GIS-Based Multi-criteria Analysis for Sustainable Urban Green Spaces Planning in Emerging Towns of Ethiopia: The Case of Sululta Town


 BackgroundUrban green spaces are important components, contributing in different ways to the quality of human well-being. In the planning and management of urban centres, attention to the appropriate site selection of urban green spaces with regard to the possible importance that these spaces have from the perspectives of ecology, socioeconomic, mentality, etc., is an inevitable requirement. In present decades, land suitability mapping methods and GIS have been used to support urban green space planners in developed countries; however, its application and practices are limited in developing countries, like Ethiopia. Therefore, the aim of this study has to select potential sites for green spaces in Sululta town that assist an effective planning process of green areas in a sustainable way. MethodsIn this study, GIS-based Multi-criteria analysis (MCA) has been adopted to select suitable sites for urban green spaces. Existing land use, proximity to settlement, road and water body, population density, land ownership, topography, and scenic attractiveness were recognized as the key factor affecting urban green land suitability. ResultBased on GIS-based MCA, 13.6%, 34%, 28%, and 18.9% of the study area are highly suitable, suitable, moderately suitable, and poorly suitable, respectively. Furthermore, based on the suitability analysis out of the total area of the study town 5.5% of the landmass is not suitable for green spaces. ConclusionTherefore, the application of this tool has provided an effective methodology to solve a complex decisional problem in green spaces site selection in the study town and urban planning all over the country.


Introduction
In the past and present decades, the world has experienced unprecedented urban growth, with more than 50% of the global population living in urban areas now (Wu, 2014). The global urban population is projected to be 6.3 billion by 2050, almost double the global population of 3.5 billion urban dwellers in 2010 (Secretariat of the Convention on Biological Diversity, (SCBD), 2012). This rapid urbanization has posed greater pressure on natural resources and the This situation also occurs in the case of Ethiopia, which is one of the fastest growing countries in sub-Saharan Africa (Lamson-Hall et al. 2018), and studies have focused on the impacts of urban growth on green space (Abebe & Megento, 2016;Gashu & Gebre-Egziabher, 2018;Abo El Wafa et al., 2018), climate change adaptation (Lindley et al., 2015), the development of functional green infrastructure and ecosystem service (Woldegerima et al., 2017), planning aspect (Girma et al., 2018), green spaces depletion (Girma et al., 2019) and utilization pattern (Yeshewazerf, 2017;Molla et al., 2017). However, the topic of suitability analysis for green space in the urban environment has not discussed in these studies. This study therefore aimed to ll the existing research gap by using GIS-based Multi-criteria analysis to identify suitable sites for urban green space development in Sululta town.

Materials And Methods
Description of the study area Sululta town is located in Sululta district of the previous North Shewa administrative zone of Oromia region, currently under Oromia special zone surrounding Fin nne. It is situated very close to the district capital town Chancho and Addis Ababa, which are far about 15 and 23 km in the north and south direction, respectively. Astronomically, the study area is located between 9 0 30'00"N to 9 0 12'15"N latitude and 38 0 42'0"E to 38 0 46'45" E longitude. The administrative area of the town is about 4471 hectares. Sululta has the same general climatologically characteristics as that of Addis Ababa. Globally it is a part of tropical humid climatic region, which is characterized by warm temperature and high rainfall. The soils of the zone are basically derived from mesozoic sedimentary and volcanic rocks. The major soil types of Suluta area are Chromic Luvisols.

Methods
Urban green spaces have continuously played a signi cant role in enhancing the quality of life of urban inhabitants and in supporting urban metabolism.
However, urban green spaces have experienced a physical and social decline, while its heterogeneity and richness is often neglected and its contribution to the well-being of a community ignored within current urban planning instruments in Sululta town (Girma et al., 2018;Girma et al., 2019). Under this circumstance, GIS-based multi-criteria land suitability analysis is becoming critical in determining the land resource that is suitable for urban green spaces (Cetin, 2015).
Continued development and re nement of suitability analysis, particularly with GIS technology, can enable urban planners to create a suitable urban green spaces system in the urban environment (Manlum, 2003). Therefore, this study proposed the application of GIS-based multi-criteria suitability analysis using analytical hierarchy process (AHP) to support the decision-making process on selecting an appropriate site for development urban green spaces. This approach will be used as a basis for the town's administration and the planning authority to identify an appropriate and potential site for providing suitable, su cient and accessible urban green spaces to the urban dwellers. Moreover, it will be used as a benchmark to guide the sustainable land use decision in the study area.
In this study, to select a suitable site for urban green spaces using GIS-based multi-criteria analysis the following ve main steps were used: Spatial and non-spatial data collection Determination and rating of criteria and sub-criteria Criteria standardization and factor map generation Determination of weighting for factors and Weighted overlay analysis Spatial and non-spatial data collection Firstly, the primary data from the eld survey were collected through interviews undertaken with different experts in the related eld of study for identifying factors that are important for urban green spaces site selection. Secondly, various spatial data were obtained from different sources ( Table 1). The data were analysed in ArcGIS 10.2 and ERDAS Imagine 2010 for further analysis and mapping purposes. agreed criteria and factors (Jabir and Arun 2014). Therefore, by synthesizing literature review, personal experiences, experts opinions and previous related studies conducted by different researchers (Manlun, 2003;Uy and Nakagoshi, 2008;Pantalone, 2010;Ahmed et. al., 2011;Kuldeep, 2013;Heshmat et al., 2013;Elahe et al., 2014;Yousef et. al., 2014;Abebe, and Megento, 2017;Li et al. 2018;Dagistanli, et al., 2018;Ustaoglu and Aydinoglu, 2020) 12 factors were considered for selection of suitable site for development of urban green spaces (Table 2).
Besides identifying appropriate criteria and sub-criteria to select a suitable site for urban green spaces the rating has been assigned for each factors. In order to assign a rating (score) for each criterion and sub-criteria, review of previous scienti c experimental research ndings and literature on parameters were undertaken. Furthermore, reviews were consolidated through consultations and discussion with experienced experts and researchers from various disciplines.
Rating of factors has usually made in terms of ve classes: highly suitable, suitable, moderately suitable, poorly suitable, and not suitable (FAO, 2006).

Criteria standardization and factors map generation
In GIS-based multi-criteria decision-making analysis, there is a need to standardize the data in order to integrate the data measured in different units and mapped in different scale of measurement such as ordinal, interval, nominal and ratio scales (Pereira et al, 1993). Even though there are different methods that can be used to standardize criterion maps, linear scale transformation is the most frequently used technique (Malczewski, 2003). For criterion standardization in this study, all the vector maps of the criterion were converted to raster data formats. Afterward using the Spatial Analyst tool in ArcMap the raster maps were reclassi ed into ve classes with the values that range from 1 to 5, where the value of 5 was taken as highly suitable while that of 1 was unsuitable for all factors considered. This approach will enable all measurements to have an equivalent value before any weights were applied. However, it was important to note that there were some variables that did not ful l the whole range of the criteria. Once all the criteria maps were standardized, a weight of each criteria map was calculated using AHP.

Estimating weight for factors and sub-factors
One component of GIS-Based multi-criteria decision-making analysis is assigning criteria weights for each factor maps. The purpose of weighing in this process is to express the importance or preference of each factor relative to another factor effect on urban green spaces. In this study, the Analytical Hierarchy Process (AHP) using pairwise comparison matrixes were used to calculate weights for the criteria maps. AHP is a widely used method in multi-criteria decision-making analysis and was introduced by Saaty (1980). In this study, the AHP was carried out in three steps. Firstly, pair-wise comparison of criteria was performed and results were put into a comparison matrix. A Pair-wise comparison is performed in the 9-degree preferences scale, which is suggested by Saaty (1980), each higher level of scale shows higher importance than the previous lower level (Table 3). Extremely more important According to Saaty (1980), the values in the matrix need to be consistent, which means that if x is compared to y, it receives a score of 9 (strong importance), y to x should score 1/9 (little importance) and something compared to itself gets the score of 1 (equal importance). Experts are asked to rank the value of criterion map for pairwise matrix on a saaty's scale. Moreover, the pairwise comparison matrices (Table 4)    The second step was calculating criterion weights, in this case, values from each column were summed and every element in the matrix was divided by the sum of the respective column. The new matrix is called normalized pair-wise comparison matrix (Table 4). Finally, an average from the elements from each row of the normalized matrix was calculated. Once the pair-wise comparison was lled and the weight of the factor was determined, a consistency ratio (CR) was calculated to identify inconsistencies and develop the best-t weights in the complete pair-wise comparison matrix. A consistency ratio was calculated for each pairwise comparison matrix to verify the degree of credibility of the relative weights, by using the following formula (Bunruamkaew and Yuji, 2001). Where; CR=Consistency ratio CI= referred to as consistency index RI = is the random inconsistency index whose value depends on the number (n) of factors being compared; as illustrated in Table 5 (Saaty, 1980). The consistency index (CI) was calculated by the following formula: Where; n= the number of items being compared in the matrix λ max = Average value of the consistency vector

Weighted overlay analysis
Once the criteria maps and weights have been developed and established, a decision rule of multi-criteria analysis was used. As pointed by Jiang and Eastman (2000) and Malczewski, (2003) there are three common decision rules in multi-criteria analysis namely weighted linear overlay, Boolean overlay and ordered averaging. The weighted linear combination technique was applied to aggregate the standardized layers in this study. In weighted linear combination procedure, factors and parameters (Xi) are multiplied by the weight of the suitability parameters (Wi) to get composited weights and then summed. This can be expressed by using the following formula to derive the intended map i.e. urban green spaces suitability map for the towns. Where; S= total suitability score Wi = weight of the selected suitability criteria layer, Xi = assigned sub criteria score of suitability criteria layer i n =total number of suitability criteria layer Result And Discussion

AHP Weights
The result of AHP shows that the derived factors have a different degree of in uence on urban green spaces. As it is evident from Table 6, the weight assigned to the factors reveals the relative importance of each parameter in exposing an area to urban green spaces evaluation. As a result shows, an area with high population density with the normalized weight of 0.22 has the highest priority. Proximity to settlement area with the weight of 0.21 is in the second priority.
Slop with a normal weight of 0.13 has the third priority. Proximity to the road with a normal weight of 0/10 is in the fourth priority. Elevation with normal weight of 0/059 is of the fth priority. The area with vegetation cover with normal weight of 0/048 is the next priority. The ood-prone area with the normal weight of 0/04 is in the low priority. Proximity to water sources, visibility and existing land with almost similar weight of 0/032, 0/032 and 0/039, respectively, have relatively lowest priority (Table 6). These imply that the higher the weight in the percentage of a factor, the more in uence it has in suitable site selection for urban green spaces. Saaty (2008) has shown that Consistency ratio of 0.1 or less is acceptable to continue the AHP analysis. But if it's larger than 0.10, then there are inconsistencies in the evaluation process, and the AHP method may not yield a meaningful result. In this study, consistency ratio or CR of conducted comparisons has obtained 0.09, which is smaller than 0.1 and therefore the comparisons can be acceptable. The computation of consistency ratio is given in Table 6, below.
Based on the result of this study, AHP is a highly e cient instrument for determining factor weights and is more bene cial than alternative approaches since the inconsistency of the factor weights' pair-wise comparison matrix can be calculated and controlled by the Consistency Ratio (CR). In various studies (Dong et al., 2008;Tudes and Yigiter, 2010;Kumar and Shaikh, 2012;Bagheri et al., 2013;Romano et al., 2015; Abebe and Megento, 2017; Ustaoglu and Aydinoglu, 2020), this has been con rmed. Studies have shown that current land use must be considered when choosing suitable sites for the development of urban green spaces and have identi ed the suitability of different land uses based on their use type (Uy, & Nakagoshi, 2008;Zhang et al. 2013;Malmir et al. 2016;Abebe and Megento 2017;Dagistanli, et.al. 2018). Open spaces and forest land were considered to be highly suitable for urban green spaces in this study, based on knowledge obtained from the analysis of literature and expert opinion. To rehabilitate the quarry area they are considered as suitable for urban green spaces. Additional, in this study, existing building area and water body has considered as moderately suitable for urban green spaces. In this study, agriculture is regarded as poorly suited to urban green spaces ( Figure 2I and Table 2).
Various researchers have shown that low-slope areas are highly suitable for the development of urban green spaces (Heshmat et al. 2013;Mahdavi and Niknejad, 2014;Pramanik, 2016;Abebe and Megento, 2017;Dagistanli et al. 2018) and 0-10 slope areas are suitable for urban green spaces such as open spaces and parks. This study therefore considered the lower slope land to be more suitable than the higher slope land and area with slope of 0-5%, 5-10%, 10-15% and 15-20% has considered as highly suitable, suitable, moderately suitable, and poorly suitable, respectively, for identify suitable site for urban green spaces development. Area with the slope greater than 20% considered as unsuitable for developing urban green spaces in this study ( Figure 2D and Table 2).
In selecting suitable sites for urban green spaces, elevation have also signi cant role and should be considered as the major factor (Gul et al. 2006;Mahmoud, & El-Sayed, 2011;Li et al. 2018;Dagistanli, et.al. 2018). Based on the information acquired from literature review and expert opinion, the elevations between 2550-2600m, 2600-26500m, 2650-2700m and 2700-2800m in this study area were considered as highly suitable, suitable, moderately suitable and poorly suitable, respectively. In this analysis, areas with elevations greater than 2800 m were considered to be unsuitable for the development of urban green spaces ( Figure 2H and Table 2).
In any geographic analysis, proximity is always signi cant. Green spaces must be accessible to settlement areas in urban areas, since they have numerous ecological, social and economic bene ts (Zhang et al. 2013;Malmir et al. 2016;Ustaoglu and Aydinoglu, 2020). Therefore, the proximity of green spaces to the settlement area in terms of distance is very important to consider. In this research, the proximity of the settlement area was taken as a criterion. Based on this, those areas that have been identi ed within 500m distances from the settlement area have been considered as highly suitable by making Euclidian distances and the area with distances from 500m to 1000m has been considered suitable ( Figure 2G and Table 2). In addition, the area with distances of 1000m to 2000m, 2000m to 3000m and greater than 3000m form settlement area was considered to be moderately appropriate, poorly suited and unsuitable for the development of urban green spaces.
The road proximity also plays a vital role in providing convenient and feasible routes to the local population to reach local green areas in their surroundings (Bunruamkaew and Murayama, 2011;Kienast et al. 2012;Morckel, 2017). Elahe et al. (2014) and Ahmed et al. (2011) indicated that if it is situated at an acceptable distance from roads in order to access transport, the green space site is preferable. As a result, the road network proximity has been given due consideration as one aspect of infrastructural facilities in the mapping of suitability maps for urban green areas. Based on this, by making Euclidian distances, areas within the 400 m radius of the road network has considered as highly suitable, area within the 400m-800 m range was considered suitable, and area within the 800m-1000 m range was considered as moderately suitable. In addition, the area between 1000 m and 1500 m has considered as poorly suitable and the area more than 1500 m from the road network has considered as not suitable ( Figure 2F and Table 2). Studies have also shown that the types of roads have an effect on the selection of suitable urban green spaces (Gul, et.al. 2006;. Research conducted by Gul, et.al. (2006) and Chandio et.al., (2011) found that areas with access to major roads are highly appropriate for the development of urban green spaces than areas with access to local roads such as gravel-soil roads, forest soil roads. Therefore, arterial and collector roads are considered to be highly suitable in this study for the selection of suitable locations for urban green spaces, as these types of roads are highly distributed in the town. In addition, main roads and local roads are regarded as suitable and moderately suitable, respectively ( Figure 2J and Table 2).
Manlun (2003), Heshmat et al. (2013), Kuldeep (2013) and Abebe and Megento (2017) have noted that for the development of green space, lands closest to rivers, lakes and reservoirs are highly suitable. Therefore, on the basis of this claim, the distance less than 250 m from the river considered to be highly suitable and between 250 m and 500 m is considered as suitable in this study. Moreover, distances between 500m to 1000m and 1000m to 1500m is considered as moderately suitable and poorly suitable for urban green spaces, respectively. Whereas distance greater that 1500m relatively considered as totally unsuitable ( Figure 2E and Table 2).
Flood-prone areas have also been introduced as parameters for the study of suitability. Studies found that the area within the lower ood-prone area has more suitable than the land with higher ood-prone area for urban green spaces development and they indicated that urban green spaces must be free from ood prone area as most as possible (Piran et al. 2013;Peng et al. 2016). Based on the information obtained from the literature review and expert opinion, high ood risk areas has considered as unsuitable for the development of urban green spaces in this study, and low and medium ood risk areas are considered as highly and moderately suitable ( Figure 2A and Table 2).
Urban green space suitability assessment is directly or indirectly correlated with different socio-economic factors. Population density is known to be one of the socio-economic factors in uencing the appropriate selection of green space in urban areas. Places with a higher number of people with crowded places near the high population density required access to the open green spaces (Schipperijn et.al. 2010). Some researchers (Gul,et.al., 2006;Pantalone 2010;Ahmed et al. 2011;Heshmat et al. 2013;Elahe et al. 2014;Dagistanli, et.al. 2018) recommend that areas that have high population density are highly suitable for developing green space. On the basis of this claim, the study area is densely populated in the northwest, north, south and southeast, and is considered to be highly suitable for the development of urban green space. The eastern portion is sparsely populated and believed to be insu ciently suited to urban green areas. As it has a medium population density, the central and western parts of the town has considered as moderately suitable for urban green spaces development ( Figure 2B and  Dagistanli, et.al. 2018). Based on the information obtained from the literature review and expert opinion, in this study area with high vegetation cover has considered as highly suitable for urban green space development. Moreover, area with medium and low vegetation cover has considered as moderately and poorly suitable, respectively ( Figure 2k and Table 2).
The availability of land is often considered as signi cant factor in the selection of appropriate sites for urban green spaces. Studies have shown that public land is highly suitable for urban green space development as compared to private land (Chandio et.al. 2011). The study undertaken by Wang & Chan (2019) suggest that the situation with initial public land ownership status backed up by regulatory instruments is more advantageous for providing urban green spaces than that with the initial private land ownership status relying on market-based instruments. On the basis of this claim, in this study public land is considered as highly suitable and private land has considered as moderately suitable for selecting optimal location for urban green spaces in the town ( Figure   2L Table 2).
In this study, as suggested by Gul, et.al. (2006) and Nur (2017), scenic beauty is also considered to decide the best or potentially acceptable sites for urban green space development. Based on the information obtained from the literature review and expert opinion, in this study area with high, moderate and low scenic attractiveness has considered as highly, moderately and poorly suitable for appropriate site selection of urban green space development, respectively.

Final Suability analysis for urban green spaces
After weighting the criteria, as regards the relative importance of each criterion as well as suitability index, all the criterion maps were overlaid and nal urban green spaces suitability map was prepared. According to GIS-based multi-criteria analysis, the nal suitability maps have ve classes for the study town that are highly suitable, suitable, moderately suitable, poorly suitable and unsuitable. Suitability maps of Sululta towns are demonstrated in Fig. 3.
Based on table 7, out of the total area of the Sululta, town, about 13.6% (610.7 ha) area fall under the highly suitable category. The suitable area covers an area of 34% (1523.9 ha) of Sululta town. The area which is shaded by blue colour covers 28% (1276.6 ha) of Sululta town representing the moderately suitable class. Moreover, based on the table (7), out of the total area 18.9% (813ha) of Sululta towns have been covered by poorly suitable class. Out of the total area 5.5% of Sululta towns land mass is not suitable for urban green spaces. In general, the nal suitability maps show a series of spaces following a pattern and connectivity. These can be adapted to form the urban green spaces system, complete with corridors and hubs within the study area. This can increase opportunities for residents and biodiversity to enjoy the nature and bene ts of urban green spaces. Moreover, as the maps show the town have a high potential for developing the urban green spaces such as playground, sport eld, parks and the like as more than half of the town's lands mass are suitable. Therefore, the planning authority and the towns' administration can take this approach as a benchmark to provide suitable, accessible, interconnected and su cient urban green spaces in town under study. There are similar studies in the literature proposed GIS based multi criteria analysis for land uses planning both in developed and developing countries (Do Carmo Giordano & Setti Riedel, 2008;Uy & Nakagoshi, 2008;Chandio et al. 2011;Abebe & Megento 2017;Ustaoglu, & Aydinoglu, 2020).

Conclusion
In this study, GIS-based multi-criteria analysis has been used to support the site selection process for the development of urban green spaces. The study results are very signi cant in evaluating the feasibility of the use of GIS-based multi-criteria analysis for the development of urban green space. Since, by using appropriate analytical methods, the evaluation of urban green space is necessary to recognize their potential and to better select the most suitable land uses to improve their integrity and maintain the bene ts obtained from them.