Exploring GeoAI Methods for Supraglacial Lake Mapping on Greenland Ice Sheet
Open Access
- 13 November 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Abstract
The ACM SIGSPATIAL Cup 2023 proposed the challenge to identify and map supraglacial lakes in Greenland in satellite imagery. The peculiarities of supraglacial lakes pose a hard problem for semantic segmentation and object detection tasks because the definition of a lake is ill-fitted to the inner workings of such approaches. For example, lakes are often covered by ice and snow and narrow streams can connect distinct lakes, which is not directly translatable to the semantic segmentation of water. It is also not well-posed for object detection, especially the identity relation - what is a lake, what is not (yet) a lake, and what are two lakes is challenging. In this context, we worked on adapting semantic segmentation using the Segment Anything Model and instance segmentation using Mask R-CNN to the setting. The latter ended up superior in our own evaluation and even got ranked second among all participants. We are proud that our approach has led to competitive performance. The source code is available from https://github.com/tum-bgd/GISCup23.Keywords
This publication has 13 references indexed in Scilit:
- A Novel Method for Automated Supraglacial Lake Mapping in Antarctica Using Sentinel-1 SAR Imagery and Deep LearningRemote Sensing, 2021
- Lateral meltwater transfer across an Antarctic ice shelfThe Cryosphere, 2020
- Antarctic Supraglacial Lake Identification Using Landsat-8 Image ClassificationRemote Sensing, 2020
- Antarctic Supraglacial Lake Detection Using Landsat 8 and Sentinel-2 Imagery: Towards Continental Generation of Lake VolumesRemote Sensing, 2020
- Mass balance of the Greenland Ice Sheet from 1992 to 2018Nature, 2019
- Antarctic surface hydrology and impacts on ice-sheet mass balanceNature Climate Change, 2018
- Mass balance of the Antarctic Ice Sheet from 1992 to 2017Nature, 2018
- A Fully Automated Supraglacial lake area and volume Tracking (“FAST”) algorithm: Development and application using MODIS imagery of West GreenlandRemote Sensing of Environment, 2017
- Feature Pyramid Networks for Object Detection2017
- Deep Residual Learning for Image Recognition2016