Classification of Boulders in Coastal Environments Using Random Forest Machine Learning on Topo-Bathymetric LiDAR Data

Abstract

Boulders on the seabed in coastal marine environments provide key geo- and ecosystem functions and services. They serve as natural coastal protection by dissipating wave energy, and they form an important hard substrate for macroalgae, and hence for coastal marine reefs that serve as important habitats for fish. The aim of this study was to investigate the possibility of developing an automated method to classify boulders from topo-bathymetric LiDAR data in coastal marine environments. The Rødsand lagoon in Denmark was used as study area. A 100 m × 100 m test site was divided into a training and a test set. The classification was performed using the random forest machine learning algorithm. Different tuning parameters were tested. The study resulted in the development of a nearly automated method to classify boulders from topo-bathymetric LiDAR data. Different measure scores were used to evaluate the performance. For the best parameter combination, the recall of the boulders was 57%, precision was 27%, and F-score 37%, while the accuracy of the points was 99%. The most important tuning parameters for boulder classification were the subsampling level, the choice of the neighborhood radius, and the features. Automatic boulder detection will enable transparent, reproducible, and fast detection and mapping of boulders.

The Use of Green Laser in LiDAR Bathymetry: State of the Art and Recent Advancements

“…This type of research on aquatic plant habitats is presented by [ 124 , 125 ] in which machine learning techniques are used (Random Forest). The authors use single wavelength or bispectral barymetric lidar.…”

3DMASC: accessible, explainable 3D point clouds classification. Application to bi-spectral topo-bathymetric LiDAR data.

“…Shallow submerged parts usually represent a significant source of uncertainty when using only topographic LiDAR data to study land-water interface dynamics (Lague and Feldmann, 2020). Combining high-resolution data about the submerged and emerged surfaces offers new opportunities to map habitats in fluvial (Fernandez-Diaz et al, 2014;Mandlburger et al, 2015;McKean et al, 2009;Pan et al, 2015) or coastal (Chust et al, 2010;Hansen et al, 2021;Launeau et al, 2018;Parrish et al, 2016;Smeeckaert et al, 2013;Wilson et al, 2019) environments, improve high-resolution modeling of flood inundation (Lague and Feldmann, 2020;Mandlburger et al, 2015) or track sediment transport at the land-water interface. Nevertheless, to fully use these datasets and leverage the scientific potential of extensive datasets made of billions of points, automatic classification of green LiDAR data directly at the 3D PC level is essential.…”

Classification of Land-Water Continuum Habitats Using Exclusively Airborne Topobathymetric Lidar Green Waveforms and Infrared Intensity Point Clouds

“…Indeed, the vertical repartition of the points offers useful information on scene architecture, providing relevant features to determine their origin, namely for vegetation or building identification. Analysis of this geometrical context is the most frequently used method to produce maps of land and water covers [23][24][25]. Research works conducted on PCs processing mostly rely on the computation of geometrical features using spherical neighborhoods [23] and, more recently, on deep neural networks [26].…”

“…Classification of land or water covers using lidar data has been well explored recently. Even when using waveform data, most of the published research is based on 2D data classification [17,25,27,29,33] while fewer articles exploit PCs [24,34,35]. Many studies researching ways to classify lidar data used machine learning algorithms such as support vector machine (SVM), maximum likelihood (ML), or random forests.…”

“…Although machine learning classification of lidar waveform features has been explored previously, we have found no point-based studies dedicated to mapping a large number of habitats both marine and terrestrial. Previously cited studies such as [24,34,35] classified either only marine or only terrestrial habitats from PCs and [27] processed 2D data to produce their map of coastal habitats.…”