1:45 PM - 2:00 PM
[U05-06] ASSESSMENT OF LAND COVER AND URBAN SPRAWL IN JAKARTA MEGACITY (JABODETABEK), INDONESIA FROM 2000–2021
Keywords:land cover, remote sensing, shannon entropy, urban sprawl
In developing countries, urbanization is one of the challenges that affect the development of urban areas. This situation is accompanied by environmental degradation, and poverty, and is called urban sprawl. The Jakarta megacity, Jabodetabek (comprised of Jakarta, Bogor, Depok, Tangerang, and Bekasi), is characterized by polycentrism, showing the development of industries followed by population growth and land changes towards the city edge, which depicts the urban sprawl phenomenon.
In this study, land cover and urban sprawl assessment were conducted over two decades from 2000 to 2021. We used the Google Earth Engine with the Random Forest machine learning method for the land cover analysis. The primary data were obtained using Landsat Surface Reflectance (SR) Collection 1, Tier 1 in annual composites from 2000, 2007, 2014, and 2021. There were 3 land cover classes: vegetation, waterbodies, and built-up area with training sample performances per class. The sample was 250 random points of each land cover class per image. The sample points were divided into two categories: 70% for the training sample and 30% for the validation sample. The accuracy assessment used the Kappa Accuracy formula with a minimum value of 0.8. The urban intensity index was calculated to show temporal urban spatial expansion. The land cover images were then categorized into two areas: urban (built-up) and non-urban (vegetation and water bodies) in ArcGIS
For the urban sprawl, Shannon entropy was used to capture its expansion of it from the core to the periphery. To measure urban sprawl, several zones must be created within, covering the entire area. A concentric ring was used to represent the spatial context of urban sprawl. Buffer rings were created around the core city center (National Monument in Jakarta City), at 1 km intervals up to 76 km, and further divided into three main zones (core, fringe, or periphery). The core interval was extended to 30 km, the urban fringe was 50 km, and the periphery was the remainder. Zones 73 to 76 were eliminated because no urban areas were detected.
The result showed that the accuracy for the land cover assessment was higher than 0,9 so it can be used for the next analysis. The vegetation area is higher compared to built-up and waterbodies. The built-up areas (settlement and industrial areas) experience the highest gradual increase from 2000–2021 at 10.81%. Conversely, the vegetation (including agriculture and forests) suffered a gradual loss of 10.32%. Overall, the proportion of waterbodies has increased by 0.49%. The vegetation area is under strong pressure due to built-up area expansion. Meanwhile, the intensity index (UEII) showed that the study area had an expansion intensity index of <0.28, which is considered a very slow development.
Further to the analysis, The Shannon entropy value showed high sprawling characteristics from 2000 to 2021 because the value was close to 1. The relative entropy in the core and urban fringe was higher than that in the periphery. Although the relative entropy in the fringe is higher than in the core, their value is almost parallel to each other. The overall Shannon entropy value in the core indicates a slightly decreased value (-0.007) and the fringe was decreased to 0.006. It means the zone as the settlement and industrial center indicates progress compactness.
But The periphery area increased by 0.0591 in 2021. The fringe zone and periphery which consists of districts are strategic locations for developing the built-up areas. But actually, most of these areas have a local protection area such as a river border, green open spaces, and protected areas for their culture and heritage. The highest Shannon entropy has allowed the urban fringe and periphery to extend uncontrollably, leading to random and scattered development centers. This can endanger the protected areas. Therefore, urban planning for the creation of a compact city must be considered to reduce sprawl.
In this study, land cover and urban sprawl assessment were conducted over two decades from 2000 to 2021. We used the Google Earth Engine with the Random Forest machine learning method for the land cover analysis. The primary data were obtained using Landsat Surface Reflectance (SR) Collection 1, Tier 1 in annual composites from 2000, 2007, 2014, and 2021. There were 3 land cover classes: vegetation, waterbodies, and built-up area with training sample performances per class. The sample was 250 random points of each land cover class per image. The sample points were divided into two categories: 70% for the training sample and 30% for the validation sample. The accuracy assessment used the Kappa Accuracy formula with a minimum value of 0.8. The urban intensity index was calculated to show temporal urban spatial expansion. The land cover images were then categorized into two areas: urban (built-up) and non-urban (vegetation and water bodies) in ArcGIS
For the urban sprawl, Shannon entropy was used to capture its expansion of it from the core to the periphery. To measure urban sprawl, several zones must be created within, covering the entire area. A concentric ring was used to represent the spatial context of urban sprawl. Buffer rings were created around the core city center (National Monument in Jakarta City), at 1 km intervals up to 76 km, and further divided into three main zones (core, fringe, or periphery). The core interval was extended to 30 km, the urban fringe was 50 km, and the periphery was the remainder. Zones 73 to 76 were eliminated because no urban areas were detected.
The result showed that the accuracy for the land cover assessment was higher than 0,9 so it can be used for the next analysis. The vegetation area is higher compared to built-up and waterbodies. The built-up areas (settlement and industrial areas) experience the highest gradual increase from 2000–2021 at 10.81%. Conversely, the vegetation (including agriculture and forests) suffered a gradual loss of 10.32%. Overall, the proportion of waterbodies has increased by 0.49%. The vegetation area is under strong pressure due to built-up area expansion. Meanwhile, the intensity index (UEII) showed that the study area had an expansion intensity index of <0.28, which is considered a very slow development.
Further to the analysis, The Shannon entropy value showed high sprawling characteristics from 2000 to 2021 because the value was close to 1. The relative entropy in the core and urban fringe was higher than that in the periphery. Although the relative entropy in the fringe is higher than in the core, their value is almost parallel to each other. The overall Shannon entropy value in the core indicates a slightly decreased value (-0.007) and the fringe was decreased to 0.006. It means the zone as the settlement and industrial center indicates progress compactness.
But The periphery area increased by 0.0591 in 2021. The fringe zone and periphery which consists of districts are strategic locations for developing the built-up areas. But actually, most of these areas have a local protection area such as a river border, green open spaces, and protected areas for their culture and heritage. The highest Shannon entropy has allowed the urban fringe and periphery to extend uncontrollably, leading to random and scattered development centers. This can endanger the protected areas. Therefore, urban planning for the creation of a compact city must be considered to reduce sprawl.