Enhanced Land and Water Classification Results of Dual and Triple Wavelength LiDAR Data
Rapid mapping of shore and coastal areas become an indispensable task for the local authority and military for coastal management and post-disaster management. For instance, some near-shore areas suffer from running out of sand along the coastal line, since back-to-back storms and hurricanes take away the beaches from the coast that drastically influence their local and tourist industries. Also, the basic binary classification of our world between land and water is critical for modelling of water flow, flood behavior and other hydrodynamics. Existing approach of near-shore mapping mainly relies on traditional field survey. In remote area, satellite remote sensing or aerial photogrammetry techniques can aid in the land/water delineation through manual digitization or semi-/ automatic data processing techniques. The drawbacks of human-computer interaction or manual survey require extensive labor headcount. As well, the lack of 3D information and the relatively coarse spatial resolution provided by the passive remote sensing techniques may not fulfil what the authorities are looking for: a rapid and precise solution with minimal manual input. Thus, an automatic classification approach of land and water is needed more than ever.
A multispectral LiDAR, like the Optech Titan, is usually considered to have at least three wavelengths. However, there have been a great many dual wavelength bathymetric systems since the 1970s. Some of these have been collinear and some have been biaxial in nature. This presentation will cover the results of land-water classification from a triple wavelength LiDAR (532 nm, 1064 nm and 1550 nm), the Optech Titan and also from another two wavelength bathy LiDAR (532 and 1064 nm), the CZMIL system. A comprehensive data processing workflow is developed by making use of both the intensity and elevation data from the dual/triple wavelength LiDAR data. A comprehensive analysis will be presented to assess the capability of using individual and dual/triple wavelength LiDAR data for improving the mapping accuracy. In addition, the data processing workflow is being tested under different environments, including beach, man-made shore, rocky shore, marsh area/wetland, natural shore with land depression, inland waterway, river, stream and nullah. This presentation aims to demonstrate and promote the use of multispectral LiDAR that can significantly help to improve the automation and accuracy of water mapping over the traditional passive remote sensing approach.