EXAMINATION OF INTER-IMAGE COMPATIBILITY OF WATER DEPTH PREDICTORS BASED ON VISIBLE AND NEAR-INFRARED IMAGERY

Abstract

Multispectral remote sensing methods of water depth can potentially play an important role in shallow water bathymetry. However, they have not been widely used because they require auxiliary information about optical properties of target water area for each image in order to build or calibrate the prediction model. In this study, we examined the inter-image compatibility of the prediction model of the popular method by Lyzenga et al. using three QuickBird images of different water areas. As a result, predicted and measured depths were found to be highly correlated regardless of the combinations of the image used for calibration and the image predicted. The interchangeability can be utilized to reduce the amount of the auxiliary information necessary to apply the method, and therefore to improve its feasibility.