Abstract:
Pak Phanang bay is an important area in both terms of economics and ecosystems. Because, it hosts a shipping route from Gulf of Thailand into Pak Phanang river. In the past, a canal which located in the center of the bay was canalized. But, from the study in 2019 found that the depth in the canal had changed from the criteria of the Marine Department which defines the standard depth equal to 4 meters. There were changes in both the deeper parts from erosion and the shallower parts from the sediment deposition. According to the changing of the depth in the canal that causes problems for navigation, there is a project to assess the environment of Pak Phanang Bay for new trench dredging in the future. So, this study will make the bathymetry and study the geomorphology of Pak Phanang bay to be a supporting information for the future construction project by using satellite derived bathymetry together with machine learning. Sentinel-2A band 5 was choosed to be a satellite data. Gaussian Process Regression (GPR) is one of the machine learning which was used for creating bathymetry from pixel values. Another machine learning is K-means clustering which was used To help differentiate the depths. The result of from this study proves that this method was able to map the depth of the canal in shallow area and the depth relative to the real data. But the GPR also generated noise and still unable to distinguish the area with little change in depth. Geomorphological processes can be divided for two areas including shallow area and deep area. In shallow area are formed by the deposition of the sediments from the Pak Phanang rivers and the suspended sediments in the seawater. And the deep area was inferred the cause by the flowing of the flows in both sides of coasts causing the current to erode in this area.
