DUAL-CHANNEL MODEL FOR SHALLOW WATER DEPTH RETRIEVAL FROM WORLDVIEW-3 IMAGERY A CASE STUDY OF KARIMUNJAWA WATERS, CENTRAL JAVA, INDONESIA

Abstract

This research aims to estimate shallow water depth using Worldview 3 satellite imagery and dual-channel models in Karimunjawa waters, Central Java – Indonesia. To build dual-channel models, we used spectral data that had been validated in the field. Twenty-three depth data were recorded synchronous to the spectral data used in forming the semianalytical dual-channel models. Twelve models were tested using 633 depth data with a non-linear model using multiple polynomial regression analysis degrees 1 and 2. This research has shown that the proposed model has been confirmed to improve depth accuracy. Models using blue and green channels of Worldview 3 image result in good accuracies especially for estimating depths with interval from 5 to 20 meters with RMSE of 1,592 meters (5–10 meters), 2,099 meters (10–15 meters), and 1,239 meters (15–20 meters). The wavelengths of two channels have a low absorption rate to penetrate deeper waters than other wavelengths. The research also finds out that there are still models that meet the IHO standard criteria.