SPATIOTEMPORAL ANALYSIS AND WATER QUALITY PREDICTION MODEL NEAR A NICKEL MINE UTILIZING DRONE MULTISPECTRAL IMAGING

• Main Author: Setyo Kuntoro, Wahyu
• Format: Theses
• Language: Indonesia
• Subjects: Teknik saniter dan perkotaan; teknik perlindungan lingkungan
• Online Access: https://digilib.itb.ac.id/gdl/view/77961
• Institution: Institut Teknologi Bandung

Changes in land use due to community plantation activities and nickel mining activities are thought to have put pressure on the water quality of the Akelamo River. Long-term water quality monitoring is needed to provide basic information in an effort to determine environmental management policies. In-situ and laboratory measurement methods generally require a relatively long time and cannot provide a detailed spatial distribution of water quality. The use of satellite imagery has been extensively developed to address these challenges. However, the high resolution of satellite images and the intermittent data availability due to cloud cover render satellite imagery inapplicable in many locations. Currently, drone technology equipped with multispectral cameras is capable of providing image data with higher spatiotemporal resolution. In this study, a machine learning algorithm, specifically supervised learning, is employed to predict water quality based on multispectral drone image data using a point and buffer scheme with a 1-meter radius for extracting reflectance values. The research results indicate that the linear regression model for predicting total suspended solids (TSS) and turbidity yields the highest coefficient of determination (???? 2 ) of 0.79. The use of the buffer scheme in extracting drone reflectance values significantly enhances the ???? 2 value. The optimal prediction of optical water types (OWT) at the Akelamo River estuary is achieved using the random forest (RF) algorithm. Clear water and very turbid categories are dominant OWT types identified in this study. Furthermore, the spatiotemporal distribution of TSS and turbidity values at the Akelamo River estuary is heavily influenced by rainfall conditions in the catchment area. Consequently, multispectral drone imagery can be employed to create a water quality prediction model at the Akelamo River estuary, providing a more detailed and accurate spatiotemporal analysis.