Using Support Vector Machine To Identify Land Cover Change During Covid-19 Pandemic In Komodo National Park, Indonesia

Abstract

The Covid-19 pandemic affects many areas of life, including the tourism sector. Furthermore, it significantly reduced the number of people visiting tourist destinations, and the reduction has helped to improve the environment in the National Park. Therefore, this study aims to present a satellite image classification method using Support Vector Machine to identify changes in the vegetation area of Komodo National Park. The satellite image used was created with Google Earth Pro with a resolution of 1920 x 1280 pixels using data collected in 2019 and 2020 before and during the pandemic. This study focuses on six tourist destinations in Komodo National Park: Loh Liang, Loh Buaya, Padar Island, Kanawa Island, Pink Beach, and Loh Sebita. The image was pre-processed using radiometric calibration, atmospheric correction, and contrast enhancement. The results of the pre-processing showed that segmentation will be performed to distinguish the area between one class and another. Furthermore, the image will be classified into five classes using the Support Vector Machine, including Soil, Vegetation, Built-Up Area, Deep Water, and Shallow Water. The measurement of the area of vegetation from 2019 and 2020 using Otsu’s thresholding showed environmental changes. Meanwhile, environmental improvements occurred in seven areas in the vegetation area category, with a 31.86% rise from 2019 to 2020. The increase in the area of green areas in the Komodo National Park all because tourist restriction and there is no climate fluctuations during the time of study.

Keywords

• vegetation area change
• environmental recovery
• machine learning
• support vector machine
• komodo national park