Passive remote sensing using drone and hd camera for mapping surf zone bathymetry

Abstract

This paper presents the shallow-water bathymetry survey using high definition color images obtained at the altitude of 150 m above sea level using a drone. Image processing algorithms using multi-variable linear regression applied to red, green, blue and grey band images obtained from HD camera has been developed. Optical passive remote sensing analysis conducted at Wolpo beach showed promising results. Estimated water depths of within 6m water depth in one HD camera image with dimensions of 270m by 180m showed the coefficient of determination of 0.990, RMSE 0.144m, and the maximum error of 0.38m compared with water depth surveyed through manual as well as ship-board echo-sounder measurements. In order to increase the accuracy of estimated water depth, we also applied artificial neural network(ANN) technique. Estimation of water depth using ANN showed improved results with the coefficient of determination of 0.995, RMSE 0.088m, and the maximum error of 0.217m. Sources of error in the estimation of water depth include sunglint, waves and breaking waves, as well as turbidity and sea bottom characteristics. Sun glint, waves and breaking waves can be removed rather easily through relative luminance and time average filtering technique, respectively. However, errors caused by turbidity of water and sea bottom characteristics are relatively difficult to calibrate. We could identify the large errors in the depth estimation due to seagrass habitat on the sea bottom which cause reduced reflectance of the RGB band. By applying the Beer-Lambert law on each RGB band separately, we could pinpoint the existence of the seagrass on the sea bottom.