the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Jun 2025
A Workflow to Identify and Monitor Slow-Moving Landslides through Spaceborne Optical Feature Tracking
Abstract. Identifying and monitoring unstable slopes is essential for preventing both direct and cascading disasters triggered by landslides. To this end, we introduce TerraTrack, an open-source, cloud-based, end-to-end optical feature tracking tool for slow-moving landslide detection and monitoring. The tool is designed to run on the Sentinel-2 Level 1 harmonized collection. Users can define the area of interest, automatically download and pre-process Sentinel-2 data, and compute displacements using different feature tracking techniques. Outputs include a landslide binary mask, velocity maps and time series. The tool can operate entirely in the Google Colaboratory cloud environment. Hence, it removes the need for local computational resources and for a fast internet connection, avoids software conflicts, and is accessible even to those with limited experience in Python programming. We validated the workflow on cases with independent displacement measurements, including the Slumgullion, Chaos Canyon, and Tessina-Lamosano landslides, showing that its results are consistent with independent estimates. Owing to its ease of use and versatility, the tool is a valuable resource for the multi-hazard and landslide communities, complementing InSAR for monitoring surface motion in space and time. The tool can estimate motion shortly after the satellite imagery is acquired, making it an asset for early warning systems that rely on time of failure estimates (i.e., inverse velocity) or predictive modelling of future motion, within the limitations of satellite-based approaches. Lastly, its capacity to identify unstable slopes can guide targeted, detailed investigations into landslide dynamics, enhancing situational awareness and supporting risk mitigation at scale.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(28024 KB)
-
Supplement
(23529 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(28024 KB) - Metadata XML
-
Supplement
(23529 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
- CC1: 'Comment on egusphere-2025-2795', Mahmud Muhammad, 05 Sep 2025
-
CC2: 'Comment on egusphere-2025-2795', Mahmud Muhammad, 05 Sep 2025
- Addition to my first comment, I suggest you might add Akhdefo software to (Table 1) list of codes and toolboxes that may be used for feature tracking on spaceborne data.
This software is written in python developed for Landslide monitoring , uses optical flow algorithms and is capable TimeSeries Generation.
Below is list of citation for the software:
Muhammad, M. (2024, August 9). Multidisciplinary fusion: Structural geology, remote sensing and geotechnical analysis for geothermal exploration and natural hazard assessment in southwestern British Columbia, Canada (Doctoral dissertation). Simon Fraser University. Retrieved from Summit: summit.sfu.ca/item/38467 (See chapters 2 and 5) - Mahmud Muhammad and Maqsad Suriev (2025) Optical Flow: A Multifaceted Approach for Analyzing and Observing Mass Movements through Optical and Radar Images, (Accepted Manuscript) Progress in Landslide Research and Technology, Volume 4 Issue 1, 2025. Progress in Landslide Research and Technology. Springer.
- Muhammad, M., Williams-Jones, G., Stead, D., Tortini, R., Falorni, G., & Donati, D. (2022). Applications of image-based computer vision for remote surveillance of slope instability. Frontiers in Earth Science, 10, 909078.
Citation: https://doi.org/10.5194/egusphere-2025-2795-CC2 - Addition to my first comment, I suggest you might add Akhdefo software to (Table 1) list of codes and toolboxes that may be used for feature tracking on spaceborne data.
-
RC1: 'Comment on egusphere-2025-2795', Anonymous Referee #1, 12 Sep 2025
The authors presents a software to process satellite images using image correlation and show a few examples.
Without entering into the details of the study, which seems good at a quick reading, I think that it is off topic for NHESS, but it should be submitted to more appropriate journals that focus on software like, e.g., Geoscientific Instrumentation, Methods and Data Systems or Computers & Geoscience.
Regards
Citation: https://doi.org/10.5194/egusphere-2025-2795-RC1 -
AC1: 'Reply on RC1', Lorenzo Nava, 14 Sep 2025
We thank the reviewer for their comment. We disagree with the suggestion that this paper is out of scope.
In its aims and scope page (https://www.natural-hazards-and-earth-system-sciences.net/about/aims_and_scope.html), NHESS explicitly lists as within its scope “the design, development, experimentation, and validation of new techniques, methods, and tools for the detection, mapping, monitoring, and modelling of natural hazards and their consequences” as well as “databases, GIS, remote sensing, early warning systems, and monitoring technologies”. TerraTrack is precisely such a contribution: a workflow designed to detect and monitor slow-moving landslides, hazard covered by NHESS.
In addition, our three case studies are not merely software demonstrations but hazard-relevant applications: (i) Landslide detection and monitoring (Slumgullion), (ii) Complementarity with InSAR to obtain a more complete picture of slope instabilities (Tessina-Lamosano), (iii) Generation of displacement time series and failure time estimation to support early-warning systems (Chaos Canyon). Through these examples we also showcase the general applicability of TerraTrack for landslide hazard monitoring and demonstrate its value for operational risk reduction. The readership of NHESS – geohazard experts with a range of technical backgrounds - is much more closely aligned with our intended user base than the journals mentioned in the review.
Citation: https://doi.org/10.5194/egusphere-2025-2795-AC1
-
AC1: 'Reply on RC1', Lorenzo Nava, 14 Sep 2025
-
RC2: 'Comment on egusphere-2025-2795', Anonymous Referee #2, 19 Sep 2025
The authors present TerraTrack: an open-source, cloud-based processing chain that computes the displacement of slow-moving landslides. They discuss the performance of the different implemented methods through synthetic data and three landslide study sites. The codes are already available on GitHub and seem well documented. The tool implements different methods, offering users flexibility. This is a really nice effort of open-science, which could be usefull for landslides study, and teaching of remote sensing, and possibly more.
The paper is very well written and the quality of the analysis is good. However, I think the quality of the paper could be improved by addressing a couple of points.
- The author could add more explanations and references on the different methods they are implementing.
- Some choices should be discussed. For now you are not masking shadows : what’s could be the impact?
- The analysis of subpixel refinement method is bit hard to follow... You mention the benchmark analysis in the case of Slumgullion, but without giving the results of this analysis. Simarly, in the discussion you summarize the results on Slumgullion, then give as recommandation the PCC (without giving a recommandation on the subpixel refinement). Then, you analyse the results on the benchmark to discuss subpixel refinement performances again. It could be clearer to add a small section to present the resultts on synthetic data before the landslides cases. It will help you to explain why you selected the PCC and 4 subpixel refinement methods for Slumgullion. At the end of Slumgullion, you could say that you are selecting os3 for the other landslides. In the discussion, you could discuss in this order : 1) your recommandation for the Feature Tracking algorithm 2) your recommandation for the subpixel refinement method, using the benchmark and Slumgullion case.
- In the abstract you wrote « We validated the workflow on cases with independent displacement measurements ». I was expecting comparison with GNSS or InSAR data. However, even in section 4.2 you are not comparing InSAR and FT results on the points which are in common... Please consider either modifying the abstract or adding more analysis..
See attached file for specific comments.
- AC2: 'Reply on RC2', Lorenzo Nava, 17 Feb 2026
-
RC3: 'Comment on egusphere-2025-2795', Anonymous Referee #3, 04 Feb 2026
The authors present the TerraTrack tool, an open-source workflow based on a cloud processing chain that analyses Sentinel-2 images to identify and monitor the displacements of slow-moving landslides. The tool uses different methods for optical image analysis, and its outputs (velocity maps and displacement time series) are compatible with GIS software for further interpretation. Overall, this tool represents a valuable advancement, providing enhanced support for landslide detection and monitoring efforts.The paper is well-structured and well-written; however, some aspects should also be considered.Please review the attached document for detailed general and specific comments.
- AC3: 'Reply on RC3', Lorenzo Nava, 17 Feb 2026
Interactive discussion
Status: closed
-
CC1: 'Comment on egusphere-2025-2795', Mahmud Muhammad, 05 Sep 2025
Dear Lorenzo,
Thank you for sharing your work — it's very impressive. Optical sensors and optical flow indeed serve as powerful tools in landslide monitoring, especially considering their rapid processing capabilities, low cost, and high temporal acquisition frequencies. Platforms like Planet Labs, with near-daily imagery, offer particularly valuable opportunities in this field.
I wanted to take a moment to share some of my related work from 2022 to 2024, which may be of interest to you. It includes the development of open-source Python software designed specifically for landslide monitoring using optical flow techniques. The tool is capable of processing:
- Spaceborne optical imagery
- Radar backscatter images
- Ground-based photographic data
GitHub - mahmudsfu/AkhDefo: Computer Vision for Slope Stability: Land Deformation Monitoring
Frontiers | Applications of Image-Based Computer Vision for Remote Surveillance of Slope Instability
-
CC2: 'Comment on egusphere-2025-2795', Mahmud Muhammad, 05 Sep 2025
- Addition to my first comment, I suggest you might add Akhdefo software to (Table 1) list of codes and toolboxes that may be used for feature tracking on spaceborne data.
This software is written in python developed for Landslide monitoring , uses optical flow algorithms and is capable TimeSeries Generation.
Below is list of citation for the software:
Muhammad, M. (2024, August 9). Multidisciplinary fusion: Structural geology, remote sensing and geotechnical analysis for geothermal exploration and natural hazard assessment in southwestern British Columbia, Canada (Doctoral dissertation). Simon Fraser University. Retrieved from Summit: summit.sfu.ca/item/38467 (See chapters 2 and 5) - Mahmud Muhammad and Maqsad Suriev (2025) Optical Flow: A Multifaceted Approach for Analyzing and Observing Mass Movements through Optical and Radar Images, (Accepted Manuscript) Progress in Landslide Research and Technology, Volume 4 Issue 1, 2025. Progress in Landslide Research and Technology. Springer.
- Muhammad, M., Williams-Jones, G., Stead, D., Tortini, R., Falorni, G., & Donati, D. (2022). Applications of image-based computer vision for remote surveillance of slope instability. Frontiers in Earth Science, 10, 909078.
Citation: https://doi.org/10.5194/egusphere-2025-2795-CC2 - Addition to my first comment, I suggest you might add Akhdefo software to (Table 1) list of codes and toolboxes that may be used for feature tracking on spaceborne data.
-
RC1: 'Comment on egusphere-2025-2795', Anonymous Referee #1, 12 Sep 2025
The authors presents a software to process satellite images using image correlation and show a few examples.
Without entering into the details of the study, which seems good at a quick reading, I think that it is off topic for NHESS, but it should be submitted to more appropriate journals that focus on software like, e.g., Geoscientific Instrumentation, Methods and Data Systems or Computers & Geoscience.
Regards
Citation: https://doi.org/10.5194/egusphere-2025-2795-RC1 -
AC1: 'Reply on RC1', Lorenzo Nava, 14 Sep 2025
We thank the reviewer for their comment. We disagree with the suggestion that this paper is out of scope.
In its aims and scope page (https://www.natural-hazards-and-earth-system-sciences.net/about/aims_and_scope.html), NHESS explicitly lists as within its scope “the design, development, experimentation, and validation of new techniques, methods, and tools for the detection, mapping, monitoring, and modelling of natural hazards and their consequences” as well as “databases, GIS, remote sensing, early warning systems, and monitoring technologies”. TerraTrack is precisely such a contribution: a workflow designed to detect and monitor slow-moving landslides, hazard covered by NHESS.
In addition, our three case studies are not merely software demonstrations but hazard-relevant applications: (i) Landslide detection and monitoring (Slumgullion), (ii) Complementarity with InSAR to obtain a more complete picture of slope instabilities (Tessina-Lamosano), (iii) Generation of displacement time series and failure time estimation to support early-warning systems (Chaos Canyon). Through these examples we also showcase the general applicability of TerraTrack for landslide hazard monitoring and demonstrate its value for operational risk reduction. The readership of NHESS – geohazard experts with a range of technical backgrounds - is much more closely aligned with our intended user base than the journals mentioned in the review.
Citation: https://doi.org/10.5194/egusphere-2025-2795-AC1
-
AC1: 'Reply on RC1', Lorenzo Nava, 14 Sep 2025
-
RC2: 'Comment on egusphere-2025-2795', Anonymous Referee #2, 19 Sep 2025
The authors present TerraTrack: an open-source, cloud-based processing chain that computes the displacement of slow-moving landslides. They discuss the performance of the different implemented methods through synthetic data and three landslide study sites. The codes are already available on GitHub and seem well documented. The tool implements different methods, offering users flexibility. This is a really nice effort of open-science, which could be usefull for landslides study, and teaching of remote sensing, and possibly more.
The paper is very well written and the quality of the analysis is good. However, I think the quality of the paper could be improved by addressing a couple of points.
- The author could add more explanations and references on the different methods they are implementing.
- Some choices should be discussed. For now you are not masking shadows : what’s could be the impact?
- The analysis of subpixel refinement method is bit hard to follow... You mention the benchmark analysis in the case of Slumgullion, but without giving the results of this analysis. Simarly, in the discussion you summarize the results on Slumgullion, then give as recommandation the PCC (without giving a recommandation on the subpixel refinement). Then, you analyse the results on the benchmark to discuss subpixel refinement performances again. It could be clearer to add a small section to present the resultts on synthetic data before the landslides cases. It will help you to explain why you selected the PCC and 4 subpixel refinement methods for Slumgullion. At the end of Slumgullion, you could say that you are selecting os3 for the other landslides. In the discussion, you could discuss in this order : 1) your recommandation for the Feature Tracking algorithm 2) your recommandation for the subpixel refinement method, using the benchmark and Slumgullion case.
- In the abstract you wrote « We validated the workflow on cases with independent displacement measurements ». I was expecting comparison with GNSS or InSAR data. However, even in section 4.2 you are not comparing InSAR and FT results on the points which are in common... Please consider either modifying the abstract or adding more analysis..
See attached file for specific comments.
- AC2: 'Reply on RC2', Lorenzo Nava, 17 Feb 2026
-
RC3: 'Comment on egusphere-2025-2795', Anonymous Referee #3, 04 Feb 2026
The authors present the TerraTrack tool, an open-source workflow based on a cloud processing chain that analyses Sentinel-2 images to identify and monitor the displacements of slow-moving landslides. The tool uses different methods for optical image analysis, and its outputs (velocity maps and displacement time series) are compatible with GIS software for further interpretation. Overall, this tool represents a valuable advancement, providing enhanced support for landslide detection and monitoring efforts.The paper is well-structured and well-written; however, some aspects should also be considered.Please review the attached document for detailed general and specific comments.
- AC3: 'Reply on RC3', Lorenzo Nava, 17 Feb 2026
Peer review completion
Journal article(s) based on this preprint
Data sets
TerraTrack: A Workflow to Identify and Monitor Slow-Moving Landslides through Spaceborne Optical Feature Tracking Lorenzo Nava et al. https://doi.org/10.5281/zenodo.15609754
Model code and software
TerraTrack: A Workflow to Identify and Monitor Slow-Moving Landslides through Spaceborne Optical Feature Tracking Lorenzo Nava et al. https://doi.org/10.5281/zenodo.15609754
Interactive computing environment
TerraTrack Lorenzo Nava https://github.com/lorenzonava96/TerraTrack
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 6,660 | 3,692 | 283 | 10,635 | 1,401 | 170 | 161 |
- HTML: 6,660
- PDF: 3,692
- XML: 283
- Total: 10,635
- Supplement: 1,401
- BibTeX: 170
- EndNote: 161
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Maximilian Van Wyk de Vries
Louie Elliot Bell
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(28024 KB) - Metadata XML
-
Supplement
(23529 KB) - BibTeX
- EndNote
- Final revised paper
Dear Lorenzo,
Thank you for sharing your work — it's very impressive. Optical sensors and optical flow indeed serve as powerful tools in landslide monitoring, especially considering their rapid processing capabilities, low cost, and high temporal acquisition frequencies. Platforms like Planet Labs, with near-daily imagery, offer particularly valuable opportunities in this field.
I wanted to take a moment to share some of my related work from 2022 to 2024, which may be of interest to you. It includes the development of open-source Python software designed specifically for landslide monitoring using optical flow techniques. The tool is capable of processing:
GitHub - mahmudsfu/AkhDefo: Computer Vision for Slope Stability: Land Deformation Monitoring
Frontiers | Applications of Image-Based Computer Vision for Remote Surveillance of Slope Instability