DEVELOPMENT OF METHOD FOR MONITORING ROCK REEFS USING SATELLITE IMAGERY

Abstract

 Satellite imaginaries are expected as the base data for monitoring the deformation of rock reefs, because the most of important rock reefs locates in isolated islands. Satellite image classification and field surveys were conducted on rock reefs located in Inubousaki (Chiba Prefecture) and Bonten-no-hana (Kanagawa Prefecture), in order to establish a new monitoring method using satellite imagery. Four satellite image classification methods (K-means, threshold, maximum likelihood, and nearest neighbor) were applied to detect changes in the rock reefs captured via satellite imagery. The nearest neighbor method showed the most accurate results, and post-hoc detection using blue-band satellite imagery improved the accuracy of rock reef detection. Three-dimensional digital elevation models (3D-DEM) of the rock reefs were established based on the three-dimensional point cloud data acquired by a green LiDAR drone. 3D-DEM satisfactorily simulated the shapes of the rock reefs at the position where rock reef surfaces were covered by wave run-up. The accuracy of the geometric correction of the satellite images was the highest when the 3D-DEM was utilized as reference point data. The shapes of the rock reefs projected on the satellite image were estimated on 3D-DEM, considering the water level and sunlight direction when the satellite images were captured. The estimated shapes of the rock reefs were consistent with the shapes projected on the satellite images. The results suggest that 3D-DEM can provide good reference data when surveying the time-series erosion of rock reefs. Based on these results, a new rock reef monitoring method based on satellite imagery and 3D-DEM is proposed, and some recommendation for practical application are discussed.