the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Sep 2024
Brief communication: Potential of satellite optical imagery to monitor glacier surface flow velocity variability in the tropical Andes
Abstract. We present the first analysis of glacier dynamics in the tropical Andes of Peru and Bolivia using satellite data from 2013 to 2022. Despite the challenges posed by small-size glacier, low velocities and high cloudiness during the monsoon, we map annually aggregated surface velocities and quantify the seasonal variability in the fastest parts of the glaciers. Limited trends are observed on the annual velocities over the last decade, but significant seasonal changes between the wet and the dry seasons are found, likely controlled by the seasonality in melt water production and the related changes in the hydrological conditions at the glacier-bedrock interface.
- Preprint
(1675 KB) - Metadata XML
-
Supplement
(614 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2662', Andres Rivera, 15 Oct 2024
This is a short, interesting and well written contribution about ice flow velocities of small glaciers in the tropical Andes of southern Peru and northern Bolivia. The authors have processed a great number of Landsat and Sentinel images from 2013 to 2022 following procedures and methods already applied to bigger glaciers and ice caps elsewhere. The main assets of this manuscript is the positive application of this methodology for detecting velocities of small glaciers in a region with limited data. They were able to detect very slow motions in regions with steep slopes and limited cloud free conditions. They have also detected some seasonal variations without clear inter annual trends. Even if the method is not novel, the main contribution is its application in small glaciers. It will be very nice if an outcome of this papers is building a freely available database with the resulting velocities in the region.
I’m having few comments and suggestions aiming to improve the text. My main concern is related to the quality of the figures, especially Nº1 that requires some improvements. For somebody not very familiar with the study area, it is difficult to locate Figure 1 a, c and e. There are very small points with colors at main Figure 1 that I presume are the insets locations. Maybe adding a bigger symbol for each inset box to Figure 1 could help. Each box must have co-ordinates. The resolution of the boxes is quite low and very limited details could be seen. Maybe having a box of only the main glaciers and not for the whole mountain center? The time series locations are not visible and I struggled to sea each start¡¡ The Figures b, d and e are extremely noisy and looks like there is no trend at all¡¡ By the way, in the text it is used m/yr2 and in the figure says m. yr-2. I suggest using only one form in the whole text. In Figure 1 (main) it is mentioned the Randolph glacier inventory (RGI), but nothing is said in the text. I suggest adding a reference and a phrase about the RGI used polygons and why not using the national inventory. I presume Ames’s inventory from 1989 is outdated but you can have a look and compare your outlines with the Peru’s national glacier inventory available at (https://sinia.minam.gob.pe/documentos/inventario-nacional-glaciares-lagunas-origen-glaciar-2023).
Regarding the seasonal comparison shown in Figure 2 is very interesting especially for such small glaciers. Some comments: The name of each analyzed glacier is visible in figure 1, but the numbers in figure 2 are a mean for the ablation zone? or are just one spot on the glacier? Increasing the size of Figure 1a, b and C can help also for identifying the location of your series in Figure 2.
As a general approach to the problem, I’m missing a brief but informative discussion of the literature available about glacier changes in the region, the relationship with ENSO and some trends justifying the analysis that has been done. How the resulting ice velocities are improving our understanding of ongoing and forecasted glacier behavior in the region? This is a brief communication, so not space for a detailed overview, but a phrase or two about this will setup the context.
How the obtained results are compared to field and remotely sensed data? There is one mention to GNSS data at Zongo, but nothing is said about the compared values. There is an agreement with some results in the Alps, but there is nothing else in the Andes? There are more works in the region, not based on GNSS, but using remotely sensed imagery that can be compared. See for example Kos et al 2021 https://doi.org/10.3390/rs13142694.
Two of the co-authors published a nice compilation of ice velocities in Cordillera Blanca (Millan et al., 2019), but seems to me there is no overlapping areas with this study for comparison purposes. In this sense, this brief communication is not about the whole tropical Andes of Peru and Bolivia, but from a limited region of Southern Peru and Northern Bolivia. Maybe this could be said from the early beginning to avoid misunderstandings regarding the extension of the study area.
There are previous works in Peru and Bolivia about velocities of debris-covered glaciers (e.g. Hubbard and Clemmens, 2008 https://doi.org/10.3189/002214308785837057.). The measured values are extremely low but similar to your minimum velocities. Maybe it is worth mentioning these efforts and how your method could be applied (or not) to these glacier types.
Regarding Chaupi Orco North West: A possible surging event is not totally absent in the tropics as mentioned by Basantes-Serrano et al., (2022), but they indicated that one possibility explaining this process is subglacial geothermal heating increase as the glacier is on top of a at Antizana volcano (even if has no activity in the last 4 centuries). Is Chaupi on a volcano? Could be related to geothermal activity increase at the glacier bed?
The velocity changes at Chaupi during the possible surge event is in the order of 100 m from a mean of 60-70 m/yr (roughly speaking) between 2013-2021 to a maximum of 165 m/yr in May 2022. If you see the seasonal variations in the other 3 glaciers, there is a gap (explained in the text due to the lack of images) where there are jumps of near 90-100 m/yr between August/November and March/April. Could be possible that the limited data at Chaupi is precluding to see the seasonal “jump” and looks like there is a sudden increase 2021-2022?
In synthesis I think this is a valuable contribution that deserves be published.
Citation: https://doi.org/10.5194/egusphere-2024-2662-RC1 -
AC1: 'Reply on RC1', Romain Millan, 20 Nov 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2662/egusphere-2024-2662-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Romain Millan, 20 Nov 2024
-
RC2: 'Comment on egusphere-2024-2662', Whyjay Zheng, 16 Oct 2024
I am thankful that the authors took a step into the ice flow velocity measurements using the optical feature tracking technique for the tropical glaciers in the Andes. As the authors stated in the preprint, this region suffers from suboptimal cloud conditions for the optical images to be useful. The authors ventured into this challenge and presented the aggregated annual velocity map for many small glaciers, some even with seasonal time series. Despite having a non-trivial number of pixel voids (e.g., Figure 1c), I still think the workflow presented here is state-of-the-art.
This work is concisely well written, and the results are worth sharing with the glaciology community. I have listed my specific comments below to hopefully improve the communication between the authors and future readers.
(Not so) major comments:
- Justification of the processing methods. The processing methods used in this work are basically the same as the three papers mentioned in L38. Therefore, the authors skipped many details when they described their workflow. I checked the three references, and it seems that methods, including the feature tracking parameters and annual map extraction, are fine-tuned for the Alps. Millan et al.’s 2019 paper uses a Peruvian region for one of the test cases, but it does not quite convince me that the parameter set is good enough for the tropical Andes glaciers because the derived velocity map does not have the same quality as the Alps glaciers in the same paper. In addition, highly imbalanced data availability during monsoon and non-monsoon seasons may also affect the robustness of the linear regression for the annual map and lead to extra uncertainty. This is also different from the case in the Alps. Could you explain why you think the same workflow applies to the tropical Andes, if there are reasons other than convenience? I think it is worth sharing any thoughts under the hood even if the authors do not aim for the best tracking parameters and processing workflow (cf. to what is suggested in Millan et al.’s paper, “The size of the sub-images may also be sub-optimal for correlation, but seems more appropriate…”). Lastly, for your future reference, I’d like to share my team’s recent work with the authors (https://doi.org/10.5194/tc-17-4063-2023), which aims to help the fine-tuning process of a feature-tracking workflow.
- Time series. Section 2.2 is titled “Time series extraction.” I see multiple annual maps are a type of time series, but with Figure 2, I was a bit misled in the beginning and was excited to search for the seasonal/monthly signals for all glaciers until I realized the time series with a high temporal sampling rate, as in Figure 2, is only for a few glaciers. With the current text structure, I would change the title of section 2.2 to “Annual maps production and selected seasonal signal extraction” or something similar to reflect the presented results. Also, I am curious to know if the authors have a map showing where (or for what pixels) the LOWESS method for the seasonal signal is applicable. This information may be valuable to the community.
Minor and technical comments:
- I do not see a Code availability and Data availability section in the preprint. For this work, I think it is essential to provide readers with guidelines about how to get the new data or at least how to reproduce the results.
- For the colormap used for surface flow velocity (Figure 1), please ensure it adheres best to the TC suggestions described here: https://www.the-cryosphere.net/submission.html#figurestables. I recommend a perceptually uniform colormap.
- Supplementary figures S1 and S2 lack key information to help readers understand what is presented, such as the satellite/instrument names and scale. Did the authors review both descending and ascending tracking for both locations? Is the loss of coherence caused by the terrain effect or the climatic conditions?
- L60: Is it possible to specify what trend maps use a Gaussian filter and what others use a median filter?
- L75-80: I do not quite understand what the authors want to achieve by comparing the Bolivia case with the Alps case. Zongo Glacier has a higher coefficient of variation (σ/μ ~1) than Rabatel et al.’s Alps test (σ/μ ~0.1), which makes the argument that “D-GNSS measurement falls within the level of error” much less powerful.
- Figure 1: Panel D (and maybe B and F as well) contains lots of non-trend pixels, which block the view of the trend color. At this figure resolution, I can barely see if there is any area with color other than yellow (0).
- The authors mentioned the glacier flow response to climate change at least twice in the manuscript (L16-17 & L140-141). Do we have any in the results? Does the surge signal count? It might be worth adding a sentence or two to address these questions.
Citation: https://doi.org/10.5194/egusphere-2024-2662-RC2 -
AC1: 'Reply on RC1', Romain Millan, 20 Nov 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-2662/egusphere-2024-2662-AC1-supplement.pdf
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 173 | 47 | 12 | 232 | 20 | 5 | 4 |
- HTML: 173
- PDF: 47
- XML: 12
- Total: 232
- Supplement: 20
- BibTeX: 5
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1