Segmenting submarine boulders with Random Forest and nnU-Net: from bathymetric rasters to survey-scale size distributions
Abstract
In the soft-bottom-dominated southern Baltic Sea, geogenic hard substrates are biodiversity hotspots. The lack of open-source, pixel-level segmentation approaches for boulders in MBES data has made it difficult to advance the ecological understanding of boulder fields in the Baltic Sea. This study developed and evaluated two segmentation approaches. A deep learning model (nnU-Net) and a classical machine learning model (APOC-25, based on Random Forest classification) were trained to segment boulders using multibeam echosounder (MBES) bathymetry data at scale. Both models were trained on manually annotated data. Additionally, a pipeline for extracting per-boulder dimension metrics from the resulting segments was developed. The trade-off between annotation effort and segmentation performance was assessed. The nnU-Net model achieved the best overall performance, with a recall of 0.77, a precision of 0.84, and an F1 score of 0.8. APOC-25, trained on a sparse subset of labelled pixels, reached a recall of 0.73, a precision of 0.51, and an F1 score of 0.60. Upscaling the input data improved detection performance for both models. Additional analyses of areal coverage bias and detection accuracy were performed, and predictions were compared against side-scan sonar (SSS) annotations at two sites. This study provides open-source tools for systematic boulder mapping, enabling the generation of ecologically relevant baseline data and facilitating habitat mapping, monitoring and restoration.
Related
- An improved polarization-sensitive semi-analytic monte carlo simulation method for bathymetric lidar considering complex water-column stratification
- Hidden periodicity in LiDAR point-count rasters
- Late Holocene deposition of coral reef boulders and cyclonic activity in Tuamotu–Gambier Islands (French Polynesia): an overview
- Sudden collapse of ice mélange under increasing submarine melt at tidewater glaciers
- Factors controlling asymmetry in submarine channels and submarine canyons: Insights from global morphological analysis
BibTeX
@misc{naumann2026segmenting,
title = {{Segmenting submarine boulders with Random Forest and nnU-Net: from bathymetric rasters to survey-scale size distributions}},
author = {Aïcha Naumann and Robert Haase and Peter Feldens and Svenja Papenmeier},
howpublished = {EarthArXiv},
year = {2026},
month = oct,
doi = {10.31223/x5tj8d},
url = {https://eartharxiv.org/repository/view/15379/},
note = {CC BY-NC 4.0},
}
RIS
TY - UNPB
TI - Segmenting submarine boulders with Random Forest and nnU-Net: from bathymetric rasters to survey-scale size distributions
AU - Aïcha Naumann
AU - Robert Haase
AU - Peter Feldens
AU - Svenja Papenmeier
PY - 2026
DA - 2026/10/07
PB - EarthArXiv
DO - 10.31223/x5tj8d
UR - https://eartharxiv.org/repository/view/15379/
AB - In the soft-bottom-dominated southern Baltic Sea, geogenic hard substrates are biodiversity hotspots. The lack of open-source, pixel-level segmentation approaches for boulders in MBES data has made it difficult to advance the ecological understanding of boulder fields in the Baltic Sea. This study developed and evaluated two segmentation approaches. A deep learning model (nnU-Net) and a classical machine learning model (APOC-25, based on Random Forest classification) were trained to segment boulders using multibeam echosounder (MBES) bathymetry data at scale. Both models were trained on manually annotated data. Additionally, a pipeline for extracting per-boulder dimension metrics from the resulting segments was developed. The trade-off between annotation effort and segmentation performance was assessed. The nnU-Net model achieved the best overall performance, with a recall of 0.77, a precision of 0.84, and an F1 score of 0.8. APOC-25, trained on a sparse subset of labelled pixels, reached a recall of 0.73, a precision of 0.51, and an F1 score of 0.60. Upscaling the input data improved detection performance for both models. Additional analyses of areal coverage bias and detection accuracy were performed, and predictions were compared against side-scan sonar (SSS) annotations at two sites. This study provides open-source tools for systematic boulder mapping, enabling the generation of ecologically relevant baseline data and facilitating habitat mapping, monitoring and restoration.
ER -