The Study of the Lithospheric Magnetic Field over Xinjiang and Tibet Areas Based on Ground, Airborne, and Satellite Data
A.V. Pospeev
Round 1
Reviewer 1 Report
(1) I am sorry to say that the language of the submitted manuscript is so terrible and hard to understand. There are too many grammar mistakes. Some sentences are too lengthy without a specific subject (Lines 37-40, Lines 46-50, to list a few). Lines 239-242 are just copies of Lines 232-235.
(2) The descriptions in sections of Research data and Involved models are quite plain. It’s better to give more details for each modelling method, especially for the three-dimensional surface spline method, which the presented work highly depends on.
(3) To be honest, I am not familiar with the 3DSS method, but through Eq. (1) and (2), the resultant model by 3DSS should be continuous, just like SHA model. It is easily seen that the fields on the two sides of the border of Xinjiang and Tibet (as shown in Figure 2b to e, black solid line) do not meet the continuous condition, unlike SHA, NGDC, RSHA, LCS models (as shown in Figure 2f to j). I suggest the authors carefully check that.
Furthermore, the results calculated by 3DSS model shown in Figure 3 are also doubtful. Let Figure 3k to be ground true, Figure 3a to j are just downward continuation of Figure 3k, and Figure 3I to u are upward continuation of Figure 3k. Theoretically speaking, the downward continued results should have more details than the original data, which is not true for Figure 3a to f. Is that because the 3DSS method is not stable or something else?
For upward continuation case, the results should be smoother, but considering the resolution of original data (about 1.1km) and upward continued distance (no more than 1 km), the fields should not change so much as shown in Figure 3p to u.
Author Response
(1) I am sorry to say that the language of the submitted manuscript is so terrible and hard to understand. There are too many grammar mistakes. Some sentences are too lengthy without a specific subject (Lines 37-40, Lines 46-50, to list a few). Lines 239-242 are just copies of Lines 232-235.
I am so sorry about my poor English language, and the whole manuscript was checked by native speaker, hope it looks much better.
(2) The descriptions in sections of Research data and Involved models are quite plain. It’s better to give more details for each modelling method, especially for the three-dimensional surface spline method, which the presented work highly depends on.
About the data, due to the selection criteria is basically same compare with other model, so I just illustrate several important points there. Here I have added more descriptions in satellite data part. I have also attached the computing algorithm of 3DSS model, and the math expression of RSHA.
(3) To be honest, I am not familiar with the 3DSS method, but through Eq. (1) and (2), the resultant model by 3DSS should be continuous, just like SHA model. It is easily seen that the fields on the two sides of the border of Xinjiang and Tibet (as shown in Figure 2b to e, black solid line) do not meet the continuous condition, unlike SHA, NGDC, RSHA, LCS models (as shown in Figure 2f to j). I suggest the authors carefully check that.
This is a good point. The main reason is lack of the measuring data outside the Chinese boundary, so the obvious boundary distortion can be found outside the black line even I had added the supplementary points to alleviate it. But so lucky that the distribution in research area is not influenced and have a good consistent with other models, so I think it does not very matter. I have added some related descriptions in article.
By the way, other four models were derived by measuring data worldwide, so the distributions are continuous.
(4) Furthermore, the results calculated by 3DSS model shown in Figure 3 are also doubtful. Let Figure 3k to be ground true, Figure 3a to j are just downward continuation of Figure 3k, and Figure 3I to u are upward continuation of Figure 3k. Theoretically speaking, the downward continued results should have more details than the original data, which is not true for Figure 3a to f. Is that because the 3DSS method is not stable or something else?
For upward continuation case, the results should be smoother, but considering the resolution of original data (about 1.1km) and upward continued distance (no more than 1 km), the fields should not change so much as shown in Figure 3p to u.
Actually the 3DSS model is seriously affected by the data density both in horizontal and vertical directions, like I said, there are only 519 grounded points distributed from 0-0.7km and account for 1.52% of total points, moreover, the quality of ground data is worst compare to other data due to different equipment, measuring methods, and methods to deal with raw data in different epochs, so it is not strange that the distribution around the surface is not clear. On contrary, at the 1km level, good distribution can be seen due to the high density and consistency aeromagnetic data, which account for 36.75% of total points.
In order to verify our results, the distribution of field just based on 519 ground data is listed to compare with (figure 4).
As for the upward continuation, I think the main reason is the measuring points decrease rapidly along the vertical direction, and the distribution is only controlled by the nearest data, e.g. aeromagnetic and CHAOS-7.11 model data.
I have added some descriptions in the conclusions part.
Author Response File: 
Author Response.docx
Reviewer 2 Report
A. General comments
This paper firstly applied the three-dimensional surface spline (3DSS) and regional spherical harmonic analysis (RSHA) methods to the geomagnetic field measurements from ground, airborne and CHAMP satellite surveys, and obtained two regional lithospheric magnetic field models over the Xinjiang-Tibet areas. Then, the constructed models were compared with the latest global/regional lithospheric magnetic field models in the literatures. Finally, the total magnetic intensity anomaly fields at different altitudes were predicted based on the regional 3DSS model and were preliminarily interpreted from the geological and geothermal viewpoints. The lithospheric magnetic field study is an important remote sensing way to probe and understand the magnetization structures in the lithosphere. Thus, this point fits for the object of the “Remote Sensing”. Despite there is no innovation point for the modelling methods and data processing techniques utilized in this paper, this study presents the three-dimensional lithospheric magnetic field models with a very high resolution over the Xinjiang-Tibet areas. However, in my personal view, there are some points need to be further improved, which are provided as follows. These comments are just given to the authors for reasonable consideration. Therefore, the paper is recommended for publication in Remote Sensing after a minor revision.
B. Specific comments
(1) The English level should be improved drastically. There are many spelling and grammatical mistakes. Many sentences do not read smoothly.
(2) The ground measurements are sparsely distributed and the aeromagnetic anomaly data is very dense in the spatial distribution. Thus, the role of the ground data in the modelling should be reconsidered. In my opinion, it’s better to adopt the ground data to check the modelling accuracy.
(3) In the 146th line, the manuscript described that the aeromagnetic data are scalar magnetic anomaly at 1 km altitude. However, in fact the 1 km altitude means the terrain clearance. Therefore, the question is that whether the modelling considered the topography variations or not. Furthermore, the sub-figures (a)~(i) in the Figure 3 are very strange. The results do not obey the decay law of potential field.
(4) There is a slight significance for testing the sensitivity of the data number of the 3DSS model. On the one hand, the selected data not used by the model should be presented as the supplementary figure. On the another hand, the selected data set not used by the model is suggested as the accuracy check data, because utilizing only 11 points to check the RMSE of the model in the section 4.3.2 lacks persuasion. Moreover, what’s the meaning of the word “abscent” in the Figure 5?
(5) In the equation (1), why a linear term is added in the expression? Besides, the different kinds of data sets are used for the modelling, but the 3DSS method can not physically fuse different component data sets. So, the question is that how the authors perform the data fusion by the 3DSS method.
(6) In the section “3. Involved models”, the title is suggested to be revised as “Modelling methods”. For the 3DSS modelling method, only the model expressions are provided and the calculating process and computing algorithm are lack. However, as for the RSHA modelling method, only the calculating process and computing algorithm are provided but the model expressions are lack. So, this section is suggested to be revised by adopting a uniform form and contents.
(7) As for the 3DSS modelling method, the ε is a small value that controls the changes on the surface curvature and thus it is very important in the modelling. However, the manuscript does not provide the method for setting the optimal value. And in the practical application, is there a uniform criterion for setting the optimal value?
(8) In the geomagnetic field modelling, the data misfit is an important way to judge the quality of the model. Therefore, the data misfit and its statistical parameters are suggested to be added in the revised manuscript.
(9) In the field modelling, multi-source data sets with different errors are utilized. So, the questions are that how to evaluate the errors and how to consider the data errors in the modelling.
(10) Despite the three 3DSS models and SHA1000 model are compared with the existing global and regional lithospheric field models, however, the paper only give the comparison in the spatial domain, and as an important analysis way the power spectrum analysis is lack. So, I suggest the authors add the comparison from the wave-number perspective in the revised manuscript.
(11) The heat flow and geothermal gradient models as presented in the Figure 9 are very strange and unbelievable. In fact, there are many global and regional heat flow and geothermal gradient models in the literature. So I suggest that it’s better to directly use the existing terrestrial heat flow models in the literature.
(12) As shown in Figure 9, there is a weak comparability between the total magnetic intensity anomaly distribution and the heat flow and geothermal gradient maps, and in fact they have different spatial resolutions. Therefore, the long wavelength components of the total magnetic intensity anomaly distribution are suggested to be extracted from the models first and then the correlation analysis can be performed. Alternatively, the spatial variations of the depth to the Curie isotherm can be calculated first and then some comparison analysis can be carried out.
C. Technical comments
(1) The physical quantity or the geomagnetic component in the Figure 2 should be provided.
(2) The words “m” and “meter” as well as the “LCS-1” and “LCS_1” should be unified in the manuscript.
(3) The expression “3DSS34039” should be changed as “3DSS34039” in the 311th line of the 11th page.
(4) In the 64th line, the symbols (i.e., F, D and I) should be expressed by the italic or removed.
(5) In the 31th line, the crustal field and the lithospheric field are different. So, the expression should be revised.
(6) The keyword “Swarm” is suggested to be revised as “CHAMP”, and the “Xinjiang” or “Tibet” is suggested to be added as the keyword.
(7) In the Figure 6, a tectonic map over the topography variations is suggested to be added as a sub-figure.
(8) In the geological interpretation, some important references are lack, such as the lines 358~361. Please check the full manuscript.
(9) The units of the physical quantities in the Figure 9 are lack. And the meaning of the symbol * is not consistent with the description in the main text.
(10) There some negative values in the color bar of the Figure 9, however, the heat flow are positive in the practical world. Moreover, it is better to fill the heat flow measurement points shown in the Figure 9 by the same color variations with the map’s color bar. Thus, it is more easily to assess the quality of the heat flow model.
(11) The expression “Geological explanation” is suggested to be revised as “Geological interpretation”.
(12) As for the Figure 2 and Figure 3, if the color bars of the subfigures are the same, it is better to provide only one color bar.
Author Response
- Specific comments
(1) The English level should be improved drastically. There are many spelling and grammatical mistakes. Many sentences do not read smoothly.
The language has been revised by a native speaker. We have also checked the manuscript several times, hope it looks much better.
(2) The ground measurements are sparsely distributed and the aeromagnetic anomaly data is very dense in the spatial distribution. Thus, the role of the ground data in the modelling should be reconsidered. In my opinion, it’s better to adopt the ground data to check the modelling accuracy.
Good idea, we have done this work (figure 4) and found a good consistency between them.
(3) In the 146thline, the manuscript described that the aeromagnetic data are scalar magnetic anomaly at 1 km altitude. However, in fact the 1 km altitude means the terrain clearance. Therefore, the question is that whether the modelling considered the topography variations or not. Furthermore, the sub-figures (a)~(i) in the Figure 3 are very strange. The results do not obey the decay law of potential field.
To be honest, we did not consider the topography variation. The airborne data is just the grids data at 1 km level, we suppose that the research area is a flat plain and use different data with different altitudes.
The other reviewer also put forward the similar question, so we’ve calculated the 3DSS model based only ground data (figure 4), and verified the reliability of distribution at ground. Results imply the 3DSS model is seriously affected by the data density both in horizontal and vertical directions, like the descriptions in article, there are only 519 ground points distributed from 0-0.7km and account for 1.52% of total points, moreover, the quality of grounded is worst compare to other data due to different equipment, measuring methods, and methods to deal with raw data during 1936-2000, so it is not strange that the distribution around the surface is not clear. On contrary, at the 1km level, good distribution can be seen due to the high density and consistency aeromagnetic data, which account for 36.75% of total points.
I have added some descriptions in the conclusions part.
(4) There is a slight significance for testing the sensitivity of the data number of the 3DSS model. On the one hand, the selected data not used by the model should be presented as the supplementary figure. On the another hand, the selected data set not used by the model is suggested as the accuracy check data, because utilizing only 11 points to check the RMSE of the model in the section 4.3.2 lacks persuasion. Moreover, what’s the meaning of the word “abscent” in the Figure 5?
Ok, I have drawn the supplementary figures of 11 points, which is just the accuracy check data. It is really hard to say how many data is enough to test the sensitivity, and we have tried different numbers of data in different altitudes and find the same results, the more data to create a 3DSS model, the better the result.
The word should be “absent”, it is a stupid mistake.
(5) In the equation (1), why a linear term is added in the expression? Besides, the different kinds of data sets are used for the modelling, but the 3DSS method can not physically fuse different component data sets. So, the question is that how the authors perform the data fusion by the 3DSS method.
The linear term added is to makeup the coefficients that need to calculated.
About the data fusion, we really considered it, the way we selected is to add CHAOS model data to alleviate the possible disorder, because the model is derived by CHAMP data, so a good transition between surface to satellite data can be obtained. Moreover, to check distributions along the different altitude to see if any sudden change happens.
(6) In the section “3. Involved models”, the title is suggested to be revised as “Modelling methods”. For the 3DSS modelling method, only the model expressions are provided and the calculating process and computing algorithm are lack. However, as for the RSHA modelling method, only the calculating process and computing algorithm are provided but the model expressions are lack. So, this section is suggested to be revised by adopting a uniform form and contents.
I have changed the title of section 3.
As for 3DSS model, the computing algorithm is attached; as for RSHA model, whose theory is SHA based one, and just calculate the new coefficients in the research area and add it to the base coefficients, the SHA expression is added there.
(7) As for the 3DSS modelling method, the ε is a small value that controls the changes on the surface curvature and thus it is very important in the modelling. However, the manuscript does not provide the method for setting the optimal value. And in the practical application, is there a uniform criterion for setting the optimal value?
Yes, in geomagnetic field modelling, the ε is setting to this fixed value 1×10-7 after tests. I added it in the article.
(8) In the geomagnetic field modelling, the data misfit is an important way to judge the quality of the model. Therefore, the data misfit and its statistical parameters are suggested to be added in the revised manuscript.
There is no misfit between 3DSS and measuring data due to the interpolation theory, so we chose 11 points (actually tried different number of absent points) that from different kind data set, and create model based on rest data, then compare modeling values with 11 points to test sensitivity.
(9) In the field modelling, multi-source data sets with different errors are utilized. So, the questions are that how to evaluate the errors and how to consider the data errors in the modelling.
Like I mentioned in the previous question, there are no misfit due to interpolating theory, so we chose different numbers and kind of data out of modelling, and calculate them by forward modelling based on rest data. Results show the misfit of CHAOS model data is best, that of satellite and aeromagnetic data is less good, that of ground data is worst due to the different measuring equipment, measuring methods, etc.
(10) Despite the three 3DSS models and SHA1000 model are compared with the existing global and regional lithospheric field models, however, the paper only give the comparison in the spatial domain, and as an important analysis way the power spectrum analysis is lack. So, I suggest the authors add the comparison from the wave-number perspective in the revised manuscript.
This study mainly focuses on the 3DSS modeling, and compare it with other models. 3DSS model is a pure regional model and quite different with global models, even the RSHA, moreover, the coefficients are not spherical harmonic coefficients. As a result, I think to list the power spectrum of 3DSS means nothing.
Here I list the first 20 coefficients of 3DSS, SHA1000, and CHAOS-7.
|
3DSS |
SHA1000 |
CHAOS-7.13 |
|
-45997.9 |
-29496.6 |
-29654.9576 |
|
29.67809 |
-1586.29 |
-1760.726415 |
|
7.715141 |
4944.236 |
5254.221373 |
|
-52.4511 |
-2396.57 |
-2228.002307 |
|
1395.01 |
3026.034 |
3071.738183 |
|
125.8851 |
-2707.77 |
-2415.199348 |
|
43689.71 |
1668.608 |
1678.996626 |
|
1611.33 |
-576.168 |
-433.2792663 |
|
-80.7001 |
1340.097 |
1338.640397 |
|
354413.1 |
-2326.19 |
-2276.639787 |
|
3009.886 |
-160.093 |
-245.7129006 |
|
2282.209 |
1231.847 |
1252.032278 |
|
121.2864 |
251.9099 |
302.3950123 |
|
-1751.01 |
634.0987 |
742.055056 |
|
815.8723 |
-536.638 |
-454.4998193 |
|
29192.61 |
912.6094 |
939.4304326 |
|
-4640.07 |
808.9485 |
782.4059621 |
|
601.9865 |
286.3638 |
265.9817579 |
|
277.4271 |
166.6647 |
274.100606 |
|
1549.138 |
-211.147 |
-234.50487 |
So I did not calculate the power spectrum of 3DSS and compare with.
(11) The heat flow and geothermal gradient models as presented in the Figure 9 are very strange and unbelievable. In fact, there are many global and regional heat flow and geothermal gradient models in the literature. So I suggest that it’s better to directly use the existing terrestrial heat flow models in the literature.
Here I plotted the figure by directly use the heat flow data, and model it by 2D Surface Spline (2DSS) model in case of the same altitudes. And compared them with Hu’s results and lithospheric field (figure 10). Results show a good consistency between them, and the features between lithospheric field and heat flow have been analyzed in article.
(12) As shown in Figure 9, there is a weak comparability between the total magnetic intensity anomaly distribution and the heat flow and geothermal gradient maps, and in fact they have different spatial resolutions. Therefore, the long wavelength components of the total magnetic intensity anomaly distribution are suggested to be extracted from the models first and then the correlation analysis can be performed. Alternatively, the spatial variations of the depth to the Curie isotherm can be calculated first and then some comparison analysis can be carried out.
Our previous work is questionable due to is over-modeling the data, which results in the Non-full rank of matrices calculation and yielded the distortion of boundary, than I have chosen the 2DSS model, just suppose all altitudes of measuring points are same, to modeling the distribution, which was then compared to the distribution by the only measuring points.
Results of heat flows show a good consistence each other, and also consistent with Hu’s result. We just take a slight comparison whit heat flow to try to find something interesting, results show a good coincidence between a positive stripe of heat flow and a positive stripe of the lithospheric field in the southern Tibet, high heat flow areas correspond to the low intensity of lithospheric field, and vice versa in Xinjiang, particular in Sourth Tarim and Junggar Basin magnetic anomalies.
The detailed analyses are described in article.
Thanks for your kind suggestion, next step we will consider the works about the spatial variations of the depth to the Curie isotherm.
- Technical comments
(1) The physical quantity or the geomagnetic component in the Figure 2 should be provided.
“The investigations are all about the total intensity F throughout the article” was added.
(2) The words “m” and “meter” as well as the “LCS-1” and “LCS_1” should be unified in the manuscript.
They are unified in the whole manuscript.
(3) The expression “3DSS34039” should be changed as “3DSS34039” in the 311th line of the 11thpage.
It is changed.
(4) In the 64th line, the symbols (i.e., F, D and I) should be expressed by the italic or removed.
They are revised.
(5) In the 31th line, the crustal field and the lithospheric field are different. So, the expression should be revised.
Some geophysical works are carried out by crustal field, which mainly focus the near surface. So I revised as “in some cases also called the crustal field”.
(6) The keyword “Swarm” is suggested to be revised as “CHAMP”, and the “Xinjiang” or “Tibet” is suggested to be added as the keyword.
I have changed the word and attached “Tibet” in the keywords.
(7) In the Figure 6, a tectonic map over the topography variations is suggested to be added as a sub-figure.
Thanks for this point, but I think related paragraphs focus on describing the features of the lithospheric field, and few about the tectonic structure.
(8) In the geological interpretation, some important references are lack, such as the lines 358~361. Please check the full manuscript.
The references are attached there.
(9) The units of the physical quantities in the Figure 9 are lack. And the meaning of the symbol * is not consistent with the description in the main text.
This figure is deleted and replaced new one.
(10) There some negative values in the color bar of the Figure 9, however, the heat flow are positive in the practical world. Moreover, it is better to fill the heat flow measurement points shown in the Figure 9 by the same color variations with the map’s color bar. Thus, it is more easily to assess the quality of the heat flow model.
The original figure 9 is mistaken and is now replaced by new ones.
(11) The expression “Geological explanation” is suggested to be revised as “Geological interpretation”.
The title has been changed.
(12) As for the Figure 2 and Figure 3, if the color bars of the subfigures are the same, it is better to provide only one color bar.
Sure, all color bars have been deleted except the first one.
Author Response File: Author Response.docx
Reviewer 3 Report
Comments for author File: 
Comments.pdf
Author Response
As for the geological and thermal interpretation of the data obtained, there are questions here. Geomorphological – ridges and valleys are considered as primary factors, and only then references are made to the features of the geological structure. Meanwhile, the primary factor forming the anomalous magnetic field of the Earth is the distribution of the magnetic susceptibility of rocks. It seems that information about the petromagnetic properties of rocks composing such large areas as Xinjiang and Tibet, the presence of residual magnetization, including the reverse with respect to the modern polarity field. would not be superfluous here. It should be noted that the authors richly illustrated the purely magnetometric part of the article but did not present a single geological picture.
The weakest point of the presented article is the section devoted to comparison with the geothermal field. The reviewer placed below a diagram of the heat flow of the studied area from the reviewed article and a fragment of the heat flow map of China from the work (Wang et al., 2012). Firstly, the scale to the right of the map has no units of measurement. If it is MW/m2, then the range presented is infeasible and does not correspond to reality. From the scheme of Wang et al. it can be seen that the minimum surface heat flow is about 20, and the maximum is 150 MW/m2. The authors' heat flow varies from –300 to +300 MW/m2. I would like to note that negative values of heat flow are extremely rare and are associated with dynamic redistribution of heat flow by different temperature fluids, or strong technogenic warming of the near-surface part of the section.
So thanks for your kind suggestions, and thanks for your affirmation of our geomagnetic works. And of course, the heat flow part is questionable.
I have re-done the modeling of heat flow in research area, and found the problem in previous work is over-modeling the data, which results in the non-full rank of matrices calculation and yielded the distortion of boundary, than I have chosen the 2D Surface Spline (2DSS) model, just suppose all altitudes of measuring points are close, to modeling the distribution, which was then compared to the distribution by the only measuring points (figure 10)
Results of heat flows show a good consistence each other, and also consistent with Hu’s work. We just take a slight comparison whit heat flow to try to find something interesting, results show a good coincidence between a positive stripe of heat flow and a positive stripe of the lithospheric field in the southern Tibet, high heat flow areas correspond to the low intensity of lithospheric field, and vice versa in Xinjiang, particular in Sourth Tarim and Junggar Basin magnetic anomalies.
The detailed analyses are added in article.
I think these re-do works will make the manuscript better, thanks again!
Author Response File: Author Response.docx
Round 2
Reviewer 3 Report
The authors quickly and concisely took into account the remarks made by the reviewer. Since the article is more methodical than geological, we can agree that it corresponds to the level of a journal and will attract the attention of specialists studying the Earth's magnetic field. In this form, it can be published in a journal.
Author Response
So thanks for the reviewer 3's suggestions; now I have revised the manuscript, especially the explanation of symbols of equation, and an English-native speaker has checked the manuscript several times. I hope it satisfies the requirement of publishing.
Author Response File: Author Response.docx
