DEVELOPMENT OF COASTAL FLOOD SUSCEPTIBILITY ASSESSMENT THROUGH REMOTE SENSING AND MULTIMACHINE LEARNING APPROACH IN SOUTHEAST ASIAN COASTAL AREAS

• Main Author: Putri Adillah, Kurnia
• Format: Theses
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/81392
• Institution: Institut Teknologi Bandung

Coastal areas are vulnerable to disasters, one of which is coastal flooding, which has the most significant impact compared to other disasters due to its higher frequency of occurrence. The exposure to coastal flooding is exacerbated by other factors such as sea level rise, temperature, storm intensity, land subsidence, and anthropogenic activities. One area particularly vulnerable to coastal flooding is Southeast Asia. This region has the longest coastline in the coastal, stretching 234,000 km, and it is estimated that 77% of its population lives in coastal areas. The impact of coastal flooding is significant as it can threaten economic aspects, infrastructure, the environment, and residential areas, making mitigation efforts a very important aspect to undertake. Based on previous studies, there is a great potential to improve methodologies in assessing coastal flood susceptibility. Therefore, this study aims to develop a comprehensive coastal flood susceptibility assessment model using multi-machine learning methods to enhance the objectivity of the model. To achieve this goal, this research has the main objective of developing a comprehensive and objective coastal flood susceptibility assessment system to support decision-making in coastal flood mitigation and management. Furthermore, this study sets three supporting objectives: first, to analyze Normalized Difference Water Index (NDWI) changes, exposure to sea level changes, and proximity to the coastline and industrial areas. Second, to build a coastal flood susceptibility assessment model from three machine learning algorithms and integrate them. Third, to analyze susceptibility values for built-up and agricultural land cover types to determine priorities for costal flood management and mitigation in both types of land cover. The novelty of this study lies in the development of a coastal flood susceptibility assessment that utilizes remote sensing data and multi-machine learning and combines various predictor variables consisting of socio-economic, environmental, and disaster-driving groups. The study concludes that there are significant variations in NDWI changes in Southeast Asia's coastal areas. The analysis shows that most areas are experiencing an increasing trend (28.35%), a decreasing trend (31.84%), or remaining stable (39.81%). The most extreme NDWI increase was recorded in Indonesia, indicating extreme water-related changes in the distance between industrial areas and sea level exposure revealed that not all areas experiencing high sea level changes are close to industrial areas. Susceptibility modeling using multi-machine learning methods proved to increase accuracy, with an Area Under Curve (AUC) value reaching 0.905. In the high susceptibility category, the Gradient Tree Boosting (GTB) method achieved the highest coverage (25.20%), followed by Classification and Regression Trees (CART) (20.84%), multi-machine learning (17.85%), and Random Forest (12.46%). The distribution analysis of districts/cities showed that as many as 567 districts/cities in urban areas (31.7%) and 254 in agricultural areas (20.8%) have very high susceptibility to coastal flooding. The Philippines, Vietnam, and Indonesia were identified as countries requiring priority in coastal flood mitigation based on susceptibility assessments. To address these challenges, several handling suggestions have been proposed. One of them is the creation of regulations related to groundwater exploitation in residential and industrial areas to reduce pressure on water resources. Additionally, infrastructure development such as barriers, levees, and dams is also considered important to enhance the resilience of the region against the threat of coastal flooding in the future. Implementing these measures can help reduce the negative impacts of coastal flooding and improve the safety and wellbeing of communities in Southeast Asia's coastal areas. The results obtained are expected to assist policymakers in making appropriate decisions regarding spatial planning and risk management in coastal areas. Furthermore, based on approaches tailored to land cover types, policymakers can formulate effective policies to reduce coastal flood risk and increase community preparedness. This list of priority areas also enables policymakers to allocate resources efficiently, focus on areas most in need, and develop adaptation strategies that can reduce the long-term impacts of climate change and sea level rise.