Potential Influence of Urban Sprawl and Changing Land Surface Temperature on Outdoor Thermal Comfort in Lagos State, Nigeria

Urbanisation, which is the gradual shift in the residence of human populations from rural to urban areas (UN 2018), is one of the most obtrusive human actions, and gives rise to irreversible effects on the earth’s surface and biosphere. These shifts in land use and territorial occupation of urban areas, as observed in Gómez et al. (2020), are topical issues at global, regional and national levels owing to their impacts on ecological and environmental systems, among others. The world has experienced its fastest rate of urbanisation, and this is especially true in developing countries (Chadchan, Shankor 2009). While 30% of the world’s population during the late 1950s lived in urban areas, 55% lived in cities in 2018, a proportion which is projected to increase to 68% by 2050 (UN 2014, 2018). Expectations for the future indicate that India, China and Nigeria will account for 35% of the projected growth in the world’s urban population between 2018 and 2050 (UN 2018). This growth will lead to urban expansion, an increase in demand for housing and urban infrastructure and increased pressure on environmental resources for human sustenance.

Studies have argued that urbanisation has accelerated the economic and social development of cities, globally turning them into engines of economic growth and centres of innovation, while influencing their surrounding hinterlands (De Sherbinin et al. 2007, Seto et al. 2011, Cui, Shi 2012, Ohwo, Abotutu 2015). However, urbanisation has had wide-ranging impacts on the environment, from the local scale to the global (Kim, Baik 2005, Zhao et al. 2006). Evidence suggests that it results in changes to land cover, modification of hydrologic and biogeochemical cycles, fragmentation of habitats, biodiversity reduction, alteration of local climate through modification of albedo and rapid loss of productive farmlands (Hahs et al. 2009, Radeloff et al. 2010, Seto et al. 2011). Also, urbanisation leads to increased air and water pollution, decreased water supply (Liu, Diamond 2005, Zhao et al. 2006, Cui, Shi 2012), insufficient housing, increased instances of impervious surfaces which have arisen consequent to reduction in natural vegetation, altered surface albedo and loss of ecosystem services, among others (Yuan 2008, Jago-on et al. 2009).

A notable effect of urban development is increase in temperature, particularly in densely built-up areas (Shi et al. 2015). Studies have indicated that through the phenomenon of urban heat island (UHI), urban areas are significantly warmer than their surrounding undisturbed areas and countryside. The phenomenon of UHI arises from decreased vegetation cover and from the fact that cities are constructed with highly impervious and absorptive materials such as concrete and asphalt. These materials absorb huge amounts of solar heat, thereby increasing the urban surface temperature. The increase in surface temperature leads to increased outdoor and indoor thermal discomfort, along with heat-related health risks (Uejio et al. 2011, Zhou et al. 2016, Mushore et al. 2017a, Tarawally et al. 2018). Land surface temperature (LST), which has been widely used to measure UHI, is a significant parameter in understanding the earth’s surface energy balance, surface physical and chemical processes and the exchange of surface matter. LST has thus featured in many environmental and climate models (Oguz 2013, Deng et al. 2018). All over the world, the intensity of human activities, coupled with fast-paced urbanisation, is leading to alterations of the land cover and, in effect, the LST. Land cover change, including variations in surface reflectance and topography, are important indicators that influence LST (Hou et al. 2010, Obiefuna et al. 2018, Alademomi et al. 2020). Due to the physical properties of the land surface, surface temperatures are typically higher in urban areas than in vegetated areas and in the countryside; this difference is attributable to their physical properties (Nse et al. 2020). One effect of urban expansion is the increase in LST. In consequence, the link between land cover and LST needs to be understood to appreciate the ecological effects of changes in LST. Also, this understanding is necessary for an inquiry into the potential influence of LST on urban outdoor thermal comfort and human well-being.

Thermal comfort can be defined as a measure of one’s physiological response to the human body’s heat balance under varying environmental conditions. According to Yilmaz (2007), cited in Mushore et al. (2017b), thermal comfort or discomfort is understood to occur when 80–90% of the respondents express dissatisfaction with the prevailing temperature at a given time and location. The usage of any outdoor location for activities and leisure is partly influenced by thermal discomfort, since urban dwellers usually enjoy leisure in the thermally comfortable and conducive outdoor settings of parks and open spaces (Goshayeshi et al. 2013, Setaih et al. 2014). Furthermore, variety in urban surfaces lends to the users’ exposure to varying levels of thermal comfort (Zhang et al. 2009), thereby affecting their physical and psychological performance. Outdoor thermal comfort or discomfort equally influences indoor thermal conditions (Mushore et al. 2017b). Being a subjective measure, the effects of the thermal environment on different people can vary widely, thereby making an assessment of varied users a complex task (Walls et al. 2015). Generally, scientific research on thermal comfort has focused on indoor/interior spaces; consequently, studies which have been undertaken so far on outdoor thermal comfort are fairly recent ones (Achour-Younsi, Kharrat 2016). In comparison with studies conducted in relation to the indoor environment, the difficulty in dealing with outdoor thermal comfort is due to the multiplicity of key variables or drivers, including climatic, personal and contributing factors that are involved. The key climatic variables affecting outdoor thermal comfort that are the most concrete and the most frequently considered include air temperature, solar exposure, mean radiant temperature, air velocity and humidity (Achour-Younsi, Kharrat 2016, Jendritzky et al. 2012, Johansson et al. 2014). Understanding the interaction between these climatic variables and their influence on outdoor thermal comfort represents a complex challenge to urban designers (Walls et al. 2015, Achour-Younsi, Kharrat 2016). Recently, thermal comfort indices, which are commonly applicable in the outdoor environment to assess bioclimatic conditions for human beings, are being deployed.

The universal thermal climate index (UTCI), which was developed by the International Society of Biometeorology, is seen as one of the most comprehensive indices for calculating heat stress in the outdoor environment (Zare et al. 2018). The reason for this view is that heat stress has the potential as a heat-related health risk indicator (Napoli et al. 2018). The UTCI is a bioclimatic index which indicates the physiological heat load or stress that the human body experiences in attempting to achieve thermal equilibrium with the surrounding outdoor environment (Błażejczyk et al. 2013, Napoli et al. 2018). The UTCI is defined as the air temperature in a reference condition or environment (of 50% humidity, still air and full shade) which results in the same strain index value as that of the reference individual’s response to the real environment (Walls et al. 2015, Zare et al. 2018). According to Napoli et al. (2018), for a given combination of air temperature, wind, radiation and humidity, the UTCI can be defined as the air temperature of a reference location/environment which would lead to the same UTCI Fiala multi-node model (Fiala et al. 2012) response (shivering, sweat production, skin blood flow and rectal, skin wittedness, temperatures, mean skin and facial) in the human body as the actual environment. The reference environment is taken as a condition of calm air or a wind speed of 0.5 m × s⁻¹ at 10 m above the ground, radiant temperature which is equal to air temperature with no additional thermal radiation and 50% relative humidity (RH). Whether it is computed from an energy balance model or the UTCI Fiala multi-node model (Fiala et al. 2012), the UTCI is applicable in all climates, seasons and scales, in addition to being independent of personal characteristics such as age, gender, specific activities and clothing (Jendritzky et al. 2012, Walls et al. 2015).

According to Błażejczyk et al. (2012), the UTCI represents specific climates, weather and locations around the world much better than certain other indices. Błażejczyk et al. (2012) found the UTCI to be very sensitive to changes in ambient stimuli, including that of temperature, solar radiation, wind and humidity. In-situ measurements of air temperature and empirical observations of other meteorological parameters are required for the calculation of these indices. Generally, satellite-derived LST has not been substituted for air temperature despite their strong relationship (Mutiibwa et al. 2015, Good et al. 2017), mainly because of the difference in what both parameters measure or represent. LST is a measure of how hot the land surface (including the uppermost parts of trees and buildings) is to touch, whereas weather stations measure the air temperature just above the surface at 2 m height (T_{2 m}; Good et al. 2017). However, as canvassed in Good et al. (2017), there are many parts of the globe, such as Africa (including Nigeria and Lagos State), where there are very few weather stations, thereby leading to gaps in temperature data. Good et al. (2017) also noted that earlier studies have confirmed that both LST and air temperature are generally well-coupled, with correlation coefficients being in the range of 0.6–0.8. LST and air temperature are most tightly coupled over highly vegetated surfaces when insolation is low at high latitude. In another study, Choi et al. (2020) assessed the correlation between air temperature and LST in some cities of South Korea. The result of their validation showed a high correlation of 0.9 between both parameters. Mutiibwa et al. (2015) had earlier argued that this relationship varies with atmospheric, seasonal and geographic changes. Nevertheless, they maintain the two remain coupled even in the presence of greater differences. This coupled relationship has therefore brought about an increase in studies which attempt to fill in the gaps in air temperature data sets with satellite LST data (Chen et al. 2015, Good 2015, Oyler et al. 2015, Parmentier et al. 2015, Janatian et al. 2016). Ultimately, the results in Good et al. (2017) presented a justification for increased use of LST data as an independent variable and to augment air temperature data in climate and weather science.

In a similar vein, Mushore et al. (2017b) canvassed persuasively for the use of Landsat medium-resolution thermal datasets to understand and map thermal comfort patterns in data-scarce cities of developing countries, where scarcity of in-situ observations may constrict such analysis. In their study of the effect of seasonal land cover changes on outdoor human thermal comfort in Harare, Zimbabwe, they concluded that Landsat 8 data detects seasonal land use/land cover and thermal discomfort changes with high accuracy. Similarly, in an earlier study, Wei-Wu et al. (2004), cited in Mushore et al. (2017b), determined outdoor thermal discomfort, with air temperature obtained from LST and RH derived from normalised difference vegetation index (NDVI), using Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper (ETM+) imageries. Their study indicated that outdoor thermal comfort/discomfort can be modelled as a function of air temperature and RH, both being derived from surface temperatures which are retrieved from thermal bands of Landsat. Along with Mushore et al. (2017b), some earlier studies (Cheng, Ng 2006, Widyasamratri et al. 2013, Polydoros, Cartalis 2014) have shown that LST retrieved from remote sensing is highly correlated with air temperature, thereby enabling estimation of air temperature from space-borne remote sensing of surface temperature.

In a previous study by Nwilo et al. (2012), a trend of changing LST within Lagos State, and in particular Lagos Metropolis, between 1984 and 2006 was established. Similarly, Obiefuna et al. (2018) followed up this study by tracking the dynamics of LST with land cover changes between 1984 and 2015 in Lagos Metropolis. Despite these studies, Ojeh et al. (2016) had earlier lamented the dearth of UHI studies both in tropical and Nigerian cities and emphasised particularly on the absence of consideration of UHI effects in the Lagos State Climate Change Policy document of 2012–2014. Also, they equally lamented that there were only a few studies that addressed the effects of synoptic observations of UHI with remote sensing data, in Africa’s most urbanised coastal city of Lagos and in Nigeria generally. With the influence of continuing global warming coupled with unrestrained urban sprawl, continuous monitoring which is aimed at generating reliable estimates of LST and land cover changes is crucial for appropriate management and policy actions at the state level. Such monitoring can ensure more orderly growth and well-being of residents. This study, therefore, updated the investigation of changes in LST and land cover between 1984 and 2019 to determine their current spatial distribution and the continued relationship between them. As part of the response to the call for investigation of the effects of UHI on the thermal comfort and health of Lagos inhabitants by Ojeh et al. (2016), this study also inquired into the possible influence of changing LST on the outdoor thermal comfort of inhabitants of the state and metropolis.

We observe that the main driver of land cover change is built-up area or urban development which had grown by over 770% of the area covered in 1984. The reduction of the mixed forests and wetlands by 665 km² and about 116 km², respectively, is a further confirmation that urban growth has displaced areas which contain ecological assets. The slight decline in the area of water bodies is probably the result of various reclamation projects in the state which commenced in 1984, including the Majidun Foreshore and Banana Island projects.

Further analysis of land cover changes in LGAs (Table 2) shows five LGAs as the focal points of expansion of urban growth together with their corresponding areas of decline in ecological assets. Ikorodu LGA had the highest growth with over 179 km² of built-up areas and a phenomenal loss in mixed forests by 189 km² during 1984–2019. In the same period, this was closely followed by Alimosho LGA, with an increase of about 146 km² in built-up areas and a decimation of mixed forests by 108 km², with only 10 km² left in 2019. Surprisingly, Badagry LGA in the west is the third highest area of urban growth with about 110 km² increase in built-up areas during 1984–2019. However, it was still left with an appreciable store of mixed forests and wetland forests in 2019. The growth in built-up areas in these hitherto rural LGAs attests to the fact of occurrence of rapid urban growth in hitherto far-flung rural areas which are associated with lower land costs. Eti-Osa LGA in the south, which is home to the rich, has the fourth highest expanse of built-up areas. The desire by the nouveau riche to be in this area is undoubtedly among the drivers of its growth.

From 1984, the shape and locations of the high LST range of 31–35°C mirrors the form and spread eastwards, northwards and westwards of built-up areas during 2001–2019. For instance, in 1984, Alimosho and Ikorodu LGAs, with built-up areas of 4.10 km² and 2.27 km², respectively, experienced mean LSTs of 28.6°C and 27.6°C, while the mean LST then was 28.6°C. This indicated a moderating influence of ecological assets since these LGAs were principally rural at the time. By 2019, when the built-up areas of the two LGAs had expanded phenomenally to 150.88 km² and 181.42 km², respectively, their mean LSTs correspondingly increased to 32.0°C and 30.8°C. At the same time, the rural and highly forested Badagry LGA in the west, which also experienced high growth in built-up area from 0.83 km² in 1984 to 111.63 km² in 2019, maintained a mean LST of 29.0°C in 2019. This mean LST value indicated a continuing moderating influence of mixed forests, which were available to the extent of 130.91 km² in 2019. This mean LST value for Badagry aligns somewhat with the earlier findings by Ojeh et al. (2016), which observed T_mean to be 27.5°C in December in the synoptic observation of seasonal variation and intensity of UHI and air temperature over urban and rural Lagos State during 2014–2015. Lower surface temperatures in 2019 generally characterise the distant and still rural LGAs of Badagry and Epe. Apapa LGA with its port complex, which was previously a major industrial manufacturing base, is in this LST segment perhaps due to the recent decline in manufacturing, since many manufacturing concerns have either closed or relocated to neighbouring countries.

The heat stress levels from utci.org are lower perhaps due to the fact that the actual wind speed of that day of 3.33 m × s⁻¹ was used in the calculation, while a wind speed of 1 m × s⁻¹ was embedded in the heat stress levels from climatechip.org. The higher wind speed expectedly induces more evaporative cooling and hence lower temperatures. In consequence, only seven LGAs from utci.org show strong heat stress, while only five LGAs from climatechip.org display moderate heat stress. Significantly developing rural LGAs such as Badagry and Epe, which have reasonable remaining amounts of areas of mixed forests, show moderate heat stress levels, while the rapidly built-up Alimosho LGA shows strong heat stress levels on both UTCI *1 and UTCI **2.

In a related study, Wang et al. (2020) conducted city-scale UTCI mapping during periods of heat waves in the Yangtze River Delta urban agglomeration in China, using satellite data. Their findings showed that the UTCI was generally higher in urban areas and varied considerably with land cover types. The changing patterns of thermal comfort under the influence of urbanisation was also shown in the work of Fu et al. (2019), which dealt with the dynamics of urban thermal comfort in the four major capital cities of China. Their work showed that urbanisation could increase urban temperature and decrease the RH and the wind velocity. According to Yahia et al. (2018), dense trees could enhance thermal comfort conditions. This finding of Yahia et al. might explain the moderate heat stress levels and lower temperatures observed at the fringes of Lagos State in Epe and Badagry LGAs, where vegetation cover is more widespread. Other researchers have also confirmed the general trend of less comfortable temperatures within urban centres in African cities such as Abuja (Isioye et al. 2020) and Nairobi (Ongoma et al. 2016) and in 173 other African cities (Rohat et al. 2019).

These results indicate the potential discomfort levels which people in various LGAs of the state undergo as a result of excessive outdoor heat, as gauged by LST. The heat stress levels shown are signals to the potential influence of LST and UHI on urban outdoor thermal comfort in various parts of the state.

The study has monitored the changing land cover, LST and thus UHI in Lagos State during 1984–2019 through the use of spatio-temporal Landsat data. While the study established the fact of rapid and continuing urban expansion into hitherto rural areas along with continuing warming of the state, it indicated a strong link between high LST intensities and rapid urban growth in some LGAs. The UTCI heat stress index derived for the state with LST results points to a potential influence of LST and thus UHI on the outdoor thermal comfort in the state. Generally, there is lower thermal comfort arising from higher temperatures and consequent heat stress within the metropolitan areas of the state. The results further reinforce the need for policies and actions to rein in urban sprawl and to improve the thermal quality of outdoor spaces through increased urban tree canopy and green infrastructure. A few areas for further research manifest from this study. A further investigation of LST to coincide with the period of maximum daily temperature, which usually peaks between 3:00–4:00 PM local time, is needed in the dry season to ascertain how closely it mirrors air temperature. Landsat overpass time, which usually occurs in the morning, neither coincides with the time of maximum temperature nor does it lend to a night-time assessment of LST. A further use of vegetation indices is desirable to underscore the LST–land cover relationship as well as the UHI urban–rural continuum in the state.

Field measurements of meteorological data in rapidly developed LGAs such as Alimosho and Ikorodu are imperative to generate data with which to test and validate the outdoor thermal indices in the state, provided funding can ever be found. In addition, the modelling of selected open spaces/pockets of vegetated spaces in the highly built-up LGAs such as Alimosho, Agege, Ikeja and Surulere, with ENVI-MET software, for the contribution of green infrastructure to LST modulation and outdoor thermal comfort, is necessary to generate indices for urban green space/landscape planning and design. The existence of only one weather station to date in the entire state presented a limitation to any attempt to compare the derived LST to the actual surface or air temperature, especially for the developing rural LGAs.

Finally, conscious that LST is used here as a proxy for air temperature due to lack of state-wide temperature data, it is safe to emphasise that findings on outdoor heat stress remain possibilities which need further studies of in-situ measurements for affirmation.