A New Earth Crustal Velocity Field Estimation from ROA cGNSS Station Networks in the South of Spain and North Africa
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsReview of manuscript "A new Earth’s crustal velocity field estimation from ROA cGNSS stations network in South of Spain and North Africa" by Rodriguez Collantes et al.
The authors used a set of permanent GNSS stations to compute position time series and velocities in the region of South Iberia and NW Africa where Africa and Eurasia plates converge. The GNSS data were processed by the PRIDE-AR software in the IGb2020 reference frame. The time series were analysed by the SARI software.
I find the resuls as interesting and worthy to be published. I have several comments that can help the authors to present a more complete analysis of their findings:
1) it is interesting to compare the velocity products with the EPOS dataset (Fernandez et al. 2022; doi:10.4401/ag-8776). The products are available from https://gnssproducts.epos.ubi.pt/
2) some discussion is necessary about the influence of co-seismic offfsets on the GNSS position time series. For example, the velocity product may have been affected by the 2016 Al Hoceima earthquake (https://earthquake.usgs.gov/earthquakes/eventpage/us10004gy9/executive). I see several GNSS stations within a 100 km radius around the epicentre. Indeed, I see a small offset (a few mm) in the East component of station ALBO in the NGL time series [http://geodesy.unr.edu/NGLStationPages/stations/ALBO.sta]
3) Also, it is worthy to discuss the M6.8 El-Haouz earthquake [https://earthquake.usgs.gov/earthquakes/eventpage/us7000kufc/executive] which is a result of compression and its location and kinematics may confirm the strain-rate results of this study.
4) in Fig. 7 I note that the usual way to display the rotation rate is in degrees per million years and the shear strain in nanostrains per year (see doi:10.1002/2014JB011762)
5) Some stations (e.g.AVER, LIJA) show a non-linear rate of motion which is not expected for geodynamic studies and probably reflect some local effect or a transient signal. This needs to be discussed in view of using the station velocities to compute strain rates and rotations.
Therefore, I recommend revision of this manuscript and further discussion of their results.
Minor comments:
Line 209: please add more information on the 2016 earthquake (date, location, magnitude). I suggest to see this paper https://doi.org/10.1093/gji/ggab089 for removing earthquake effects from the time series of GNSS position data.
Lines 250-253: as above.
Line 328. This is Figure 8.
Fig. 2: I suggest to try using a grey shaded relief as background in this Figure. Also, please add a metric scale bar.
Fig. 3: as above.
Fig. 5: as above
Fig. 6: as above. Please note: a) The plotted axes are those from the strain rate tensor b)The strain scale bar labels shows 20 ns/yr. c) Also add the epricentre of the Eh-Haouz 2024 earthquake in this Figure. This earthquake [https://earthquake.usgs.gov/earthquakes/eventpage/us7000kufc/executive] is a result of compression and its location and kinematics may confirm your model.
References in this review.
Briole, P., et al. 2021. The GPS velocity field of the Aegean. New observations, contribution of the earthquakes, crustal blocks model, Geophysical Journal International, Volume 226, Issue 1, Pages 468–492, https://doi.org/10.1093/gji/ggab089
Fernandes R, et al. 2022. A new European service to share GNSS Data and Products. Ann. Geophys. [Internet]. 2022Jul.12 [cited 2022Jul.21];65(3):DM317 https://doi.org/10.4401/ag-8776
Author Response
First of all, thank you for your time and comments. I hope we have been able to address all the corrections and suggestions.
Comments 1) it is interesting to compare the velocity products with the EPOS dataset (Fernandez et al. 2022; doi:10.4401/ag-8776). The products are available from https://gnssproducts.epos.ubi.pt/.
Response 1) Thank you for the information about EPOS; we will certainly take it into account. In our case, as you can see in Table 3 (which is just an example), we have already compared the data we obtained directly with IGS. In future studies, we could consider it, but in this area, there are no stations from any source that have not already been utilized. Additionally, I would like to add that SARI is currently used by the EPOS network in France, so the EPOS velocities we have used, which I believe come from the IGS, are most likely the actual EPOS velocities. I hope I'm not mistaken! We have added citation [40].
Station |
Total Vel (mm/year) |
SD (mm/year) |
EURA Vel (mm/year) |
SD (mm/year) |
||||||||
East |
North |
East |
North |
East |
|
North |
|
East |
|
North |
||
SFER PRIDE |
16.55 |
16.98 |
±0.001 |
±0.001 |
-3.12 |
|
0.32 |
|
±0.001 |
|
±0.001 |
|
SFER IGS |
16.37 |
17.02 |
±0.025 |
±0.021 |
-3.28 |
|
0.39 |
|
±0.028 |
|
±0.025 |
|
Comments 2) some discussion is necessary about the influence of co-seismic offfsets on the GNSS position time series. For example, the velocity product may have been affected by the 2016 Al Hoceima earthquake (https://earthquake.usgs.gov/earthquakes/eventpage/us10004gy9/executive). I see several GNSS stations within a 100 km radius around the epicentre. Indeed, I see a small offset (a few mm) in the East component of station ALBO in the NGL time series [http://geodesy.unr.edu/NGLStationPages/stations/ALBO.sta].
Response 2) The two stations affected by this earthquake are ALBO and MELI, where the coseismic offset has been introduced in both cases. The rest of the stations either do not have data for those dates or do not appear to have been affected by this event. We have included a paragraph as indicated LINES 240-242.
Comment 3) Also, it is worthy to discuss the M6.8 El-Haouz earthquake [https://earthquake.usgs.gov/earthquakes/eventpage/us7000kufc/executive] which is a result of compression and its location and kinematics may confirm the strain-rate results of this study.
Response 3) In this case, none of the stations are affected by this event, so it has not been necessary to introduce any offset or modify the time series, which is why we have not mentioned it. If you look closely, our results indicate extension in that area, but the actual event occurs outside the area covered by the stations, making further verification impossible.
Comment 4) in Fig. 7 I note that the usual way to display the rotation rate is in degrees per million years and the shear strain in nanostrains per year (see doi:10.1002/2014JB011762)
Response 4) Solved! Thanks!
Comment5) Some stations (e.g.AVER, LIJA) show a non-linear rate of motion which is not expected for geodynamic studies and probably reflect some local effect or a transient signal. This needs to be discussed in view of using the station velocities to compute strain rates and rotations.
Response 5) It is true that both stations may exhibit local phenomena, especially AVER, which appears to have a hidden periodicity that we have not yet been able to identify. We intend to conduct a more in-depth study in the future covering this area, where these local effects could be due to the geology of the terrain, such as the presence of submarine aquifers or the accumulation of sediments. Other episodic stations exhibit similar behavior, which could help determine the behavior of the coastal plateau, but this should be discussed in another study.
Minor comments:
Line 209: please add more information on the 2016 earthquake (date, location, magnitude). I suggest to see this paper https://doi.org/10.1093/gji/ggab089 for removing earthquake effects from the time series of GNSS position data.
and Lines 250-253: as above.
Response: Although it may seem present, if you look closely at the filtering model, an offset is applied, which appears as a dotted line.
Line 328. This is Figure 8.
Response: Solved, Thanks!
Fig. 2: I suggest to try using a grey shaded relief as background in this Figure. Also, please add a metric scale bar.
and Fig. 3: as above.
and Fig. 5: as above
Response: My real intention here is for it to be more decorative rather than to display the topobathymetry of the area. That is why I included Image 1, to avoid excessive saturation of the image or the data presented, as it is not particularly relevant to what is being shown at that moment. Tanks for the suggestion.
Fig. 6: as above. Please note: a) The plotted axes are those from the strain rate tensor b)The strain scale bar labels shows 20 ns/yr. c) Also add the epricentre of the Eh-Haouz 2024 earthquake in this Figure. This earthquake [https://earthquake.usgs.gov/earthquakes/eventpage/us7000kufc/executive] is a result of compression and its location and kinematics may confirm your model.
Response: For the earthquake catalog, we have used that of the Instituto Geográfico Nacional de España, as it is our national institution and, together with Morocco, has the closest stations. This source is properly referenced. The earthquake appears as the purple marker in the southwest corner. As mentioned earlier, the arrows indicate extension in this area, but this could also be influenced by this event. Thanks for the link!
Reviewer 2 Report
Comments and Suggestions for Authors1.The technoligy of monitoring crustal plate movement using time series observed by GNSS stations is already quite mature. What is the difference between the research method used this paper and that in the existing relevant papers? What is the innovation of the method used in this paper?
2.The explanation of the velocity field solution process is insufficient. What method was used in the paper and how the velocity data was obtainted is hardly elabor? Please add an explanation.
3.The paper has spent a lot of space on the interpretation and explanation of the velocity field results, which has verified the feasibility of the method used in this, but from the application of GNSS technology, there is a lack of innovation.
Author Response
First of all, thank you for taking the time to review our article. I would like to clarify that this is not a new methodology, but rather a presentation of initial insights into a new velocity field that could help improve the understanding of the EURA-NUBIA plate convergence zone. It is important to highlight that almost all previous work in the area has been based on episodic campaigns, without permanent stations like those presented in this study.
Comment 1. The technoligy of monitoring crustal plate movement using time series observed by GNSS stations is already quite mature. What is the difference between the research method used this paper and that in the existing relevant papers? What is the innovation of the method used in this paper?
Response 1. This article does not introduce any novelty in terms of processing methodology, as that is not its purpose. As indicated in the title, the objective is to present a new crustal velocity field in a little-known area with new stations and long time series.
Comment 2. The explanation of the velocity field solution process is insufficient. What method was used in the paper and how the velocity data was obtainted is hardly elabor? Please add an explanation.
Response 2. The entire methodology used is detailed in section 2.2. The processing method is not discussed further since everything is referenced to the methodology used.
Comment 3. The paper has spent a lot of space on the interpretation and explanation of the velocity field results, which has verified the feasibility of the method used in this, but from the application of GNSS technology, there is a lack of innovation.
Response 3. I apologize for being repetitive, but as I mentioned in Comment 1, this is not the purpose of the article. The aim is to introduce a new velocity field using permanent stations in locations that have never been studied before, in an area of great geophysical and geological interest. This is not a technical study on software or processing methodology.
We sincerely regret not being able to provide more answers to your comments beyond those we have presented, but we believe we cannot add anything further. Once again, thank you for your collaboration and for helping us improve.
Reviewer 3 Report
Comments and Suggestions for AuthorsIn this work , the displacement velocities of the ROA 27 CGNSS stations have been estimated to provide for the first time, the work sound the community. The writting and organzition need improved, especially the abstract and conclusion. For details see:
1.Abstract: line 18-27, to long introduction. The beginning of the abstract needs to be simplified.
2.Figure. 4 on the right residuals of these filtered data: did not give more details on filter method? Besides, there existed "offsets" on ALBO time series, how did the author process the offset (with ,MIDAS or SARI).
3.The authors missed some literatures, e.g., lots of research pointed out that "offset" will effect the velocity derived from GNSS time series ([1]Huang, J., He, X., Hu, S., Ming, F., Impact of Offsets on GNSS Time Series Stochastic Noise Properties and Velocity Estimation, Advances in Space Research (2024), doi: https://doi.org/10.1016/j.asr. 2024.12.016. & [2] Wang, L. and Herring, T., 2019. Impact of estimating position offsets on the uncertainties of GNSS site velocity estimates. Journal of Geophysical Research: Solid Earth, 124(12), pp.13452-13467.)
4.In Tab.3 "0.001 mm/yr", can GNSS get this level of accuracy?
5.Figure. 7: enlarge the font, make it easy to follow.
6.In the conclusion "This observation is consistent with previous 303 studies referenced [2-5].", the author need pointed out the spetific value of the work [2-5], also 4.5 mm/yr? or close to it?
7.Also for "This corroborates findings from 307 several studies, such as [2,14,39,40].", The author needs to give a concluding description, not just a simple quote. It is recommaned to revised the discussion, to make it more easy to follow, especially hilgh-lighted the contribution of this work.
8. "In summary, a network of significant interest has been introduced in a sparsely populated area, which could complement many past studies. It is important to emphasize the need for and the effort involved in maintaining such networks, as they are essential for complementing other geophysical and geological studies": It is more like a subjective conclusion?
Comments on the Quality of English Languagenone
Author Response
First of all, we sincerely appreciate your time and dedication in helping us improve this article. We will do our best to incorporate all the suggested changes.
Comment 1.Abstract: line 18-27, to long introduction. The beginning of the abstract needs to be simplified.
Response 1. Done.
Comment 2.Figure. 4 on the right residuals of these filtered data: did not give more details on filter method? Besides, there existed "offsets" on ALBO time series, how did the author process the offset (with ,MIDAS or SARI).
Response 2. The offset is therefore determined using SARI, while MIDAS is only used to verify the velocities (through SARI). The MIDAS methodology is based on the median and does not account for offset correction. In the applied methodology, section 2.2, we have included the following: For model fitting, the least squares (LS) option was used [41]. From among the options, the linear trend is introduced, the discontinuities detected in the series by defining the reference epochs, and the periodic signals with a reference epoch that is the average of the available measurement data, where the phase values are estimated accordingly. For velocity verification with SARI, the Median Interannual Difference Adjusted for Skewness (MIDAS) algorithm [42] has been used. An analysis of noise is performed to estimate the complete variance-covariance matrix that optimally characterizes the model or filter residuals as a Gaussian process, using maximum likelihood estimation (MLE) to determine the parameters of a selected covariance model.
Comment 3. The authors missed some literatures, e.g., lots of research pointed out that "offset" will effect the velocity derived from GNSS time series ([1]Huang, J., He, X., Hu, S., Ming, F., Impact of Offsets on GNSS Time Series Stochastic Noise Properties and Velocity Estimation, Advances in Space Research (2024), doi: https://doi.org/10.1016/j.asr. 2024.12.016. & [2] Wang, L. and Herring, T., 2019. Impact of estimating position offsets on the uncertainties of GNSS site velocity estimates. Journal of Geophysical Research: Solid Earth, 124(12), pp.13452-13467.)
Response 3. Introduced, many thanks! It is truly interesting to take this into account, especially for future studies with our stations.
Comment 4. In Tab.3 "0.001 mm/yr", can GNSS get this level of accuracy?
Response 4. It is a typographical error; the corrected table is accurate to the tenth of a millimeter. Many thanks!!
Comment 5.Figure. 7: enlarge the font, make it easy to follow.
Response 5. Done.
Comment 6. In the conclusion "This observation is consistent with previous 303 studies referenced [2-5].", the author need pointed out the spetific value of the work [2-5], also 4.5 mm/yr? or close to it?
Response 5. The following has been introduced: where velocities ranging between 4-5 mm/yr are shown. Based on articles 2-5.
Comment 7.Also for "This corroborates findings from 307 several studies, such as [2,14,39,40].", The author needs to give a concluding description, not just a simple quote. It is recommaned to revised the discussion, to make it more easy to follow, especially hilgh-lighted the contribution of this work.
Response 7. I would like to remind you that this work that this is a very general first analysis of a highly complex area, which has no permanent stations except for those we have presented. The articles that we cite study the plate convergence in a very general way, often with only one or two permanent stations or using episodic surveis. New permanent stations from ROA provide continuous data and help clarify a complex region with different plate boundaries within the diffuse zone of the Gibraltar Arc. Once a very general comparison of the results is made, we explain each zone in as much detail as possible while keeping the discussion concise. In future work, a more exhaustive analysis of the different study areas within this region will be conducted, considering other techniques and methodologies, such as our own ROA seismic network.
Comment 8. "In summary, a network of significant interest has been introduced in a sparsely populated area, which could complement many past studies. It is important to emphasize the need for and the effort involved in maintaining such networks, as they are essential for complementing other geophysical and geological studies": It is more like a subjective conclusion?
Response 8. I have modified it by adding more information to make it sound less subjective. I just wanted to highlight the value of the work carried out to obtain these data.
Reviewer 4 Report
Comments and Suggestions for AuthorsDear Authors,
The availability of new GNSS stations in southern Spain and North Africa is very interesting because we can improve the knowledge of the geodynamics in this boundary zone between the EURA and NUBIA plates. However, the manuscript does not contain information on GNSS processing, and some inaccuracies are present.
The main critical points are as follows.
- More information is needed on the functional model used to fit the GNSS time series, as the authors didn't provide details. Which periodic signals? annual? Semi-annual? Also, I'm not sure I understand the sentence on lines 165-168. What reference epoch were the authors referring to?
- Have you checked if there are any stations affected by the earthquake in Morocco (M. 6.8 - Al Haouz, September 2023)?
- More information is also necessary on the stochastic model. The authors claim (line 170) that a noise analysis was carries out to estimate the full variance-covariance matrix. Which models were chosen?
- In Figure 4, on the right, the residuals of the filtered data are plotted. What do you mean with “filtered”? The residuals, i.e. observations minus modeling? If this is correct, I think that the term “filtered” should be removed.
- In the caption of Figure 6 and on Line 228 the authors report the stress tensor. Are you sure about this? I think you mean the strain/rate tensor. In fact, in the label of Figure 6 the unit is in strain/yr. Also check the term “shear stresses” on line 291. Do you mean strain?
- Figure 6, on the left. There is too much colors. I suggest reporting the GNSS sites all with one color.
- There are two figures 7.
- Figure 7 line 267, why do the rotation and shear deformation images (is that correct?) not cover the whole area with the GNSS stations?
- Figure 7 line 328, you plot the velocities of continuous and episodic stations together. Have you put all the solutions in the same reference frame? The solution of this study is in IGb2020, but the episodic solutions come from previous works ([4], [10], [18], where different reference frames are used.
- In Tab.2 I suggest changing the label “Total velocity” to “absolute velocity” or “IGb20 velocity”
- Line 225, 238, 271, 329: “residual velocities wrt EURA” —> “velocities wrt EURA”
Author Response
First of all, we sincerely appreciate your reviewing work. Thank you for helping us improve this article and allowing us to finalize it in the best possible way.
Comment 1. More information is needed on the functional model used to fit the GNSS time series, as the authors didn't provide details. Which periodic signals? annual? Semi-annual? Also, I'm not sure I understand the sentence on lines 165-168. What reference epoch were the authors referring to?
Response 1. We appreciate your comment and the opportunity to clarify these points. It has been added to the methodology between lines 180-189 to address the reviewer's comments.. Regarding the functional model and periodic signals: The model used to fit GNSS time series in SARI allows for the inclusion of various deterministic components, such as linear trends, higher-degree polynomials, discontinuities, sinusoidal signals, and exponential or logarithmic decays. Specifically, periodic signals are fitted using a decomposition into sine and cosine terms. Regarding the considered periods, the user can manually define the periods of the sinusoidal signals to be fitted. Additionally, SARI provides the option to include draconitic harmonics for each selected GNSS constellation, allowing the modeling of periodic variations induced by satellite orbits. These details are described in the "Fit controls" and "Notes on the sinusoidal fitting" sections of the documentation of the software. Regarding the reference epoch. The "reference epoch" mentioned in this section refers to how SARI handles the time units of the loaded time series. Specifically, the user must ensure that the time units are correctly defined when loading the series, as they will determine the temporal resolution of the analysis. If changed later, the time axis scale will adjust, but the data sampling itself will not change. This information is crucial for ensuring a consistent analysis, especially when working with different series formats (NEU/ENU, PBO, NGL, etc.). We are going to introduce some more information about it, but is sometihng about the software...
Comment 2. Have you checked if there are any stations affected by the earthquake in Morocco (M. 6.8 - Al Haouz, September 2023)?
Response 2. Yes, of course. In fact, we are working on another article where we attempt to identify earthquakes using high-frequency signals, as well as possible precursors. In the case of 30-second signals, we have not been able to observe anything.
Comment 3. More information is also necessary on the stochastic model. The authors claim (line 170) that a noise analysis was carries out to estimate the full variance-covariance matrix. Which models were chosen?
Response 3. Introduced: SARI models stochastic noise in GNSS time series using Gaussian processes and optimizes the covariance matrix through Maximum Likelihood Estimation (MLE). We use a combination of white noise and random walk, adjusting their parameters and selecting the best model based on log-likelihood. It also estimates the impact of noise on trend uncertainty and offers an option to separate noise components using a Wiener filter. To improve efficiency, we validate the fit by comparing the power spectrum of residuals with the estimated noise model.
Comment 4. In Figure 4, on the right, the residuals of the filtered data are plotted. What do you mean with “filtered”? The residuals, i.e. observations minus modeling? If this is correct, I think that the term “filtered” should be removed.
Reponse 4. Removed.Thanks!
Comment 5. In the caption of Figure 6 and on Line 228 the authors report the stress tensor. Are you sure about this? I think you mean the strain/rate tensor. In fact, in the label of Figure 6 the unit is in strain/yr. Also check the term “shear stresses” on line 291. Do you mean strain?
Responsed 5. Corrected.
Comment 6. Figure 6, on the left. There is too much colors. I suggest reporting the GNSS sites all with one color.
Response 6. It is the only way to view the different types of stations. For now, we believe we will leave it as it is unless a change is necessary. Thank you for your observation.
Comment 7. There are two figures 7.
Response 7. Changed.Many thanks!
Comment 8. Figure 7 line 267, why do the rotation and shear deformation images (is that correct?) not cover the whole area with the GNSS stations?
Response 8. If you look closely, we can only cover deformations within the area of the stations. If we place something outside of it, we would be modeling or perhaps even inventing the data, as there is no information outside the grid.
Comment 9. Figure 7 line 328, you plot the velocities of continuous and episodic stations together. Have you put all the solutions in the same reference frame? The solution of this study is in IGb2020, but the episodic solutions come from previous works ([4], [10], [18], where different reference frames are used.
Response 9. Exactly, that is correct. For this very reason, we cannot use these same results to model deformations. Additionally, these are episodic stations, and the data are not entirely reliable, as they may hide local or transient effects. They are only introduced for comparison purposes. In any case, the velocity differences between these results should be small since it is a very localized area. Considering that the stations from the previous TopoIberia and episodic studies are referenced to IGB14, we have used them as a general comparison and extrapolated this to the rest of the stations. The difference between reference frames in this area is minimal. Additionally, we have conducted comparisons, but they were not included in the article to avoid excessive length.
Comment 10. In Tab.2 I suggest changing the label “Total velocity” to “absolute velocity” or “IGb20 velocity”.
Response 10. Done. Thanks!
Comment 11. Line 225, 238, 271, 329: “residual velocities wrt EURA” —> “velocities wrt EURA”
Response 11. Done. Thanks!
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsThe author has response to all my comments and improved the manuscript. It can be accepted in present version.
Reviewer 4 Report
Comments and Suggestions for AuthorsDear authors,
Thank you for taking my comments into consideration and for your explanations.