Classification of aquatic macrovegetation and substrates with airborne lidar

Abstract: This study evaluates the potential to use waveform data from bathymetric airborne lidar for mapping seabed substratum and vegetation using a site in the Northern Baltic Sea as a study area. One objective of the study was to develop classification procedures including accurate corrections of waveform data for environmental and lidar system factors, to make them useful for mapping areas with variable water quality. Other objectives were to evaluate the classification accuracy using a combination of depth-derived variables and additional waveform variables, and to compare the results with the classification accuracy obtained when classification is performed using depth-derived variables only (without additional waveform variables). The analysis was based on two waveform variables (bottom pulse width and pulse area), in addition to two depth-derived variables (slope and depth standard deviation). The classification was performed using a model-based maximum likelihood approach. The classification models were created and evaluated with lidar data taken from locations documented with underwater video. The results show that inclusion of the waveform variables significantly improved the classification accuracy. Classification of the seabed into three classes (hard substrate; soft substrate with high vegetation; and soft substrate with low vegetation) had an overall accuracy of 86% when evaluated with an independent data set. This highlights the potential of data from airborne lidar, including waveform data (bottom pulse width and pulse area), for mapping shallow seabed habitats in coastal areas. [Copyright &y& Elsevier]