Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation

Abstract Retrogressive Thaw Slumps (RTSs), a form of thermokarst hazards, pose risks to hydrological and ecological environments and the safety of the Qinghai‐Tibet Engineering Corridor. We still lack the knowledge about the geographic locations of RTSs and their dynamically changing spatial margins...

Full description

Saved in:
Bibliographic Details
Main Authors: Yiling Lin, Xie Hu, Haoyu Lu, Fujun Niu, Gengnian Liu, Lingcao Huang, Shanghang Zhang, Jifu Liu, Yunhuai Liu
Format: Article
Language:English
Published: Wiley 2025-03-01
Series:Journal of Geophysical Research: Machine Learning and Computation
Subjects:
Online Access:https://doi.org/10.1029/2024JH000370
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850063385890127872
author Yiling Lin
Xie Hu
Haoyu Lu
Fujun Niu
Gengnian Liu
Lingcao Huang
Shanghang Zhang
Jifu Liu
Yunhuai Liu
author_facet Yiling Lin
Xie Hu
Haoyu Lu
Fujun Niu
Gengnian Liu
Lingcao Huang
Shanghang Zhang
Jifu Liu
Yunhuai Liu
author_sort Yiling Lin
collection DOAJ
description Abstract Retrogressive Thaw Slumps (RTSs), a form of thermokarst hazards, pose risks to hydrological and ecological environments and the safety of the Qinghai‐Tibet Engineering Corridor. We still lack the knowledge about the geographic locations of RTSs and their dynamically changing spatial margins. However, visual interpretation is labor‐intensive while the present‐day deep learning methods become ineffective when the model trained in one year is directly transferred to another. To enhance the model's generalization ability, here we implemented and compared three domain adaptation methods, that is, the classic supervised fine‐tuning method and two proposed unsupervised methods: Image StyleTransfer Domain Adaptation (ISTDA) and the Tversky Adversarial Domain Adaptation (TADA) network. In our proposed ISTDA, we uniformed the contextual information of multi‐temporal images by Cycle Generative Adversarial Network (CycleGAN). We introduced the Tversky loss and the automatic adjustment of weights for multiple loss functions to suppress false positives and to improve the generalization of TADA. We tested three methods' performance in Beiluhe region over the Qinghai‐Tibet Plateau using PlanetScope optical images during 2019–2022. The three domain adaptation methods are successful in generating regional, multi‐annual RTS inventories. Remarkably, TADA sustains good performance in complex transfer scenarios without additional label cost, achieving an F1 increase of 14.32%–24.17% compared to classic methods. Our work is the first to apply an unsupervised domain adaptation to automatically map the RTSs on a multi‐annual timescale, demonstrating a strong potential of its applicability for monitoring large‐scale, multi‐temporal evolution of geomorphological features.
format Article
id doaj-art-ec1fc405f66b4909b9b45eebd88e10c3
institution DOAJ
issn 2993-5210
language English
publishDate 2025-03-01
publisher Wiley
record_format Article
series Journal of Geophysical Research: Machine Learning and Computation
spelling doaj-art-ec1fc405f66b4909b9b45eebd88e10c32025-08-20T02:49:36ZengWileyJournal of Geophysical Research: Machine Learning and Computation2993-52102025-03-0121n/an/a10.1029/2024JH000370Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain AdaptationYiling Lin0Xie Hu1Haoyu Lu2Fujun Niu3Gengnian Liu4Lingcao Huang5Shanghang Zhang6Jifu Liu7Yunhuai Liu8College of Urban and Environmental Sciences Peking University Beijing ChinaCollege of Urban and Environmental Sciences Peking University Beijing ChinaSchool of Artificial Intelligence Beijing Normal University Beijing ChinaCivil Engineering at the State Key Laboratory of Frozen Soil Engineering Northwest Institute of Eco‐environment and Resources Chinese Academy of Sciences Gansu ChinaCollege of Urban and Environmental Sciences Peking University Beijing ChinaInstitute of Space and Earth Information Science The Chinese University of Hong Kong Hong Kong ChinaThe School of Computer Science Peking University Beijing ChinaFaculty of Geographical Science Beijing Normal University Beijing ChinaThe School of Computer Science Peking University Beijing ChinaAbstract Retrogressive Thaw Slumps (RTSs), a form of thermokarst hazards, pose risks to hydrological and ecological environments and the safety of the Qinghai‐Tibet Engineering Corridor. We still lack the knowledge about the geographic locations of RTSs and their dynamically changing spatial margins. However, visual interpretation is labor‐intensive while the present‐day deep learning methods become ineffective when the model trained in one year is directly transferred to another. To enhance the model's generalization ability, here we implemented and compared three domain adaptation methods, that is, the classic supervised fine‐tuning method and two proposed unsupervised methods: Image StyleTransfer Domain Adaptation (ISTDA) and the Tversky Adversarial Domain Adaptation (TADA) network. In our proposed ISTDA, we uniformed the contextual information of multi‐temporal images by Cycle Generative Adversarial Network (CycleGAN). We introduced the Tversky loss and the automatic adjustment of weights for multiple loss functions to suppress false positives and to improve the generalization of TADA. We tested three methods' performance in Beiluhe region over the Qinghai‐Tibet Plateau using PlanetScope optical images during 2019–2022. The three domain adaptation methods are successful in generating regional, multi‐annual RTS inventories. Remarkably, TADA sustains good performance in complex transfer scenarios without additional label cost, achieving an F1 increase of 14.32%–24.17% compared to classic methods. Our work is the first to apply an unsupervised domain adaptation to automatically map the RTSs on a multi‐annual timescale, demonstrating a strong potential of its applicability for monitoring large‐scale, multi‐temporal evolution of geomorphological features.https://doi.org/10.1029/2024JH000370retrogressive thaw slumps (RTSs)domain adaptationtemporal generalizationsemantic segmentationQinghai‐Tibet plateau
spellingShingle Yiling Lin
Xie Hu
Haoyu Lu
Fujun Niu
Gengnian Liu
Lingcao Huang
Shanghang Zhang
Jifu Liu
Yunhuai Liu
Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
Journal of Geophysical Research: Machine Learning and Computation
retrogressive thaw slumps (RTSs)
domain adaptation
temporal generalization
semantic segmentation
Qinghai‐Tibet plateau
title Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
title_full Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
title_fullStr Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
title_full_unstemmed Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
title_short Multi‐Annual Inventorying of Retrogressive Thaw Slumps Using Domain Adaptation
title_sort multi annual inventorying of retrogressive thaw slumps using domain adaptation
topic retrogressive thaw slumps (RTSs)
domain adaptation
temporal generalization
semantic segmentation
Qinghai‐Tibet plateau
url https://doi.org/10.1029/2024JH000370
work_keys_str_mv AT yilinglin multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT xiehu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT haoyulu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT fujunniu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT gengnianliu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT lingcaohuang multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT shanghangzhang multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT jifuliu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation
AT yunhuailiu multiannualinventoryingofretrogressivethawslumpsusingdomainadaptation