10:45 AM - 12:15 PM
[HTT17-P01] Preliminary evaluation of coastline and tidal flat extraction using wide-area and high-resolution characteristics of the Advanced Optical Satellite, ALOS-3
Keywords:ALOS-3, WISH, Near-Infrared, coastline, tidal flat
ALOS-3 was launched in FY2022, and full-scale observation operations will begin as a successor to the wide-area and high-resolution optical mission of ALOS, which ended operation in 2011. ALOS-3 is equipped with a wide-swath and high-resolution optical imager (WISH). The spatial resolution has been improved to 0.8m (panchromatic) and 3.2 m (multispectral) compared to ALOS. In particular, the multispectral observation wavelength band of ALOS-3 was AVNIR-2's visible bands (blue, green, and red) and near-infrared bands, but these 4 bands were added with the addition of 2 bands, Coastal and Red Edge, for a total of 6 bands. The ALOS-3 mission is expected to be used for environmental monitoring of coastal areas. In this study, as a viewpoint of coastal environmental conservation, we will report the possibility of extracting coastlines and tidal flat areas using simulated satellite images equivalent to the observation performance of ALOS-3 prior to full-scale operation. The testing target area was Yakata-Katabaru, Onna Village, Okinawa Prefecture. This area is a place where the dry out area appears clearly over a wide area at low tide. Therefore, it was selected as a test site because it is possible to evaluate the impact of high and low tide levels on image analysis results, and because it can be used as ancillary information by conducting field surveys in this area in the past. The satellite image uses Maxar Technologies' Worldview, which has a specification close to that of ALOS-3 WISH. Then, we use near-infrared image only, which is effective for separating land and water area because of its optical characteristics. It was resized to 3.2 m. The analysis processing procedure uses a single image of the near-infrared band, applies the respective methods of ISODATA and Otsu. It determines the water boundary from the calculated threshold, and outputs it as a shape file. Next, it is evaluated by comparing the output Shape file with the coastline data of the national land numerical information. The distance difference between the Lines represented by the Shape file is calculated, but here, the Benchmark (BM), which is the reference for distance measurement, is set as a point with high visibility on the landside on the image, and the straight-line distance between the BM point and the output result and national land numerical information coastline data is measured. The difference in the measurement distance from BM of the output results based on both ISODATA and Otsu methods for water boundaries is generally within 1-2 m, and there is no significant difference between the methods. The output results (results of the Otsu method) and the national land numerical information coastline data were evaluated according to the characteristics of the coast of the verification site, and a difference of 10~20m was observed in the rock area, and although it was generally within 2~3m on the beach, there were places where a difference of 5~6m could be confirmed. Although it was less than 1 m in the port seawall, there was a difference of about 9 m in some places near the fishing port. In the estuary area, it is generally within 2~3m, but a peculiar difference of 15m was confirmed in some places. The reason for the difference in rocky areas is that the national land numerical information is more delineated towards the sea, but the output results from satellite images suggest that it is due to the difference in the area along the land side. The extraction of the coastline at test site uses satellite images observed at the time when the tide level was relatively high. Conversely, using images of the date and time when the tide level was low, tidal flats were extracted using the same process as the coastline.