Ứng dụng viễn thám và máy học trong giải đoán địa hình ven bờ tại khu vực cửa Tam Quan, Bình Định

Abstract

Coastal terrain data is crucial for coastal engineering and management research. Access to multi-temporal terrain data, combined with meteorological, hydrological, and oceanographic information, is essential for understanding coastal dynamics. However, traditional methods for acquiring coastal terrain data often face limitations due to resource constraints, high costs, specialized equipment requirements, and weather dependency. Remote sensing technology has emerged as an effective alternative, offering advantages such as rapid data acquisition, extensive spatial coverage, and access to historical imagery archives. This study utilizes remote sensing imagery to interpret the terrain of the estuary, coastal areas, and sand dunes at Tam Quan estuary in Binh Dinh province, Vietnam. Two interpretation methods are employed and compared: Stumpf's ratio formula and machine learning algorithms. Results indicate that both methods can effectively extract terrain information from remote sensing data. Stumpf's method achieves a correlation coefficient of RMSE = 0.73 compared to field survey data, while machine learning algorithms demonstrate superior performance, with the Random Forest (RF) algorithm achieving the highest accuracy (RMSE = 0.957). This research highlights the potential of remote sensing, particularly when integrated with machine learning, for multi-temporal terrain interpretation in coastal regions. This approach contributes to a deeper understanding of coastal dynamics and supports sustainable decision-making processes.