Implementation and Evaluation of Urban Green Space Supply for Debre Berhan Town, Ethiopia

Urban green space (UGS) is part of urban land covered with vegetation which has an immense benet for sustainable urban development. The main objective of this study was urban green space supply analysis and evaluation for Debre Berhan town, Ethiopia. To realize this objective, we generated UGSs information by using normalized difference vegetation index (NDVI) which was calculated from Sentinel 2 satellite image. Then, the result of NDVI calculation was classied as small vegetation and high vegetation with over all accuracy of 87.29% and kappa value of 0.808. According to the result, there are 15.85% and 1.55% supply of urban green space for small and high vegetation respectively. In addition, the UGS supply in inner and outer part of the town for each land use was quantied. The result showed that in the inner part of the town, service, green area, and vacant land uses have 30.12%, 30.06%, and 13.32% of UGS supply respectively. Conversely, recreation, commercial, and administration land uses are least supplied with 0.11%, 0.26%, and 0.40% share of UGS. Furthermore, the UGS supply was evaluated against the new urban planning strategy of Ethiopia (30% coverage) and World Health Organization standard (9m 2 per person). The coverage of small vegetation and high vegetation was 15.85% and 1.55% respectively. The UGS per capita for small vegetation was about 75.16m 2 while the value for high vegetation was 7.33m 2 . The results of our analysis can be used as an input for urban planning and master plan revision.

In order to solve the aforementioned urbanization driven problems, urban green infrastructure through establishment of interconnected network of green spaces is identi ed as an alternative nature-based solution (Benedict and McMahon, 2002;Gill et al., 2007;Grimm et al., 2008;Ramdani, 2013;Rakhshandehroo et al., 2017;Nikolic and Yang, 2020).
Due to the environmental pressures of growing urban centers, many cities around the world have undertaken greening programs such as urban tree planting, enhancing parks, and providing incentives for green roofs to bene t from the amenities of urban green spaces (Li et al., 2015). In this respect, industrialized cities in the developed world have been trying to integrate sustainable ecological, social, and economic dimensions in all aspects of urban development (Bunce, 2009;Haase et al., 2010). However, in developing countries, the number of urban green spaces decreases, and there is unequal access to urban green space (Ramaiah and Avtar, 2019;Sabyrbekov et al., 2020;Rae, 2021). In sub-Saharan African countries, components of urban green infrastructure are depleting at an alarming rate in several cities across the continent (Mensah, 2014). However, there is a lack of empirical studies about the status, availability, and accessibility of urban green infrastructure in cities of African countries (Azagew and Worku, 2020). As part of the sub-Saharan Africa, Ethiopia is among the world's fast urbanizing countries even with more than population growth rate (Gulyani et al., 2001) but with unplanned urban growth (Zeluel et al., 2011).
Geospatial data and analysis techniques are very effective in urban greenery planning, their availability assessment, and evaluation. Many studies demonstrated the effectiveness of geospatial techniques for urban green space analysis and mapping (Hashim et al., 2019;Elderbrock et al., 2020;Woldesemayat and Genovese, 2021;Kumakoshi et al., 2020;Le Texier et al., 2018). Now a days Debre Berahn town is experiencing rapid growth of industrialization due to its proximity to Addis Ababa and its conducive climatic condition. With the expansion of industry, the town will be at high risk of emissions in the near future. To address this problem, expanding the supply of green space is an important task. Therefore, this research examines the nature of city's green space supply, identi es the underlying issues of UGS supply and provides accurate information to stakeholders.

Study area
Debre Berhan town is the administrative Centre of North Shewa Zone which is found in Amhara regional state of Ethiopia. Its astronomical location is in between 9 0 36'30''-9 0 42'32''North and 39 0 27'56''-39 0 34'26" East ( Fig. 1). In its relative location, it is found at 130 km road distance from Addis Ababa (Ethiopia's capital) in the Northeast direction. Presently, it is classi ed in to 9 administrative kebeles (the smallest administrative unit in Ethiopia). The total area of Debre Berhan town is 5710ha. In this research, the town area is further classi ed as inner part (1654ha), and outer part (4055ha) based on the level of urban development. Based on the population projection of Central Statistical Agency of Ethiopia, the current population size of the town is 120399 (CSA, 2019).

Data type and sources
In this study, we used band 4 and 8 of Sentinel 2 satellite image from o cial web site of Sentinel hub (https://scihub.copernicus.eu/). These two bands were used to calculate normalized difference vegetation index (NDVI) for Debre Berhan town. In addition, the land use plan of Debre Berhan town was obtained from town municipality. Ground control points (GCPs) were also collected from the eld using hand held GPS (Table 1).

Normalized Difference Vegetation Index (NDVI) calculation and classi cation
The Sentinel 2 satellite image covering Debre Berhan area was downloaded from the o cial web site of Sentinel hub. To minimize the effect of haze and cloud, we downloaded the dry season image of 18th January 2021. Thereafter, the downloaded image was subseted by the shape-le of our study area in ERDAS IMAGINE software.
The normalized difference vegetation index (NDVI) is the ratio of the difference between the near-infrared band (NIR) & the red band (R) and the sum of these two bands (Rouse et al., 1974) Where, NIR is near-infrared re ectance, RED is red re ectance. In this research work, sentinel 2 image was used to calculate NDVI. When calculating the NDVI, band 8 and band 4 were used for near-infrared and red bands respectively.
NDVI value for a given pixel always ranges from − 1 to + 1. A zero NDVI value corresponds to no vegetation while closer to + 1 indicates the highest density of green leaves. In addition to NDVI calculation, we classi ed NDVI result in to three classes with overall accuracy of 87.29 % and Kappa value of 0.808 (Table 2). While classifying the NDVI image, NDVI value of 0.25 was used as a minimum threshold for vegetation (Elderbrock et al., 2020). The NDVI raster data was then classi ed in to three classes of urban green space as: non-vegetation, small vegetation, and high vegetation (Hashim et al., 2019). The classi cation was done based on NDVI threshold value of < 0.25 for non-vegetation, 0. 25 to 0.5 for low vegetation, and > 0.5 for high vegetation using reclassify tool of Arc GIS software (Table 3). The availability index of urban green space is calculated by the share of land allocated to urban green space per area of a given reference surface (Le Texier et al., 2018). In this research work, availability index was computed for both the inner and outer urban areas of Debre Berhan town. The inner areas are the developed part of the town, and the outer areas are those which are currently being used for urban agriculture. The availability was also analyzed using distance from the Central Business District (CBD). In the study, the area around Semayawi (Blue) building was selected as a CBD.

Land use composition
The share of each land use classes in the land use plan of Debre Berhan town was used as an input to quantify land use composition. The composition analysis was computed for both inner and outer areas of the town.

Urban green space supply evaluation
Urban green space supply evaluation was analyzed according to the new national urban planning strategy of Ethiopia (30% coverage) and WHO (9m 2 per person) standards. This analysis helps us to examine the performance of study area in the provision of urban green space for its residents.

Urban green space composition
According to the land use plan of Debre Berhan town, there are 13 land use types which cover a total area of 5710ha. From these land use classes, urban agriculture covers the largest area of the town with 71.02% of land. The smallest land use type is mixed use with 0.23% of the town. The share of green space in the land use plan of the town is about 2.19% which is by far below the national green space supply standard (30% coverage) ( Table 4). In this research work, we classi ed our study area in to two. These are the inner city which is purely nonagricultural land use and its periphery which is purely urban agricultural land use. In the inner city, residential land use is the largest followed by service and vacant land with 463.71ha (28.02%), 365.55ha (22.09) and vacant land 220.01ha (13.3%) respectively (Table 5). Urban agriculture is the only land use in the peripheral area of inner city. The small share of urban green space (2.19%) in the land use plan of the town indicates overlooking of UGS allocation in the town. Generally, the existing UGSs are not evenly distributed and planned properly considering the national and WHO standards of green space provision (Fig. 2).

Urban Green space availability
Green space availability is a measure of green space provision of an area. It is quanti ed by using availability index which is the ratio of urban green space and total reference surface (Le Texier et al., 2018). The NDVI calculation from Sentinel 2 and its classi cation result shows the availability of green space which is not incorporated in the land use plan of the town. Thus, the result suggests a mismatch between UGS on the land use plan and UGS in reality (Fig. 3). The availability analysis was done considering the minimum NDVI threshold value of 0.25 for small vegetation and 0.5 for high vegetation. According to the result of our analysis, kebele 05 has the largest availability of UGS with a share of 32.18% followed by kebele 07 and kebele 03 having 21.24% and 20.81% respectively. The UGS availability analysis result asserted that the old part of the town (kebele 01 and 02) have a minimum supply while the recently developed part of the town has better provision of urban green space (Table 6 and Fig. 4). In line with this, Azagew and Worku (2020) found that the inner sub-cities of Addis Ababa city have a small amount of urban green infrastructure. However, in developed countries, studies revealed that the availability of higher green space in the inner city (Dobbs et al., 2014). The distribution of UGS was mapped for the town in general, the inner city, and its outer part (Fig. 5a, 5b, and 5c). The comparative analysis of UGS for the inner part of the town and its periphery shows a continuous increment of small vegetation towards the town boundary. In contrast, the coverage of high vegetation shows insigni cant reduction towards the periphery (Fig. 5d). Likewise a study by Woldesemayat and Genovese (2021) a rmed that inner parts of the urban centers in developing countries have low supply of UGS, which increases towards the peripheries.
Furthermore, green space density was calculated for each kebele. Green space density is the ratio of green space and the total area of reference surface. The result revealed that kebeles 05, 07, and 03 have higher urban green space density with respective values of 0.32, 0.21, and 0.21 (Fig. 6, Table 7). The high UGS density in Kebele 05 is mainly ascribed to its less suitability for urban development due to the rugged topography of the area. In this regard, slope steepness and terrain irregularity positively in uences the extent of green space coverage and their quality (Davies et al., 2008).

Urban green space availability in terms of land use
The amount of urban green space in each land use was analyzed using land use plan of the town and the NDVI classi cation result. According to this analysis, in the inner part of the town the service, green area, and vacant land are the rst three largest land uses with 30.12%, 30.06%, and 13.2% of UGS coverage respectively. In this part of the town recreation, commercial, and administration land uses are least supplied with a UGS share of 0.11%, 0.26%, and 0.40% correspondingly (Table). Although recreation land use is expected to have better UGS supply, unexpectedly it has the smallest (0.11%) UGS share. In addition, there is only 30% agreement between green area on the land use plan and our NDVI analysis result. Thus, 70% of green space on the land use plan has not yet developed as of planned (Table 8). This exposes the seriousness of the problem on land use planning and its implementation. Moreover, it indicates underperformance of the town municipality in UGS supply and development. Furthermore, most of the green areas in our nding are informal greenery. Hence, our evaluation of UGS supply shows signi cant amount of informal greenery which is planned as vacant land (13.32% of the vacant land is informal green) (Table 8). Those informal greens will be converted to other land uses such as commercial, residential, manufacturing etc. Hence, the current informal green space may not be available in the near future with the horizontal expansion of the town. In agreement with this, in most cases there is reduction of green spaces when urban centers expand towards the peripheries (Shen et al., 2017;Azagew and Worku, 2020).
Similarly, continuous urbanization and suburbanization together with the growing infrastructure development causes a constant decrease and severe fragmentation of green landscapes (Van Herzele and Wiedemann, 2003).

Urban Green space supply evaluation
The performance of town municipality in UGS supply was evaluated against the new urban planning strategy of  In this research work we also evaluated the change in UGS supply by generating buffer distance from the CBD.
According to the result, the rst 1km 2 buffer areas from the CBD has the smallest share of UGS availability and the largest UGS supply gap. Generally with some irregularities there is an increment of UGS supply from the CBD towards the peripheries. (Fig. 7).

Conclusion
Based on the land use plan of the town, in its inner part Debre Berhan has 13 land use types of which residential land use covers the highest share (28.02%). The share of green space in the town land use plan is about 2.19% which is by far below the new urban planning strategy of Ethiopia (30%). The small share of urban green space in the land use plan of the town indicates planning gap in UGS provision and development. The NDVI calculation from Sentinel 2 and its classi cation result shows the availability of informal green space which is not incorporated in the land use plan of the town. Thus, the result suggests a mismatch between UGS on the land use plan and UGS in reality. The UGS availability analysis result asserted that the old part of the town (kebele 01 and 02) has a minimum supply while the recently developed part of the town has relatively better provision of urban green space.
In the inner part of the town the service, green area, and vacant land uses are the rst three largest land uses with 30.12%, 30.06%, and 13.2% of UGS coverage respectively. In this part of the town recreation, commercial, and administration land uses are least supplied with a UGS share of 0.11%, 0.26%, and 0.40% correspondingly. Although recreation land use is expected to have better UGS supply, unexpectedly it has the smallest (0.11%) UGS share. In addition, there is only 30% agreement between green area on the land use plan and our analysis result. Furthermore, most of the green areas in our nding are informal greenery which will be converted to other land uses in the near future. UGS supply evaluation result based on the national and WHO standards a rmed underperformance of the town municipality in achieving expected UGS provision and development. Thus, considering all these issues the concerned planning department is expected to revise the existing land use plan speci cally in relation to urban green space provision and supply.

Declarations
Funding (Not applicable) Figure 1 Location Map. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 2
Land use composition (a is general composition, b is inner composition and c is outer composition). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. UGS availability index map. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 5
Green space availability; where a is general availability map, b is core availability map , c is outer availability map, and d is change of UGS availability from core to outer areas. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 6
UGS density map. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.
This map has been provided by the authors.