Influence of Nairobi’s Biophysical Characteristics on its Vulnerability to a Changing Climate

Urban areas experience exacerbated impacts of the regional climate variability because of their form characteristics such as imperviousness of surfaces, building density and distribution of open spaces. These are further confounded by geographical aspects such as topography, soil types, and vegetation types. Nairobi city is increasingly exposed to flood and heat risk as an aggregation of its urban form and the changing global climate. The paper sought to establish the influence of Nairobi’s biophysical characteristics on its vulnerability to both flooding and heat risks. The paper used a descriptive research design augmented with Geographic Information Systems to spatially model the landcover, soil drainage, topography, green space networks, and population density characteristics at the sub-location level. Vulnerability indices were developed using the expert ranking system and used to determine the vulnerability of the different sub-locations. The findings revealed a vulnerability pattern close to the historically segregated planning of the city. The central and eastern parts of the city exhibit high vulnerability while the western, northwestern, and southern parts of the city display moderate to low vulnerability. The paper recommends that adapting existing neighbourhoods and proactive planning of new neighbourhoods uses the ecosystem-based approach. This to entail decentralization of smaller green spaces, redesign of road medians for water management, re-specification of street vegetation species to incorporate a mix of deciduous and evergreen trees and incorporating eco-roofs and walls in high-density developments like the Central Business District.


INTRODUCTION Global Climate and Biophysical Characteristics of Urban Areas
Over 60% of the world's population is expected to be living in urban areas by the year 2050 (United Nations Department of Economic and Social Affairs, 2018). Similarly, the United Nations Framework Convention on Climate Change (UNFCCC) (2011) and World Meteorological Organization (WMO) (2019) projects that the impacts of climate change will be exacerbated the world over. As a result, the urban population is likely to experience confounded vulnerability. First, because of the increased incidences of climate-related hazards that affect the global environment (Thomas & López, 2015). Second, from the urban biophysical characteristics that influence how climatic parameters interact with urban locales and inhabitants (Oke, Mills, Christen, & Voogt, 2017).
The interaction between climatic factors and urban characteristics determines vulnerability at the urban level. Vulnerability in this context is the degree to which a system is susceptible to and unable to cope with adverse effects of climate change, including climate variability and extremes (Perry, Canziani, Palutikof, Linden, & Hanson, 2007). Its main determinants are exposure, sensitivity, and adaptive capacity (Intergovernmental Panel on Climate Change (IPCC), 2014; Krellenberg, Welz, Link, & Barth, 2017;Satapathy et al., 2014).
Exposure is the presence of people; livelihoods; environmental services and resources; infrastructure; or economic, social, or cultural assets in places that could be adversely affected (Field, Barros, Stocker, & Dahe, 2012;Fritzsche et al., 2014). Sensitivity is the degree to which a system is affected directly or indirectly, either adversely or beneficially, by climate-related stimuli (McCarthy, Canziani, Leary, Dokken, & White, 2001). Adaptive capacity is the ability of a system to adjust to climate change, moderate potential impacts through the combination of the strengths, attributes, and resources available to an individual, community, society, or organization (Perry et al., 2007). The confluence of these determinants defined by equation 1 developed from Satapathy et al. (2014) shows their contribution to vulnerability. Exposure in this paper applies to climatic parameters, while sensitivity encompasses aspects of the biophysical characterization of the city. dominant approach within the context of climate change often looks at the biophysical and socioeconomic characteristics within the confines of exposure, sensitivity, and adaptive capacity. Several factors have been developed to aid in vulnerability assessment. The key biophysical factors of urban areas that influence how it interacts with climate and by extension vulnerability include soil hydro-lithology (Mentzafou, Markogianni, & Dimitriou, 2018;Skilodimou, Bathrellos, Chousianitis, Youssef, & Pradhan, 2019), land use/ land cover: the ratio of built-up to open spaces (Dou et al., 2018;Gigović, Pamučar, Bajić, & Drobnjak, 2017) and topography (Doyle, Sullivan, Mahtta, & Pandey, 2017) and dwelling types (Table 1 and 2). For instance, soil and topography influence runoff, affecting vulnerability to floods. Dwelling types and density of built-up areas affect surface temperatures, influencing heat stress. Rainfall Intensity (Dou, et al., 2017) (Chen, Ito, Sawamukai, & Tokunaga, 2015) (Mentzafou, Markgianni, & Dimitriou, 2017) Flow Accumulation (Dou, et al., 2017) (Mentzafou, Markgianni, & Dimitriou, 2017) (Nayak, et al., 2018) (Wolf & McGregor, 2013) (Swart, et al., 2012) (Gamble, et al., 2018) (Dong, et al., 2014)

NAIROBI CITY'S CLIMATE AND BIOPHYSICAL CHARACTER
This paper is based on Nairobi as defined by the city-county boundary comprising the 112 sublocations as at the 2008 housing and population census ( Figure 1). Nairobi referred to as a citycounty since the entire city is a county. It extends between 36° 4′ and 37° 10′ East and between 1° 9′ and 1° 28′ south. The city has a population of about 4.7 million people within an area of about 695 km² as per the 2019 Kenya National Bureau of Statistics (KNBS, 2019). The city has 112 sub-locations with varying population densities. High-income sublocations have average densities as low as 500 people/km 2 while low-income locations such as those in the informal settlements have densities as high as 63 000 people/km 2 .

Figure 1: Nairobi city-county boundary showing sub-locations
Nairobi has an undulating topography with an elevation range of 1,460 m -1,920 m above sea level. The lowest elevation occurs at Athi River at the eastern boundary of the city and highest at the western rim of the city. Average annual rainfall in Nairobi is about 900 mm with peaks in April and November ( Figure 2). The mean daily maximum temperature by month ranges from 28 o C to 22 o C and the minimum ranges from 14 o C to 12 o C (Japan International Cooperation Agency (JICA), 2004). Nairobi is underlain by tertiary volcanic rocks and the soil types are the grey clays, black clays, red clays, reddish-brown clays and brown clays with varying soil drainage properties (University of Nairobi, 2018).

Procedure for Data collection and analysis
The paper used a descriptive approach and modelling of the parameters in ArcMap 10.6. The modelling developed thematic maps of the biophysical parameters. Heat and flood vulnerability mapping followed this. Vulnerability mapping was based on an aggregation of the experts' ranking of parameters. The ranking determined their contribution to vulnerability within the classes of sensitivity and exposure. The census data further aided in developing population density maps. Even though population density is a socio-economic parameter, it was selected to represent density data since the landcover images only show the 2-dimensional density.
The biophysical data comprised Landsat images, Digital Elevation Models (DEMs) and Soil data. The topography, flow accumulation and landcover were modelled from Landsat and DEMs. Kenya Agriculture and Livestock Research Organization (KALRO) and Survey of Kenya (SoK) supplied soil data. Kenya Meteorological Department (KMD) and the IGAD Climate Prediction and Applications Center (ICPAC) supplied the climatic data.
The landcover data underwent pre-processing and post-image processing to enhance the output of the images in terms of quality by correcting any image distortions. The satellite imagery was geometrically corrected and projected to ARC 1960 Universal Traverse Mercator (UTM) 37 S; a recognized projected co-ordinate system in Kenya. A pixelbased hybrid classification method was adopted. It combined supervised and unsupervised classification methods. The supervised method was adopted for its effectiveness and accuracy, while the unsupervised method was used to plug the gaps and minimize errors in the supervised method. The resultant images were then merged using the reclassify tool in ArcMap 10.6. The resultant landcover classes were built-up areas, forest, bare lands, grasslands, water bodies ( Table 3)

……… [2]
The soil data parameter used was soil drainage properties. These were classified by International Soil Reference and Information Center (ISRIC) as excessively drained, somewhat excessively welldrained, moderately well-drained, imperfectly drained, poorly drained, very poorly drained and a non-classified section. The non-classified section is the built-up area. A thematic map based on the classifications found in Nairobi was developed.
Terrain data was modelled from the 2 m spatial resolution DEM using ArcMap 10.6. The DEM that was projected to Arc 1960 UTM zone 37S. The output was used to create a slope raster file. The slope classifications were 0% -4%, 4% -8%, 8% -12%, 12% -16% and 16% -100%. The selection of the classifications was based on studies by Nassif & Wilson (1975) and Huang, Kang, Yang, & Jin (2017) who note those slopes as the ones that impact stormwater runoff. The terrain dataset was further used to develop Flow Accumulation Maps.
Relying on the DEM, the spatial analyst tool in ArcMap was used to generate fill raster image which was then used to calculate the flow direction and develop the flow accumulation map.
The values of the parameters were normalized (eq. iii) as recommended by Swanson, Hiley, & Venema (2007), Elmoustafa (2012) and Inostroza, Palme, & de la Barrera (2016). This is to eliminate any misinterpretation that could result from the different units of measurements used. The normalized data were used to generate thematic vulnerability maps.
Where Xnor. -The normalized X; X -The values X1-Xn; Xmin -The lowest value of X; Xmax -The highest value of X A vulnerability index was developed using the statistical mean. Vulnerability ranking used a fourlevel scale of Very High Vulnerability (VHV), High Vulnerability (HV), Moderate Vulnerability (MV) and Low Vulnerability (LV). From the vulnerability maps of the different parameters, the overall vulnerability of the county to heat and flood risks was computed. It was based on the expert rankings of the contributions of the unique aspects of heat and flood risks. The ranking was based on the Likert type scale (Figure 3). The paper considered 12 experts in the fields of architecture, urban planning, landscape architecture and hydrology.

Land use/Landcover
Dominant land covers in the city are grassland at 244 km 2 and built-up area at 228 Km 2 while the least land cover is a water body at 4 km 2 ( Figure 5). The built-up area encompasses buildings and paved surfaces such as roads. It is dominant in the central zone of the city, reducing towards the periphery. This translates to a Very High Vulnerability (VHV) to flooding and heat stress in the central zone ( Figure 6). The densest sub-locations with normalized densities above 0.6 include Lindi, Laini Saba, Kibra, Gatwikira, Korogocho, Mathare 4A and Uhuru (Figure 7). They host some of the largest informal settlements in the city like Kibra, Mukuru Kwa Njenga, Mathare, and Korogocho. The high population density implies a large population at risk to both flood and heat stress. As a result, the same sub-locations show VHV (Figure 8).

Soil Drainage Properties
The Soil Mapping of Kenya by ISRIC lists soil drainage properties within Nairobi into 5 classes. They are Well-drained, Imperfectly drained, Poorly drained, Very poorly drained and a non-classified section (Fig. 10). The unclassified, poorly and very poorly drained soils are concentrated along a central horizontal belt. The unclassified zone is a zone of dense development. It is therefore classified among the poorly drained soils due to its imperviousness. This translates into VHV at the centre of the city and LV to the north and west of the city (Fig. 11).

Topography
Majority of the slopes on the eastern side of the city fall below 8% and the western side above 16%. This translates to an eastern side concentrated MV ( Figure 11). 19 % of the sub-locations manifest MV and the rest LV ( Figure 12). Figure 11: Nairobi city's slope analysis Figure 12: Nairobi city's vulnerability to flooding risk based on slope.

Flow Accumulation
The flow accumulation considered the flow accumulation model and natural drainage patterns ( Figure 13). Those with high flow accumulation connections but no natural drainage pattens were ranked as being moderately vulnerable. MV was selected to account for possible artificial drainage. The MV is concentrated on the central and northern zones of the city (Figure 14). The open space patterns were mapped by the UN-Habitat (Figure 15). The central part of the city reveals VHV to HV while the peripheral zones of the city reveal LV (Figure 16). The NDVI model for the year 2018 manifests a western and northwestern concentration of vegetation whose reflective index shows tree cover of good health in the dry season; which is the period Landsat images extracted ( Figure 17). Based on this model, the central and eastern zones, and the northern tip of the city display HV compared to the western, northwestern, and southern zones ( Figure  18). The experts ranked the magnitude of the different parameters in contributing to flood and heat stress vulnerability on the 1-7 Likert scale (Table 4). They ranked in descending order as flow accumulation, land cover, slope and soil drainage properties. Parameters of heat stress in descending order were land cover and NDVI. The LV zones ate to the northern and western sides ( Figure 19).

DISCUSSION
The sub-locations show various levels of vulnerability to the individual biophysical parameters. However, zonal segregation of the different vulnerabilities is clear. For instance, very high and high vulnerability sub-locations are concentrated in the central zone of the city. This changes to moderate vulnerability spreading to the east and south with a narrow belt extending westwards. However, the northern and western sublocations reveal low vulnerability.
This zonal vulnerability compares with the documented historical segregated approach to the city's planning. The eastern side of the city was low-income African neighbourhoods and the northern and western sides of the city were high income, low-density white settlements. This segregation has played a significant role in increasing the city's biophysical vulnerability in a changing climate. Informal settlements and formal low-income sub-locations like Korogocho, Kibra, Huruma, Laini Saba, Mathare, and Lindi manifest top levels of vulnerability to biophysical characteristics. This emanates from the very high population density that exposes the residents to climate risks in two aspects. First is the high population that becomes vulnerable in case of hazards such as flooding. Second is the high densities that translate to high urban heat island and surface sealing that increase heat and flood vulnerability, respectively. This is at pace with the poorly drained soils that increase runoff and lack of adequate tree cover to improve local temperatures.
High-income sub-locations like Karen, Hardy, Garden, Kyuna, and Muthaiga manifests low biophysical vulnerability because of the low density of built-up areas and large swathes of open and green spaces that provide resilience attributes such as temperature regulation and percolation of stormwater. The neighbourhoods also have welldrained soils and slopes above 16% both of which ensure good drainage of stormwater.

CONCLUSION AND RECOMMENDATION
These findings support the argument by Satterthwaite, Huq, Pelling, Reid, & Lankao (2007) who classify the urban poor high among the vulnerable urban populations that require urgent action. Similarly, the multifaceted nature of their vulnerability as discussed by Carmin & Zhang (2009) is evidence in Nairobi where the urban poor inhabit locations at risk due to multiple parameters such as soil drainage properties, slopes percentages, and landcover types. Therefore, to achieve urban climate resilience, urban planning and development in Nairobi must be sensitive to biophysical parameters like the nature and distribution of open spaces, types of vegetation, soil types and density of development, especially in the informal and lowincome neighbourhoods ( Figure 22).

Figure 22: Sample adaptation approaches in high-density neighbourhoods and informal settlements in Kibra sub-location
Development of new neighbourhoods, urban renewal and regeneration projects such as the proposed Eastlands Urban Renewal Project must consider biophysical characteristics and their role in climate change resilience. For instance, the highdensity development can be concentrated in the poorly drained soils and the green open spaces located on the well-drained deep soils. This would aid in the proper management of stormwater runoff.
Some of the ecosystem-based adaptation approaches such as river re-naturalization, green infrastructure and green corridors proposed by Geneletti & Zardo (2016) and Scarano (2017) are ideal for application in Nairobi city as they deal with both biophysical and socio-economic aspects. Their downscaled implementation can include decentralization of smaller green spaces in highdensity sub-locations such as Umoja and Kibra, redesign of road medians for water management, re-specification of street vegetation species to incorporate a mix of deciduous and evergreen trees and the incorporation of eco-roofs and walls in high-density developments like the Central Business District (Figure 23).

Figure 23: Climate change adaptation within a section of the Nairobi CBD
Further studies on biophysical vulnerability in Nairobi City should incorporate the parameters of dwelling types and materials.