Introduction

Code and dataset for paper Fine-scale Antarctic Grounded Ice Cliff 3D Calving Monitoring based on Multi-temporal UAV Photogrammetry Without Ground Control. Ice cliff collapse detection and volume estimation using multitemporal aligned mesh generated from photogrammetry.

Note: Due to the randomness of the clustering algorithm, the results obtained by the code will be slightly different from the results of the paper.

Requirements

Software

1. MATLAB (tested on R2023a)
2. CloudComPy (tested on v3.9)

MATLAB file exchange

1. Triangle/Ray Intersection
2. LASRead/LASWrite

Dataset

The dataset contains geo-registered multitemporal meshes recording the calving process of an ice cliff in Antarctica and ice cliff calving detection results. It can be accessed here: https://data.mendeley.com/datasets/fvt6r84zmm/1

Multitemporal meshes

The meshes were generated from UAV photogrammetry and the co-alignment technique. The dataset is a single CloudCompare .bin file, Ice_cliff_multitemporal_mesh.bin. It contains meshes generated by Metashape from each epoch. The file structure is shown in the image below. Each subfolder contains a mesh of the ice cliff on yyyymmdd day.

The ice cliff is located between the Qinling station area $(74^\circ56' S, 163^\circ 42' E)$ and the Nansen Ice shelf, Inexpressible Island in Victoria Land, East Antarctica. The length of its coastline is approximately 0.89 km.

The UAV for collecting aerial images was a DJI Mavic 2 Pro drone, which carries a Hasselblad L1D-20c gimbal camera. Its 28 mm-equivalent lens has a $77^\circ$ field of view (FoV). From 30-Jan to 25-Feb 2022, 10 flights were performed with the same flight parameters in 26 days. The date of flights were 30-Jan, 31-Jan, 1-Feb, 5-Feb, 7-Feb, 10-Feb, 13-Feb, 16-Feb, 17-Feb and 25-Feb.

Parameter name	Value
Flying height	100 m
Image front/sidelap	80% / 60%
# of images planned	330
GSD	2.1 cm
Image coverage	159 $\times$ 106 m

Calving detection results

In the \result folder stores a series of point cloud files mmdd-mmdd_diff.ply. They are calving object detection results of our paper. The snow_volume_0.66.xlsx is a table containing the numerical results of calving detection. Below are two screenshots showing ice cliff mesh and calving object detection results from our paper.

Run

Generate valid space for change detection

1. Set the parameter STEP in CloudComPy_scripts\run_change_detection.py to 9, and the parameter path to the location of Ice_cliff_multitemporal_mesh.bin.
2. Run CloudComPy_scripts\run_change_detection.py (CloudComPy required), it creates a point cloud file mmdd-mmdd.las under the subfolder of distance_threshold_xx. This compares the mesh between the first and the last epoch.
3. Set the folder in line 4 of run_generateValidSpace.m to the subfolder created in step 2.
4. Run run_generateValidSpace.m, it will create a pointcloudValidSpace.mat storing the point cloud for valid space in the export subfolder.

Ice cliff change detection and volume calculation

1. Set the parameter STEP in CloudComPy_scripts\run_change_detection.py to 1, and the parameter path to the location of Ice_cliff_multitemporal_mesh.bin.
2. Run CloudComPy_scripts\run_change_detection.py (CloudComPy required), it creates a series of point cloud file mmdd-mmdd.las under the subfolder of distance_threshold_xx. This compares the mesh between each epoch.
3. Set the folder in line 4 of run.m to the subfolder created in step 2.
4. Run run.m, it will export the calving volume result to the console, and store it to an Excel table. It will also create a series of point cloud file mmdd-mmdd_diff.ply. They are calving objects detected by the algorithm.

Citation

@article{ZHANG2025104620,
title = {Fine-scale Antarctic grounded ice cliff 3D calving monitoring based on multi-temporal UAV photogrammetry without ground control},
journal = {International Journal of Applied Earth Observation and Geoinformation},
volume = {142},
pages = {104620},
year = {2025},
issn = {1569-8432},
doi = {https://doi.org/10.1016/j.jag.2025.104620},
url = {https://www.sciencedirect.com/science/article/pii/S1569843225002675},
author = {Shuhang Zhang and Lei Zheng and Huizhou Zhou and Qiuyang Zhao and Jie Li and Yinyue Xia and Wuming Zhang and Xiao Cheng},
keywords = {Grounded ice cliff, Calving detection, UAV photogrammetry, Co-alignment}
}