Rapid detection of centimeter-scale change in a Mars analog environment using ground-based mobile LiDAR

Vanga, Sashank; Perkins, Reid P.; Neish, Catherine D.; Zanetti, Michael R.; Carr, Brett B.; Hamilton, Christopher W.

doi:10.5194/egusphere-2026-2733

Preprints

https://doi.org/10.5194/egusphere-2026-2733

Preprints

29 May 2026

| 29 May 2026

Rapid detection of centimeter-scale change in a Mars analog environment using ground-based mobile LiDAR

Sashank Vanga, Reid P. Perkins, Catherine D. Neish, Michael R. Zanetti, Brett B. Carr, and Christopher W. Hamilton

Abstract. Landed missions to dynamic planetary surfaces require tools capable of measuring the small-scale topographic changes driven by active surface processes. A ground-based LiDAR scanner is one of the few instruments capable of detecting the centimeter-scale changes that are expected due to both natural processes such as wind erosion and sedimentary reworking, as well as artificial processes resulting from robotic/human exploration. At landing site scales, a ground-based LiDAR scanner on a mobile platform can collect cm-scale topographic information more efficiently than using a scanner on a stationary mount. To demonstrate the viability of mobile LiDAR for planetary missions, we conducted scans at a Mars analog site in central Iceland. Scans before and after a windstorm detected erosion and deposition of sand on the order of a few cm, with a detection threshold of ~1 cm. The ability of the mobile platform to collect data over extremely short timescales (<30 minutes) made it possible to conduct spontaneous scans to record opportunistic targets after the windstorm, while maintaining cm-scale resolution. This makes a compelling case for the inclusion of mobile LiDAR on future missions to explore planetary surfaces such as the Moon, Mars and Titan.

Received: 13 May 2026 – Discussion started: 29 May 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Sashank Vanga, Reid P. Perkins, Catherine D. Neish, Michael R. Zanetti, Brett B. Carr, and Christopher W. Hamilton

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-2733', Anonymous Referee #1, 01 Jun 2026
The topic is very interesting and relevant, good introduction into it.

There is no GNSS on Mars. That (and the consequences) should already be mentioned in the introduction, not only in the discussion at the end.

Why was the Riegl VUX-1HA chosen? Is 1550 nm the ideal wavelength for the purpose? Why didn’t you choose a scanner with phase shift based distance measurement, which should provide significantly better distance measurement precision?

2.1: Why is it necessary to manually pick point pairs in two datasets here? With very good approximate valuas available, multi-temporal point cloud registration could easily be automated, see for instance the ISPRS Com. II proceedings 2024 (https://isprs-annals.copernicus.org/articles/X-2-2024.

The results look good on first glimpse. However, a detailed analysis of precision and reliability id missing. You mention cm-precision, but is that really substantiated by the results? Especially, it would have been nice to have independent external reference measurements, which shouldn’t be too difficult in such a study on Earth.

I found the statements on the possibilities of registration in GNSS-denied settings too vague, especially concerning the effects on the desired cm-precision. Why were these alternatives not examined in the study? And could SLAM be an option?
Citation: https://doi.org/10.5194/egusphere-2026-2733-RC1
RC2:
'Comment on egusphere-2026-2733', Anonymous Referee #2, 30 Jun 2026
The manuscript presents an interesting field study using a backpack-mounted mobile LiDAR system to measure centimeter-scale dune migration at a site in Holuhraun (Iceland). The data collection is robust, and the terrestrial topographic processing is well applied. However, the manuscript's main scientific justification—that it is a Mars analog demonstration—is not sufficiently supported by the methods and approach used. In its current state, the article documents a conventional terrestrial surveying application without addressing the actual constraints of a planetary mission.
I recommend rejecting the manuscript in its current form. However, I strongly encourage the authors to reprocess their existing data to simulate realistic planetary constraints and resubmit the manuscript, as the underlying dataset has great potential.
Major Comments:
The proposed technical methodology (mobile LiDAR) and the change-detection algorithms (M3C2) are mature and widely documented in the literature. The presented results are exactly what is expected from a high-end commercial instrument like the Riegl VUX-1HA. Thus, the paper does not present an instrumental or algorithmic breakthrough. Furthermore, the methodology should be improved or justified in the following areas:
The authors should justify the use of the Riegl VUX-1HA (a time-of-flight scanner). For centimeter-to-millimeter precision at close range, phase-shift scanners typically provide better distance measurement precision, although this comes at the cost of a shorter maximum range. The choice of instrument should still be briefly justified in the text, even if range and portability are the reason.

Registration is done using nine manually picked point pairs. This works, but it is a highly outdated approach. With good approximate values available, multi-temporal point cloud registration should be automated to avoid user bias and improve reproducibility

The claimed cm-level precision is entirely based on relative cloud-to-cloud distances. But independent ground control points were not used. These would have been easy to set up in this terrestrial field site and would allow direct validation of the precision, instead of relying only on internal comparison. This is especially important, since the precision value highlighted as a key result is exactly what an external validation would need to confirm.

However, the novelty of this paper relies on the use of this instrument in a Mars-analog environment. But the study fails to address the realities of deploying such a system on Mars. This is evident in several critical areas:
The authors achieve their absolute precision and trajectory mapping using a GNSS antenna and a reference station. Mars does not have a GPS network. The authors address this point in the Discussion, where they note that GNSS-denied LiDAR mapping techniques exist and cite prior work, including their own, on this topic. However, this discussion is not reflected in the data or methods presented in this study, as the dataset itself still relies on GNSS for its results; the claim of Mars-relevant precision is therefore discussed but not actually demonstrated.

This constraint should also be explicitly stated in the Introduction, not only in the discussion. To prove this concept works for planetary science, the authors should have processed their raw data without the GNSS input, relying solely on the IMU and SLAM algorithms, to demonstrate if the required cm-level precision can actually be maintained in a GNSS-denied environment.

The 175 m² region of interest produced over 5 million points. The manuscript does not discuss the data volume problem for deep-space communication. Deep space communications have severe bandwidth limitations, and raw LiDAR point clouds are incredibly data-heavy.

A mobile backpack scanner does not reflect the severe constraints of a rover. The authors do not address the high energy consumption required to run a continuous laser scanner, nor the physical size and weight limitations that are critical in spaceflight engineering.

Also, a human operator provides a completely different vantage point, walking speed, and vibration dampening compared to a rigid, slow-moving rover. Additionally, in a terrestrial analog, a human operator will naturally shelter the LiDAR equipment during a severe windstorm to protect it. On Mars, the instrument cannot be put away; it would be continuously exposed to harsh environmental conditions, abrasive dust, and intense winds, which calls into question the operational durability of the proposed setup.

So, in general the change-detection result itself is solid science but not novel. This paper could have a significant scientific impact if it shifts its focus: rather than demonstrating that "commercial LiDAR can detect millimeter changes on Earth," it should explore "how Martian constraints degrade the capability of LiDAR to detect changes."
I highly encourage the authors to resubmit a revised version of this paper that performs adapted simulations on their existing data. A few examples:
Process the trajectory without GNSS (IMU/SLAM only) to calculate the accumulated error.

Perform massive downsampling on the point clouds to simulate the lighter file sizes dictated by deep-space data transmission limits.

Adjust the point density and perspective to match a slow-moving rover (or install the system in a small rover simulator)

If the authors can demonstrate that cm-scale change detection is still possible under these artificially impoverished, Mars-simulated conditions, the manuscript would be an exceptionally valuable contribution.
Citation: https://doi.org/10.5194/egusphere-2026-2733-RC2
AC1:
'Comment on egusphere-2026-2733', Sashank Vanga, 08 Sep 2026
We thank the reviewers for their helpful feedback. Below is our response to each comment, in bold.
Reviewer 1
The topic is very interesting and relevant, good introduction into it.

We thank the reviewer for this comment.
There is no GNSS on Mars. That (and the consequences) should already be mentioned in the introduction, not only in the discussion at the end.

We thank the reviewer for this suggestion. We modified sections 1 (line 48-49, 51-57) and 2 (line 83-88) to highlight the comparison of our data with GNSS-denied LiDAR data from the same region. Lines 51-55 in particular discuss the implications of the lack of GNSS.
Why was the Riegl VUX-1HA chosen? Is 1550 nm the ideal wavelength for the purpose? Why didn’t you choose a scanner with phase shift based distance measurement, which should provide significantly better distance measurement precision?

While phase-shift sensors may offer comparable or better precision than time of flight (ToF) sensors, they are generally used indoors on static mounts and have much shorter range than ToF sensors (e.g., see doi.org/10.3390/s22165946). The VUX-1HA was chosen for its sub-cm precision over hundreds of meters necessary for measuring cm-scale topographic features and directly comparing the results to cm-scale synthetic aperture radar data, which is a major focus of our lab group (see, for example, Tolometti et al., 2022, Journal of Geophysical Research). While in Iceland to scan rough lava flows, we utilized the instrument in an opportunistic way to detect cm-scale change in sediment after a particularly strong windstorm. 1550 nm is one of only a handful of wavelengths used in ground-based and aerial LiDAR sensors as it is considered eye-safe. Other sensors (such as a phase-shift sensor) were not considered due to the costs associated with purchasing new equipment and the opportunistic nature of the change detection study.
2.1: Why is it necessary to manually pick point pairs in two datasets here? With very good approximate valuas available, multi-temporal point cloud registration could easily be automated, see for instance the ISPRS Com. II proceedings 2024 (https://isprs-annals.copernicus.org/articles/X-2-2024.

While automated registration methods such as ICP-based registration are generally preferred for point cloud alignment, we observed that point picking performed better in our case compared to ICP. Lines 93-97 in section 2.1 now read:
"Registration via point picking resulted in a root mean square error (RMSE) of ~2 cm compared to the final RMSE of ~4 cm from automated point cloud registration using the iterative closest point (ICP) (Besl and McKay, 1992) approach. Surficial change due to the windstorm was likely present across most of the ROI, thus, the higher RMSE value for ICP likely resulted from incorporating actual change in sand depth into the registration process (Carr et al., 2019; Hou et al., 2026).”
The results look good on first glimpse. However, a detailed analysis of precision and reliability id missing. You mention cm-precision, but is that really substantiated by the results? Especially, it would have been nice to have independent external reference measurements, which shouldn’t be too difficult in such a study on Earth.

It is typical to use artificial ground control points (GCPs) as an external means of reference during predetermined survey(s) for change detection. However, we used the undisturbed boulders as ‘natural’ ground control in this study. Our M3C2 differencing process between the boulder points highlight mean vertical and horizontal change of 3 mm and <1 mm respectively, which we think adequately establishes the threshold for detectable change by similar instruments.
I found the statements on the possibilities of registration in GNSS-denied settings too vague, especially concerning the effects on the desired cm-precision. Why were these alternatives not examined in the study? And could SLAM be an option?

We thank the reviewer for this comment. We added section 4 to the manuscript, where we discuss the comparison of our original data with GNSS-denied data processed using a SLAM algorithm.
Reviewer 2
The manuscript presents an interesting field study using a backpack-mounted mobile LiDAR system to measure centimeter-scale dune migration at a site in Holuhraun (Iceland). The data collection is robust, and the terrestrial topographic processing is well applied. However, the manuscript's main scientific justification—that it is a Mars analog demonstration—is not sufficiently supported by the methods and approach used. In its current state, the article documents a conventional terrestrial surveying application without addressing the actual constraints of a planetary mission.
I recommend rejecting the manuscript in its current form. However, I strongly encourage the authors to reprocess their existing data to simulate realistic planetary constraints and resubmit the manuscript, as the underlying dataset has great potential.
Major Comments:
The proposed technical methodology (mobile LiDAR) and the change-detection algorithms (M3C2) are mature and widely documented in the literature. The presented results are exactly what is expected from a high-end commercial instrument like the Riegl VUX-1HA. Thus, the paper does not present an instrumental or algorithmic breakthrough. Furthermore, the methodology should be improved or justified in the following areas:
The authors should justify the use of the Riegl VUX-1HA (a time-of-flight scanner). For centimeter-to-millimeter precision at close range, phase-shift scanners typically provide better distance measurement precision, although this comes at the cost of a shorter maximum range. The choice of instrument should still be briefly justified in the text, even if range and portability are the reason.

While phase-shift sensors may offer comparable or better precision than time of flight (ToF) sensors, they are generally used indoors on static mounts and have much shorter range than ToF sensors (e.g., see doi.org/10.3390/s22165946). The VUX-1HA was chosen for its sub-cm precision over large areas necessary for measuring cm-scale topographic features and directly comparing the results to cm-scale synthetic aperture radar data, which is a major focus of our lab group (see, for example, Tolometti et al., 2022, Journal of Geophysical Research). While in Iceland to scan rough lava flows, we utilized the instrument in an opportunistic way to detect cm-scale change in sediment after a particularly strong windstorm. 1550 nm is one of only a handful of wavelengths used in ground-based and aerial LiDAR sensors as it is considered eye-safe. Other sensors (such as a phase-shift sensor) were not considered due to the costs associated with purchasing new equipment and the opportunistic nature of the change detection study.

Registration is done using nine manually picked point pairs. This works, but it is a highly outdated approach. With good approximate values available, multi-temporal point cloud registration should be automated to avoid user bias and improve reproducibility

We thank the reviewer for this comment. We modified section 2.1 to clarify the choice of registration method used. Lines 93-97 now read:
"Registration via point picking resulted in a root mean square error (RMSE) of ~2 cm compared to the final RMSE of ~4 cm from automated point cloud registration using the iterative closest point (ICP) (Besl and McKay, 1992) approach. Surficial change due to the windstorm was likely present across most of the ROI, thus, the higher RMSE value for ICP likely resulted from incorporating actual change in sand depth into the registration process (Carr et al., 2019; Hou et al., 2026).”
The claimed cm-level precision is entirely based on relative cloud-to-cloud distances. But independent ground control points were not used. These would have been easy to set up in this terrestrial field site and would allow direct validation of the precision, instead of relying only on internal comparison. This is especially important, since the precision value highlighted as a key result is exactly what an external validation would need to confirm.

We agree with the reviewer’s comment that use of ground control points (GCPs) would allow for external validation of precision. However, as this was an opportunistic study, we relied on unchanged surfaces such as rocks and boulders within the scanned area as control. We posit that the lack of external ground control would make this work more analogous to performing a similar survey on a planetary surface where the use of external ground control may not be feasible. Lines 97-99 added to section 2.1 now reflect this.
However, the novelty of this paper relies on the use of this instrument in a Mars-analog environment. But the study fails to address the realities of deploying such a system on Mars. This is evident in several critical areas:
The authors achieve their absolute precision and trajectory mapping using a GNSS antenna and a reference station. Mars does not have a GPS network. The authors address this point in the Discussion, where they note that GNSS-denied LiDAR mapping techniques exist and cite prior work, including their own, on this topic. However, this discussion is not reflected in the data or methods presented in this study, as the dataset itself still relies on GNSS for its results; the claim of Mars-relevant precision is therefore discussed but not actually demonstrated.

We thank the reviewer for this suggestion. We modified sections 1 (line 48-49, 51-57) and 2 (line 83-88) to highlight the comparison of our data with GNSS-denied LiDAR data from the same region. Lines 51-55 in particular discuss the implications of not having GNSS on Mars.
This constraint should also be explicitly stated in the Introduction, not only in the discussion. To prove this concept works for planetary science, the authors should have processed their raw data without the GNSS input, relying solely on the IMU and SLAM algorithms, to demonstrate if the required cm-level precision can actually be maintained in a GNSS-denied environment.

We thank the reviewer for this suggestion. We added section 4 to the manuscript, where we discuss the comparison of our original data with GNSS-denied data processed using a SLAM algorithm, to substantiate our claims.
The 175 m² region of interest produced over 5 million points. The manuscript does not discuss the data volume problem for deep-space communication. Deep space communications have severe bandwidth limitations, and raw LiDAR point clouds are incredibly data-heavy.

Despite the high point density of our data, the compressed M3C2 distance point clouds of the region of interest (ROI) are smaller than 42 MegaBytes (MB) each. We don’t consider the data volume to be inhibitive of uplink/downlink for Moon/Mars-based communications. Newly added section 4 also highlights the comparison between the original data and downsampled data from the same region. We show that the downsampled data reduces data volume considerably without significant change in detection thresholds.
A mobile backpack scanner does not reflect the severe constraints of a rover. The authors do not address the high energy consumption required to run a continuous laser scanner, nor the physical size and weight limitations that are critical in spaceflight engineering.

We agree with the reviewer’s comment regarding the importance of size, weight, and power (SWaP) specifications of the instrument for space applications. However, we don’t consider these engineering constraints to be within the scope of this study.
Also, a human operator provides a completely different vantage point, walking speed, and vibration dampening compared to a rigid, slow-moving rover. Additionally, in a terrestrial analog, a human operator will naturally shelter the LiDAR equipment during a severe windstorm to protect it. On Mars, the instrument cannot be put away; it would be continuously exposed to harsh environmental conditions, abrasive dust, and intense winds, which calls into question the operational durability of the proposed setup.

Instrument design and operation dynamics are beyond the scope of our work.
So, in general the change-detection result itself is solid science but not novel. This paper could have a significant scientific impact if it shifts its focus: rather than demonstrating that "commercial LiDAR can detect millimeter changes on Earth," it should explore "how Martian constraints degrade the capability of LiDAR to detect changes."
I highly encourage the authors to resubmit a revised version of this paper that performs adapted simulations on their existing data. A few examples:
Process the trajectory without GNSS (IMU/SLAM only) to calculate the accumulated error.

Perform massive downsampling on the point clouds to simulate the lighter file sizes dictated by deep-space data transmission limits.

Adjust the point density and perspective to match a slow-moving rover (or install the system in a small rover simulator)

We thank the reviewer for these suggestions. Our revised manuscript includes a comparison of our original data with GNSS-denied SLAM data from the same region (section 4). We also downsampled our original data significantly to match the point density of the SLAM data. The original and downsampled versions of the data were determined to be within the transmission limits from a Mars communications standpoint.
If the authors can demonstrate that cm-scale change detection is still possible under these artificially impoverished, Mars-simulated conditions, the manuscript would be an exceptionally valuable contribution.
Citation: https://doi.org/10.5194/egusphere-2026-2733-AC1

Sashank Vanga, Reid P. Perkins, Catherine D. Neish, Michael R. Zanetti, Brett B. Carr, and Christopher W. Hamilton

Data sets

Backpack LiDAR scans from a sand covered lava margin at Holuhraun (Iceland) Vanga, S., Perkins, R. P., Neish, C. D., Zanetti, M. R., Carr, B. B. and Hamilton, C. W. https://doi.org/10.5069/G92805VK

Sashank Vanga, Reid P. Perkins, Catherine D. Neish, Michael R. Zanetti, Brett B. Carr, and Christopher W. Hamilton

Viewed

Total article views: 196 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
121	55	20	196	13	11

HTML: 121
PDF: 55
XML: 20
Total: 196
BibTeX: 13
EndNote: 11

Views and downloads (calculated since 29 May 2026)

Month	HTML	PDF	XML	Total
Jun 2026	37	15	8	60
Jul 2026	29	21	2	52
Aug 2026	32	6	5	43
Sep 2026	23	13	5	41

Cumulative views and downloads (calculated since 29 May 2026)

Month	HTML	PDF	XML	Total
Jun 2026	37	15	8	60
Jul 2026	29	21	2	52
Aug 2026	32	6	5	43
Sep 2026	23	13	5	41

Viewed (geographical distribution)

Total article views: 165 (including HTML, PDF, and XML) Thereof 165 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 17 Sep 2026

Short summary

Landers and rovers deployed during planetary surface missions are expected to encounter small-scale surficial changes due to natural processes such as wind erosion and/or artificial processes caused by robotic/human exploration. By measuring centimeter-scale changes in sand cover at a Mars analog site, we show that a ground-based LiDAR system situated on a mobile platform can identify similar changes on planetary surfaces in short timeframes and without compromising accuracy.


Total:	0
HTML:	0
PDF:	0
XML:	0