively changed to built-up areas with an area of 11.5 km² (8.32%) and 30.97 km² (22.4%) from 2002 to 2022 from the total of LULC classes. This means that the city's built-up regions grew at the expense of its vegetation cover.

In order to comprehend the potential factors behind the area's changes in LULC, the study takes into account local perspectives. The area's identified causes of LULC change were population growth, economic development, governmental actions, and changes in institutional structure and policy. As a result, the informant study conducted by key informant interview (KII) and the Focus Group Discussion (FGD) participants saw LULC change in numerous locations throughout the city and sub-city. They also observed that the pattern of urban expansion in the city and its sub-cities is accelerating at an alarming rate. The elders have agreed that the sub-city is growing quickly and the city is expanding. They gave two explanations. The first is the nation's investment policy, which encourages people to make investments in the suburbs. As a result of increased investments, the sub-city's land usage and cover have worsened due to an increase in the built-up area of the city. In addition, the rural farming communities on the periphery of the sub-city have been exposed to displacement and dislocation by the city administration through large-scale expansion and renewal programs. The second reason mentioned by the FGD and KII informants was that since the land policies of the country were revised, local and international industry owners gave way to establishing their factories and industries in the sub-city.

According to FGD and KII's discussion, the main factors influencing the changes in the region's LULC are population dynamics (including size, growth, and migration) and the urbanization process. This is consistent with the findings of Ahlam (2017), who suggested that the rapid growth of built-up regions was a contributing factor to the alarming rate of population growth brought on by natural growth, rural–urban migrations, and urban–urban migrations.

In accordance with the satellite image results (Table 2), between 2002 and 2022, settlements or built-up areas were created on 17,99 km² of sparsely vegetated land. These data were further reinforced by observations of Adama City made from various perspectives while ground control points were being collected. As a result, Adama City's LULC was modified as a result of the alteration in the state of the national economy, which encouraged the expansion of urban land at the expense of vegetative land. In general, the findings of earlier research carried out in various sites around Ethiopia and the outcomes of satellite image were similar to the respondents' assessments of LULC change.

A map of the research area's thermal pattern distribution was made by dividing the LST distribution into permissible ranges and assigning each range a different color. The spatial distribution of LST in Adama City is shown in Figure 3(a) and 3(b) for the years 2002 and 2022, respectively. The results of this investigation show that the LST values of various LU and LC classes vary. The LST summary statistics for the specified years are shown in Table 4. A region that was 8.23 km² (5.95%) in size in 2002 and had LST values between 30 and 34 °C has grown to 38.97 km² (28.20%) by 2022. The rapid expansion of impermeable surfaces in the city was the main cause of the significant decline in this rating. Maximum LST values were found in both populated and unpopulated areas. This result reveals unequivocally that both built-up and open space are significant potential sources of the rise in surface temperature in the study area. Due to changes in LULC, the LST value ranges from 30 to 34°C increased from 8.23 to 38.97 km².

The mean LST of Adama City is accelerated by the high rate of activity in the industry, transportation, and power generation sectors. The findings concur with those of Kerr et al. (2004), Maitima et al. (2009), and Khin et al. (2012) that uncontrolled use of land resources and changes in LULC are the main causes of the rise in LST in both rural and urban areas.

The finding revealed that the average surface temperature of Adama City has increased from 28.25°C in 2002 to 31.78°C in 2022 at a rate of 1.77°C per decade. The study clearly shows that an increment in surface temperature is mainly due to the decrease in green spaces that are replaced by impervious surfaces. It was found that 67.12 km² (48.57%) of Adama City was covered by built-up areas in 2002 and 86.0 km² (62.24%) in 2022. More than half of the city is packed with man-made features, which make the temperature higher than in the surrounding regions. Table 5 shows the comparison between the LST results of 2002 and 2022. It reveals that all the LULC classes identified recorded an increase in the surface temperature over the study period. This result is in agreement with Varshney (2013) that the distinctive LST patterns are associated with the thermal characteristics of land cover classes. Alemayehu (2008) also showed the evidence for climate change including occurrence of drought, rising temperature, flood, reduced annual rainfall, and rising sea levels are resulted from LULC dynamics.

The northeast and southwest regions of the research area had greater NDVI values, according to the results of the NDVI for both 20002 and 2022. Additionally, dense residential areas with little vegetation cover showed low NDVI values. Table 6 compares the NDVI values from various years and reveals that there has been a noticeable change in vegetation cover over the sequence of the 20-year study period (Figure 9).

The findings of this investigation showed a statistically significant correlation between NDVI and LST. The study also revealed a negative association between NDVI and LST, indicating the necessity for urban greening and plans to increase vegetation cover in order to preserve the city's ecosystem and lessen the impact of the UHI. LST and NDVI readings were shown to be inversely correlated. This finding is consistent with those of several investigations. For instance, Feyisa et al. (2014) demonstrated how the presence of greenery can reduce air and surface temperatures by preventing land surfaces from being directly heated by sunlight. Other studies also confirmed that green spaces, whether parks, urban forests, or even small gardens, play a crucial role in mitigating the UHI effect, a phenomenon where cities experience significantly higher temperatures than surrounding rural areas. This is because these green spaces act as natural temperature regulators, effectively lowering both air and surface temperatures (Nolon 2012). The findings of this investigation are supported by additional literature that highlights the inverse relationship between NDVI and LST, emphasizing the critical role of vegetation in urban environments. For instance, Chen et al. (2015) found that urban areas with higher NDVI values tend to have lower LST, reinforcing the importance of vegetation in moderating urban temperatures. Similarly, Tamba et al. (2021) documented that the presence of green spaces within urban areas can significantly reduce surface temperatures, thereby mitigating the UHI effect. These studies, along with others, underscore the necessity of urban greening initiatives to manage rising temperatures in cities effectively.

The study aimed to investigate the impacts of land cover change on LST. The analysis revealed that urbanization is the primary driver of decreasing vegetation cover, significantly influencing land cover dynamics in the study area. The results showed that the built-up region, which represents the prospective land cover class (LULC), dominated the city. Over the two decades, of study period the built-up area has substantially increased compared to other land cover types, primarily due to rapid urban expansion. The development of residential areas and the loss of vegetative cover are the two most significant changes, which may be the primary cause of the UHI effect in Adama City and the resulted increase in LST. The NDVI in 2002, with a value > 0.2, decreased from 64.54 km² (46.71%) to 47.84 km² (34.62%) in 2022. It is interesting to note that the distribution of LST in Adama City is closely related to the distribution of NDVI and NDBI. The mean LST in the study area has significantly increased which could reduce urban thermal comfort and increase the risk of temperature. Accordingly, the study findings can be used to highlight the importance of developing urban green spaces to manage the increasing trend of LST and reduce the risk of temperature-related heat stress in the city. Therefore, the City administration, in collaboration with relevant agencies, should prioritize adherence to the national standards for green space coverage. Also, local communities should actively participate in protecting and enhancing the city's green spaces, fostering a sense of ownership and responsibility.

The authors acknowledge experts and the local community for providing the existing facilities to conduct this study. The authors also acknowledge Ethiopian Meteorological Agency. Additionally, the authors acknowledge Mattu University for the existing facilities to conduct the study.

Z.R.R. participated in research design, data analysis, and manuscript writing and final draft edition. T.G.A. and T.T.G. participated in data collection, literature review, and document analysis. All authors read and approved the final manuscript for publication.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.