Identifikasi Tutupan Lahan Menggunakan Support Vector Machine di Kawasan Perkotaan Cekungan Bandung

Abstract

The current mapping challenge is to obtain land cover information with a very high level of accuracy. This information can be useful for regional development, especially in the Bandung Basin Urban Area (KPCB). One of the problems in KPCB is spatial planning issues regarding land cover development. One approach to obtaining land cover information is by utilizing remote sensing technology using the Support Vector Machine (SVM) method. The research method employed in this study is remote sensing, with a unit of analysis based on Regency/City administrative boundaries. The data used in this research consists of Landsat imagery recorded in 2023. The aim of this research is to classify land cover and assess accuracy using the SVM method in KPCB. The SVM results provide information on six land covers: water bodies, secondary forests and mixed gardens, wet agricultural land, dry agricultural land, plantations, built-up land, and primary forests. The accuracy test yielded an overall accuracy of 90% and a kappa value of 0.88. The obtained accuracy in land cover classification is very high, indicating that the data used can be employed for further analysis. For instance, spatial analysis reveals that built-up land development in KPCB tends towards the south, highlighting the phenomenon of urban sprawl. Thus, the utilization of remote sensing technology for land cover information can offer valuable policy insights for addressing spatial planning and development in KPCB.