Relative Water Column Correction Methods for Benthic Habitat Mapping in Optically Shallow Coastal Water

Relative water column correction (WCC) is a method to minimize the effect of water column energy attenuation on remote sensing images that requires fewer inputs compared to more robust absolute WCC methods, hence the broad applicability. Lyzenga developed the frequently used relative WCC method Lyzenga (Applied Optics 17:379–383, 1978) called the “water depth invariant bottom index” (DII). This method is known for its simplicity and accuracy. However, this method does have several limitations; hence, several relative WCC methods have been developed to improve it, among others, by Conger et al. (IEEE Transactions on Geoscience and Remote Sensing 4: 1655-1660, 2006) and Sagawa et al. (International Journal of Remote Sensing 31: 3051-3064, 2010). This research compared the accuracy of benthic habitat mapping based on applying the aforementioned relative WCC methods on WorldView-2 images to recommend the most effective and efficient relative WCC method for mapping the benthic habitat in a complex underwater topography environment. Random forest classification was used to classify water column corrected pixels into different benthic habitat classes. Confusion matrix analysis was used to assess the accuracy of classification results. This research showed that the improved relative WCC methods increased the mapping accuracy. The accuracy differences between these relative WCC methods are within 3.03%. Therefore, these relative WCC methods are applicable to high-transparency optically shallow water areas and complex underwater topography, such as Kemujan Island. Nevertheless, depending on the specific mapping objectives, it is recommended to use the improved relative WCC methods if the necessary inputs are available.