Detection and Classification of Embung Land Cover using Support Vector Machine

The agricultural sector is the mainstay sector in the economy of Malang Regency. However, Malang Regency has experienced a decrease in rice harvested area caused by drought. One of the Government's efforts to overcome this is by carrying out embung for agriculture. The use of remote sensing technology is one of the practical tools to monitor the phenomenon of change that occurs continuously and in a large area, in this case, the reservoir. This study aims to determine and analyze the use of SVM classification in satellite imagery to detect embung in Malang Regency. This research uses PlanetScope satellite imagery and Support Vector Machine (SVM) to classify land cover types. This research consists of three main tasks: satellite image preprocessing, satellite image classification, and land cover detection. The results showed that the increase in the number of sample areas in the SVM algorithm impacted the computational time and accuracy of the embung classification. The number of sample areas was small, the computation time was 16 seconds, and the accuracy was 0.5641. While the number of sample areas is large, the computation time is 307 seconds, and the accuracy is 0.7093.