Vegetation Cover Change and Its Diversity in Urban Areas of Medan

Vegetation plays an important role in maintaining the environmental quality of urban areas. Increase in population and development of cities has led to land conversion with lesser vegetated areas. Land cover change analysis in urban areas is needed, especially for urban regional planning with green open space consideration. This research was conducted to analyze urban vegetation cover and its changes in two sub-districts of Medan between the years 1999 and 2019. Normalized difference vegetation index (NDVI) and change analysis were conducted in the research. The diversity of plant within this areas was observed. The results showed changes in vegetation cover areas in the mentioned years. In 1999, most of the areas were under a highly dense vegetation class while in 2019, they were under a low-density vegetation class. This indicates a decrease in vegetation cover due to changes to non-vegetation cover or land cover areas with less vegetation. There are a diverse of plants within the area such as paddy, cassava, corn etc and also many tree species. It is recommended to optimize the land by replanting in the area with no or less vegetation to maintain the environmental quality.


Introduction
Urban areas are a place of more than half of the world's population. By 2050, it is expected 66% of the population will be living in those areas. The continued population growth and urbanization have caused many problems that hamper environmental sustenance (Uçar et al 2020).
Cities as a mosaic of habitat that change through time. It shows environmental heterogeneity due to variation within and even among cities (Rivkin et al (2019). Guha et al (2020) mentioned the urban landscape is considered the most complex and heterogeneous landscape among the different land surface features.
Complex environmental problems have accompanied the enormous growth of cities. City environment is in uenced by global and local climate changes, pollution from transport, industries, and local heating sources (Rozova et al 2020). Urbanization is considered as one of the main factors affecting global change. (Qing et al 2020). Climate change and urbanization are the two primary drivers that can alter vegetation growth processes in the urban environment .
The development of urban areas was led by the increase of population (Dewantoro 2020). It is in relation with the urbanization within the areas. It gives an impact to land cover change due to the need for housing and economical related actions. As Frimpong (2021) there are transformation of various Land Use Land Cover (LULC) types into urban/built-up areas due to rapid urbanization. Enoguanbhor (2019) has also mentioned that a signi cant contributor to land cover change is a rapid urban expansion.
Vegetation formations are an important component in the urban structure due to a wide range of ecosystem services in the area. Rozova et al. (2020) contested one of the important functions to improve the environmental and residential quality of the city is climate modi cation.
Providing information on land cover and use change, including in vegetation areas, is a must for improving urban spatial planning in relation to the improvement of environmental quality. Remote sensing and geographic information systems are powerful scienti c tools for supporting analysis and monitoring land use and land cover change. Sha  Medan is one of the major cities in Indonesia's that faces the same problem as any major city in the world. An increase in the city's population and development has led to the conversion of vegetated land for other purposes, especially settlements and buildings. Medan has 21 sub-districts with a total area of 265 km 2 . Each sub-district has its own area characteristics. Medan Baru and Medan Selayang subdistricts are among the areas that are overgrowing and experiencing many land cover changes, including increase in built-up areas and decrease in vegetation areas.
NDVI and change analysis were conducted in this research. The NDVI has been proven to accurately detect various land cover changes ). To detect changes in land cover and NDVI quickly and accurately, this research aimed to analyze the vegetation cover areas and its changes in Medan Baru and Medan Selayang's sub-district. Changes between the years 1999 and 2019 were analyzed. The NDVI is a powerful tool for understanding past vegetation, monitoring its current state, and predicting its future (Xing et al. 2020;). Phinzi and Szabo (2020) argued that the NDVI remains one of the most widely used tools to assess atmosphere and bare soil background through imagine spectral properties, facilitating image interpretability. It has been used for land use/built-up cover areas.  found vegetation cover changes to be related to differences in urbanization rates, gross domestic products, population densities, and stages of urban development among the cities. Landuse change is one of the main impacts of human activities and profoundly impacts vegetation change ). To ensure that ecosystem services are provided, vegetation cover should be considered in urban planning in urban core areas and peripheral areas ). Therefore, having information on vegetation cover change in urban areas will be useful in planning a better environment by assuring the optimal existence of green space within the areas.

Materials And Methods
The research was conducted in the sub-districts of Medan Baru and Medan Selayang (Figure 1). Processing and analysis of the research data were carried out in Forest Management Laboratory, Faculty of Forestry, Universitas Sumatera Utara.
Landsat images of the years 1999 and 2019 were downloaded through www.earthexplorer.usgs.gov. The software used for remote sensing and GIS analysis were Erdas Imagine 8.5 and ArcGIS 10.3. Ground check had been conducted in the research areas. Radiometric corrections were performed using Erdas 8.5 to correct errors that occurred in satellite imagery by sharpening the contrast. Image cropping was conducted to obtain a speci c research location.
NDVI transformation was carried out using band infrared and red of Landsat. The principle of the NDVI is to measure the level of greenness. The intensity of greenness is correlated with the density of vegetation crown, which is related to leaf chlorophyll content. A comparison between the red (R) re ectance and the near-infrared (NIR) parts of the electromagnetic spectrum was calculated. The selection of these wavelengths results from the absorption and re ection characteristics of vegetation. Due to absorption processes, especially in the red part of the electromagnetic spectrum, it was associated with the leaf chlorophyll content. In the near-infrared part of the electromagnetic spectrum, there is a very strong re ection that corresponds to multiple re ections in the leaf inner cell structure due to the cells' water content (Juergens and Meyer-Heß 2020).
The NDVI can estimate vegetation coverage. If the value of the NDVI is greater than others, this indicates that vegetation coverage is better (Wen and Zhang 2020). The more the NDVI value tends to +1, the more it is related to vegetation cover and its vigor (Juergens and Meyer-Heß 2020). The value ranging between -1 to +1 resulting from this NDVI transformation has a different presentation on its land use. Clouds, water, and non-vegetation objects have the NDVI value of less than zero. The greater the value of the NDVI, the higher the density, and vice versa for a lower value. The following formula was used: where, IR = re ectance value of infrared band R = re ectance value of the red band With the aid of NDVI transformation, information can be obtained on vegetation elements' existence in the whole area. The ground check gave valuable information on land use and land cover classes found in the eld. So, we can obtain the range of values of the NDVI within some classes. The classi cation was conducted to divide the areas into different vegetation density classes. The range of values was de ned by considering the vegetation existence in the eld and the NDVI values. Phinzi and Szabo (2020) stated that different NDVI threshold values were used for classifying various land-use/cover classes (water bodies, badlands, bare soil, and built-up land, agriculture, grassland, and forest).
In this research, the location was classi ed into ve classes, namely non-vegetation, low dense, medium dense, dense, and high dense classes. The classi cation can show the difference in vegetation density within the whole area . In order to see the change in vegetation density, change analysis was conducted (Zaitunah et al. , 2020. From the analysis, we were able to assess the change of vegetation cover within the observation years. This research provides information on the change of vegetation cover in any speci c areas in the research location.

Distribution of NDVI values in Medan Baru and Selayang Subdistricts
The NDVI compares the total amount of visible red light absorbed with the amount of re ected nearinfrared light by a surface (Fusami et al. 2020). It basically uses a mathematical ratio to compare the amount of absorbed visible red light and the re ected near-infrared light. In the year 1999, the Medan Baru and Medan Selayang sub-districts had the highest NDVI value in the range > 0.4, which is 609.66 ha (29.96%) of the total area, while the smallest in the range of NDVI < 0.1, which is 151.94 ha (7.47%) of the total area (Table 1). These are quite different from the year 2019 (Table  2), which shows the NDVI of the Medan Baru and Medan Selayang subdistricts, where they had the highest NDVI value in the range of 0.1-0.2. It comprised the area of 768.09 ha (37.73%). The smallest NDVI range > 0.4 covered 226.04 ha (11.1%) of the total area. A comparison of each NDVI area class is illustrated in Figure 2.  Based on eld checks, there are 9 land cover classes, namely buildings, roads, mixed gardens, oil palm, settlements, trees, grass, rice elds, and shrubs. The division of objects and the range of NDVI values can be seen in Table 3.  The denser the vegetation, the higher the NDVI value. The low the NDVI value will represent the low vegetation density. There are variations of the NDVI values in each land cover class due to the variation of objects found in the eld; for example, there are trees alongside the road and trees in between houses in settlement areas. Some trees can also be found near buildings, and some near rice elds, grass land, and shrub. The higher the NDVI values found in areas with trees, shrub and oil palm. The lower value is found in settlements and roads.
The spatial resolution of Landsat images contributes to the variation of NDVI values of each land cover classes as there are many objects found in areas of 30 m. There is also due to existence of vegetation in each land cover classes as mentioned above.

Non-Vegetation Class
Non-vegetation classes are roads, dense settlements, and tall buildings. There is little or no vegetation in this class. The non-vegetation class consists of residential areas, business areas represented by tall buildings near the main road, and congested streets in the yover area. Some campus areas of the University of North Sumatra, which is located in Medan Baru, show non-vegetation density classes. The library, faculty building, and administration o ce center building belong to this class. Areas that belong to the non-vegetation class illustrated in Figure 5.

Low Dense Class
In the low dense class, the vegetation density is low. Areas that belong to this class are roads, buildings, settlements, and bare lands with less vegetation. Low dense class is found in some areas in the city surrounded by buildings and existing vegetation. There are also trees along the path of the road and around houses. The species found here are Plumeria rubra, Mangifera indica, Pterocarpus indicus, Gnetum gnemon, Polyalthia longifolia, and Areca catechu. In the campus area of the Universitas Sumatera Utara (USU), there are some buildings with trees around them. The species found here are Terminalia catappa, Casuarina equisetifolia, Mangifera indica, and Ficus benjamina. Areas that belong to the low dense class can be seen in Figure 6.

Medium Dense Class
In the medium dense class, the density of vegetation is rather high. There are roads, buildings, and settlements, where those covers have vegetation; for example, individual trees near those covers.
Medium dense class is found in settlements with a large yard with various trees. There are also gardens in the area of settlement.  Figure 7.

Dense Class
In the dense class, rice elds, shrubs, trees, and mixed gardens were found. People grows cassava, sugar cane, corn, banana and oranges. Trees are found around the areas. Areas that belong to the dense class can be seen in Figure 8.

High Dense Class
While in high dense, there are dense vegetation found including trees, oil palm, rice elds, very dense shrubs, and mixed gardens.High dense class includes land overgrowing with trees accompanied by shrubs, land planted with oil palm, vast rice elds, and mixed gardens owned by the community planted with various plants such as Citrus nobilis, Zea mays, Cymbopogon citratus, Solanum torvum, Artocarpus altilis, Saccharum o cinarum, Cocos nucifera, Etlingera elatior, Theobroma cacao, Carica papaya, Manihot esculenta, Musa paradisiaca, and Psidium guajava. The high dense class was found in the campus forest which is consisting of some species. The species are Swietenia mahagoni, Artocarpus heterophyllus, Psidium guajava, Paraserianthes falcataria. In USU administration center park is also found Terminalia catappa, Terminalia mantaly, Ficus elastica, Swietenia mahagoni, Syzygium oleana, Ficus benjamina, Durio zibethinus, and Pterocarpus indicus. Areas that belong to the high dense class can be seen in Figure 9.
The increase of population growth and development of the Medan areas as a big city has increased builtup areas and settlements. It decreases vegetated land. Changes in rice elds and gardens into housing areas are examples of conversion. The ground check shows dense settlements and multi-storey buildings are found more than vegetated lands such as trees, rice elds, and community-owned gardens. The decrease in vegetation areas could have a bad impact on the quality of the environment, i.e., problems of ood due to a decrease in water absorption areas.
Many vegetated lands were converted into housing and buildings. Moreover, some vegetation areas were replaced by roads and public facilities. Vegetation cover changes could lead to environmental problems, such as the incident of ood in the rainy season, bad air quality due to pollution, and health problems. In urban contexts, vegetation surfaces are very important for the well-being and health of the urban population. The NDVI is often correlated with socioeconomic and/or sociodemographic data to demonstrate the inequality in environmental settings that themselves in uence individual health and questions of environmental justice (Juergens and Meyer-Heß 2020).
Medan has faced some ood incidents, especially in the areas near the river. Some of the areas have no or less vegetation on the side of the river. This situation can worsen unless there is an improvement of areas by replanting trees and better planning the areas. This requires preparing both the community and land condition. Land-use planning process was used to provide alternative and possible land uses as well as management activities to ensure land-use sustainability . It is necessary to initiate efforts and actions to preserve the environment to prevent further damage. Four criteria were used for determining suitable plants that could be planted as well as cultivated by the community: A big part of the urban land occupation is composed of a highly dense buildings and large paved and impermeable areas, which support the high increase in the air temperature in urban centers, which creates microclimates in different zones of the city, causing heat islands (Giacomelli et al. 2020).
In that case, the modeling of the urban sprawl effect on vegetation-cover is realized by the NDVI) After observing and characterizing the areas altering by the urban expansion, the results displayed that in 17 years, the urban growth of Annaba decreased the vegetation cover by 28.50 % (Saouli et al. 2021).
Urban studies shows wide scope of discussion with the same aim that is to overcome problems and built a good environment for man and nature. Such studies including study on urban resilience (

Conclusions
The research found a decrease of high and dense vegetation areas into lower and non-vegetated areas within 20 years. This indicates a decrease in vegetation cover due to changes to non-vegetation cover or land cover areas with less vegetation. The NDVI could further explore the change of vegetation cover in urban areas and take part in monitoring and urban planning. In the research area, it is recommended to optimize the land by replanting the area with no or less vegetation and maintaining the vegetated areas to improve the environmental quality. Research 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.    The low dense class.

Figure 7
The medium dense class on-site research Figure 8 Dense class areas.

Figure 9
High dense class