Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China

Round 1

Reviewer 1 Report

 

Dear Authors,

 

I think that your manuscript entitled “Stratified Data Reconstruction and Spatial Pattern  Analyses of Soil Bulk Density in the Northern  Grasslands of China” could be scientifically interesting and might bring new and valuable ideas and results to large are studies of soil bulk density.

This work could be also important from practical point of view, it is rather also tidy It is also appropriately concise and clear.

 

However, I founded some factual errors, or at least inaccuracies, and I do not agree entirely with the conclusions, which should be rewritten.

Major remarks

 

1. You write that in lines 191-194 that “Spatial distribution of grassland resource pools: based on the vector data for distribution mapping of grassland types obtained from the national grassland survey in 1980, the spatial distribution of grassland resource pools was interpreted and compiled in the northern temperate grassland regions in accordance with TM data in 2010, with a precision of 20 m.” But remote sensing and climatic data you took from the range from 2013 to 2015. This means that you used data from very different, distant years. The spatial distribution of grassland coul change.
2. In Fig. 3 R² is too small to draw so optimistic conclusion you made as “its distribution or trend is consistent with grassland type, which is mainly affected by soil organic carbon content.” You found just some rough correlation using the very very simple pedotransfer function, as you admitted.
3. I do not understand the sentence: “By comparing with the actual measurement results of other scholars, it is feasible and accurate to establish a stratified Pedotransfer function using the existing soil data, and which providing insights into the construction of a Pedotransfer function model for soil bulk density estimate at a large scale”. You did not do it, and you wrote such conclusions. Why? Make appropriate calculations, comparisons etc. Change thoroughly conclusions, please.
4. NDVI is rather rough, traditional remote sensing index, however it is greatly sensitive to many parameters which are essential for the condition of grasslands, as  for example soil moisture depending not only on climatic conditions, but also on many local conditions. Why you did not use other more sophisticated indices?  It would be much better if the conclusions would be more realistic but included more thoughtful directions for the future development of the pedofransfer function, e.g. much better use of satellite observations

 

 

Minor, but very important remarks – examples

Line: 230 Replace NDIV by NDVI.

Lines: 236-238 Non-finished sentence.

Lines: 238-239 There are 4 symbols in 237, but only are 3 explained.

Line: 255 lack of the space before (0-50cm).

Line: 350 lack of the space before "[35]".

Line: 354 lack of the space before "[44]".

Errors in eqs (3)-(6) (index "j" is lacking on the left side)

…….

Taking above into account check again carefully the whole text and improve it.

 

So, I recommend the manuscript for publication, in IJGI but only after major revision.

 

                                                                                                                                                                                                                                                        Sincerely yours

 

Reviewer

 

 

Author Response

Revision notes for “Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China”

We have revised the manuscript according to the comments/suggestions from the reviewers. The following revision notes record these changes (Underlined are comments/suggestions by the reviewers, and the text immediately followed is our response).

 

Response to Reviewer#1

 General comments

I think that your manuscript entitled “Stratified Data Reconstruction and Spatial Pattern  Analyses of Soil Bulk Density in the Northern  Grasslands of China” could be scientifically interesting and might bring new and valuable ideas and results to large are studies of soil bulk density.

This work could be also important from practical point of view, it is rather also tidy It is also appropriately concise and clear.

However, I founded some factual errors, or at least inaccuracies, and I do not agree entirely with the conclusions, which should be rewritten.

Response: Thank you very much for your constructive comments below. Your pertinent and insightful points/suggestions are greatly appreciated. We have incorporated almost all of the suggested changes into the new version of our manuscript.

Specific comments

Major remarks

1.You write that in lines 191-194 that “Spatial distribution of grassland resource pools: based on the vector data for distribution mapping of grassland types obtained from the national grassland survey in 1980, the spatial distribution of grassland resource pools was interpreted and compiled in the northern temperate grassland regions in accordance with TM data in 2010, with a precision of 20 m.” But remote sensing and climatic data you took from the range from 2013 to 2015. This means that you used data from very different, distant years. The spatial distribution of grassland could change.

Response: Thanks very much for your beneficial comments. The 1980s' data contained the most typical grassland types, commonly located in central areas of each grassland types. In this study, we applied 1980s' data as calibration value for compiling and revising the spatial map of grassland of 2010 with TM data, assuming that unlike border areas, the typical and central area of each grassland type would not change (Gao et al., 2013; Xu et al., 2019). And the 2010 spatial map of grassland resource type is the latest version which could reflect the real condition of grassland. We rewrote the sentences in line 209-214 to make our presentation clearer as below:

" Spatial distribution of grassland resource map: based on the vector data for distribution mapping of grassland types obtained from the national grassland survey in 1980, which is calibration value for compiling and revising the spatial map of grassland of 2010 using TM data, with a spatial resolution of 20 m."

Referred publication

Gao T, Yang X, Jin Y, Ma H, Li J, et al. (2013) Spatio-Temporal Variation in Vegetation Biomass and Its Relationships with Climate Factors in the Xilingol Grasslands, Northern China. PLOS ONE 8(12): e83824.

Weipei Xu, Lin Pan, Xuebiao Hu, Qi Li, Qiuyue Zhang, Xuting Shao, Changxiu Wang, Chenchen Wang, Xiaoxiao. (2019). Spatio-temporal variations in vegetation types based on a climatic grassland classification system during the past 30 years in Inner Mongolia, China. CATENA. 185. 104298. 10.1016/j.catena.2019.104298.

 

2.In Fig. 3 R² is too small to draw so optimistic conclusion you made as “its distribution or trend is consistent with grassland type, which is mainly affected by soil organic carbon content.” You found just some rough correlation using the very very simple pedotransfer function, as you admitted.

Response: Thanks very much for your professional suggestion and constructive comments, according to which, we have rewritten the referred conclusion sentences, and the discussion on the correlations has been replaced to the discussion section. The deeper analysis of relationships between soil organic carbon and soil bulk density were not presented in this manuscript, and related issues will be addressed in the future studies. We rewrote the sentence in line 413-414 as below:

" Soil bulk density shows spatial heterogeneity, and its distribution trend is consistent with grassland type. "

 

3.I do not understand the sentence: “By comparing with the actual measurement results of other scholars, it is feasible and accurate to establish a stratified Pedotransfer function using the existing soil data, and which providing insights into the construction of a Pedotransfer function model for soil bulk density estimate at a large scale”. You did not do it, and you wrote such conclusions. Why? Make appropriate calculations, comparisons etc. Change thoroughly conclusions, please.

Response: Thanks very much for your beneficial comments. It was not expressed clearly enough in the manuscript. We rewrote the sentence in line 418-420 as below:

"According to the validation results (MPE=0.018, RMSE=0.223, relative error= 16.2 %, R²=0.5386), our Pedotransfer function can provide insights into the construction of stratified data for soil bulk density estimating at a large scale, which raised accuracy of soil organic matter calculation."

The whole conclusion section was checked and modified to avoid similar ambiguity.

 

4.NDVI is rather rough, traditional remote sensing index, however it is greatly sensitive to many parameters which are essential for the condition of grasslands, as for example soil moisture depending not only on climatic conditions, but also on many local conditions. Why you did not use other more sophisticated indices?  It would be much better if the conclusions would be more realistic but included more thoughtful directions for the future development of the pedofransfer function, e.g. much better use of satellite observations.

Response: Thanks very much for your suggestions. Compared with other satellite observations products, such as LST and NDMI, many previous researches prefer to use NDVI in Pedofransfer function (Iqbal et al., 2005; Khalili et al., 2015) because NDVI has a more correlation than other Moderate Resolution Imaging Spectroradiometer (MODIS) data products (Al-Shehhi et al., 2011; Han et al., 2010;), this issue makes NDVI is an appropriate index for soil bulk density estimation.

Referred publication

Al-Shehhi M R, Saffarini R, Farhat A, et al. Evaluating the effect of soil moisture, surface temperature, and humidity variations on MODIS-derived NDVI values. Geoscience & Remote Sensing Symposium. IEEE, 2011:3160-3163.

Han Y, Wang Y, Zhao Y. Estimating Soil Moisture Conditions of the Greater Changbai Mountains by Land Surface Temperature and NDVI. IEEE Transactions on Geoence & Remote Sensing, 2010, 48(6):2509-2515.

Iqbal J, Read J J, Thomasson A J, et al. Relationships between Soil–Landscape and Dryland Cotton Lint Yield. Soil science Society of America Journal, 2005, 69(3):872-882.

Khalili Moghadam B, Afyuni M, Jalalian A, et al. Estimation of soil saturated hydraulic conductivity in part of central zagroos using regression and ANNS method. JWSS - Isfahan University of Technology, 2015, 228-238.

 

Minor, but very important remarks – examples

Line: 230 Replace NDIV by NDVI.

Response: Thanks for your professional concern and close attention, the manuscript was checked and corrected in line 252.

 

Lines: 236-238 Non-finished sentence.

Response: Thanks for your close attention. The sentence was checked and corrected in line 239-242. We also checked throughout the whole manuscript to avoid similar flaws.

 

Lines: 238-239 There are 4 symbols in 237, but only are 3 explained.

Response: Thanks for your professional concern and close attention, we have added the symbols in line 259-260.

 

Line: 255 lack of the space before (0-50cm).

Response: Thanks for your close attention. Similar flaws in the manuscript were checked and corrected in line 280.

 

Line: 350 lack of the space before "[35]".

Response: Thanks for your close attention. Modified in line 372.

 

Line: 354 lack of the space before "[44]".

Response: Thanks for your close attention. Modified in line 376.

 

Errors in eqs (3)-(6) (index "j" is lacking on the left side)

Response: Thanks very much for your beneficial comments. We revised the notation to use Pi instead of Ti, and Mi instead of Tj. The sentences in line 258-260 were rewritten as below:

" In order to evaluate the estimation accuracy of sample points and model results, the average prediction error (MPE), root mean square error (RMSE), relative error (E) and multiple correlation coefficient (R²) were applied. In the equation (3)-(6), , ,  and   represent the measured value, predicted value, average predicted value and average value respectively, and n is the number of samples."

 

Taking above into account check again carefully the whole text and improve it.

 Response: Thanks for your close attention. We carefully checked the whole text, and corrected other similar flaws in manuscript as below:

Modified the sentence in line 233 to make it clear.

Rewrote the original sentences in line 239-242.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

Regarding the MS ijgi-963725,

it seems a very interesting topic. as correctly mentioned, bulk density is an important factor for calculating the soil organic carbon stock. therefore, estimation of BD with accuracy method could be practical. 

 

Author Response

Revision notes for “Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China”

We have revised the manuscript according to the comments/suggestions from the reviewers. The following revision notes record these changes (Underlined are comments/suggestions by the reviewers, and the text immediately followed is our response).

Response to Reviewer#2

Dear Authors,

Regarding the MS ijgi-963725,

it seems a very interesting topic. as correctly mentioned, bulk density is an important factor for calculating the soil organic carbon stock. therefore, estimation of BD with accuracy method could be practical. 

Response: We would like to thank reviewer #2 for your positive remarks. Thank you for helping us to better understand bulk density prediction and to improve the quality of our manuscript.

Reviewer 3 Report

General comments

The manuscript proposes a study on soil bulk density in northern China. The topic is of great interest to IJGI readers, but it is not clear the novelty of the study.

The authors state “a novel approach was proposed for soil bulk density measurement based on the comprehensive evaluation of multiple climatic factors and the reports on the soil bulk density of grasslands across northern China” (line 81-83), but they use the multi-factor weighted regression model (MWRM) proposed by Wang at al. 2019 [ref. 35] over a different area. They need to better highlight the novelty of their approach, which part of the method is different, or which data is treated differently.

Therefore, I suggest a strong rearrangement of the manuscript to be published in IJGI journal.

 

Specific comments

Introduction: Start from this section to clearly highlight the differences between the state of the art and the objective of the proposed novel approach.

Line 90: Section 2. Materials and Methods: Add a flowchart that shows the different steps of the method adopted and comment in detail the steps containing novelty compared to the current literature.

Figure 1: Please indicate in different colors the sample points used for the implementation of the model from those used for the validation phase.

Line 122: Section 2.2 Soil bulk density survey and soil sampling: To better individuate the characteristics of the two soil sampling used and the finality of their use, it is better to separate the two descriptions (eg., with a bullet list)

Lines 180-183: Please justify the selection of NDVI MODIS data from mid-August. Why don’t select the standard years' cycle Jan-Dec ?

Line 235: Model validation and accuracy evaluation

First of all, the equations are not entirely corrected. The sum on i=1 to n refers to the i-th sample point, therefore, both the predicted and measured values have to refer to i-th sample point. Hence, the term T_j has no meaning; there is no sum on j. It is better to change the notation (e.g., P_i instead of T_i and Mi instead of Tj with the respective mean).

Moreover, in this section, it is useful to clarify which samples are used for validation (clearly distinct from the samples used for model implementation).

Table 1 and Table 2: Please specify the depth of the shown data.

Table 3: This table shows the comparison of the apparent soil density data from different studies. The study on Yili, Xinjiang (ref 33) provides information for the layer 0-30 cm; why do you compare the layer 0-10 cm from the present study? Your results also include 10-20 and 20-30 cm. Perhaps it is better to compare the mean of your three layers.

Author Response

Revision notes for “Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China”

We have revised the manuscript according to the comments/suggestions from the reviewers. The following revision notes record these changes (Underlined are comments/suggestions by the reviewers, and the text immediately followed is our response).

 

Response to Reviewer#3

General comments

The manuscript proposes a study on soil bulk density in northern China. The topic is of great interest to IJGI readers, but it is not clear the novelty of the study.

The authors state “a novel approach was proposed for soil bulk density measurement based on the comprehensive evaluation of multiple climatic factors and the reports on the soil bulk density of grasslands across northern China” (line 81-83), but they use the multi-factor weighted regression model (MWRM) proposed by Wang at al. 2019 [ref. 35] over a different area. They need to better highlight the novelty of their approach, which part of the method is different, or which data is treated differently.

Therefore, I suggest a strong rearrangement of the manuscript to be published in IJGI journal.

Response: Thank you for your professional and important suggestions. Our original presentation did not express the novelty clearly enough. We proposed a new soil stratified Pedotransfer function with a dataset (Pilot Project of the Chinese Academy of Sciences). After that, part of another dataset (Investigation on the Degradation and Causes of Grassland Resources in Key Pastoral Areas of Temperatea Grassland in China) was applied to construct the spatial model. Besides that, the multi-factor weighted regression method (MWRM) was only used to the spatial interpolation with environmental factors and spatial the soil bulk density data. To better highlight the novelty of this approach, we reorganized the sentences in the method section in line 192-195 as below:

" We proposed a new soil stratified Pedotransfer function with a dataset (Pilot Project of the Chinese Academy of Sciences) which help construct the missing stratified data in a large scale, and used MWRM (multi-factor weighted regression model) model[35] to interpolate soil bulk density of temperate grassland of northern China."

We rearrange the manuscript based on the reviewer’s suggestion as below:

Added the sentences in line 63-69.

Deleted the sentences in line 70-72.

Deleted the sentences in line 81-83.

Added the sentences in line 83-90.

Deleted the sentences in line 95-97.

Rewrote the sentence in line 98-99.

Added Table 1 in line 137.

Rewrote the sentence in line 192-195.

Rewrote the sentence in line 209-214.

Deleted the sentences in line 415-417.

Added the sentences in line 418-420.

 

Specific comments

Introduction: Start from this section to clearly highlight the differences between the state of the art and the objective of the proposed novel approach.

Response: Thank very much for your comments. According to your suggestion, the novel SPTF that we constructed and the innovation of this study were highlighted in line 98-99 in the introduction section.

Additionally, to better highlight the novelty of this approach, we rearranged the introduction section, adding some content in line 63-69, deleting the sentences in line 70-72, and 81-83 and adding sentences in line 83-90. The new added contents was marked with yellow background. The revised introduction is as below:

"Soil, an essential component of terrestrial ecosystem, is fundamental for the vegetation survival. Its physical and chemical properties affect the growth of plants as well as restrict the productivity [1,2]. Soil bulk density, defined as the soil mass (or weight) per unit volume of undisturbed soil column[3], is an important physical property of soil that has a critical impact on soil permeability, infiltration, water-holding capacity, solute transport, and soil erosion resistance[4,5]. Hence, it quantitatively characterizes the ecological functions of soil and is one of the major indicators for evaluating the environmental soil quality [6,7]. In addition, soil bulk density is an indispensable index for estimating the soil water-holding capacity and is one of the main parameters for accurately estimating soil carbon and nitrogen storage[8,9]. Therefore, the establishment of complete systematic database pertaining to soil bulk density is practical significance for basic soil science research, ecological environment assessment, and soil quality monitoring.

The cutting ring method is the most commonly and directly used soil bulk density method[10]. However, the approach for soil bulk density measurement must be requires the collection of undisturbed soil samples[11]. This particular approach is time-consuming, labor intensive, and expensive in large-scale practical implementation [12-14]; on the other hand, sampling uncertain may cause systematic errors, that all limit the number of sample points and data quality of soil bulk density measurement. In particularly, bulk density data of large-scale projects and special region, such as watersheds and forests, is difficult to obtain. Therefore, current research is focused on the surface soil bulk density, with only few reports available on deep soil data[15]. In recent years, to overcome the problem of the lack of bulk density data, a Pedotransfer function model has been proposed using other soil properties for its estimation as an alternative, and this method proved by several scholars because of its good predictability[16-19]. As a simplified and convenient approach for soil bulk density measurement, soil bulk density prediction model and its related application are receiving increasing attention worldwide[18,20].

The SPTFs are mainly derived from two conceptions. One of which is multiple regression analysis based with on auxiliary data such as organic matter carbon, soil particle and depth [21-24], the other one is empirical model [25] Other scholars wanted to improve the estimation accuracy of multiple regression model by incorporating additional parameters such as soil morphological and physiographic properties [26]. But there are few studies about application of Pedotransfer function model on environmental and/or climatical factors, such as differences in land use, vegetation, and heterogeneity of soil types. Therefore, its accuracy and precision often rely on the study of specific regions[1,21,25]. In contrast, for the carbon accounting, a spatial coordinate-based approach is recommended, and carbon storage should be expressed by mass of soil organic carbon per unit land area to a depth of 30 cm[8,26]; however, the estimation of carbon storage in this way may be uncertain due to the evident occurrence of soil swelling or compaction because of a change in the soil bulk density[27]. It is primarily caused by the swelling or contraction of clayey soil resulting from the perennial fluctuations of the soil water content and soil depth, thereby leading to the alterations in soil bulk density[28,29].

Soil properties especial soil bulk density are vital parameters to estimate soil total carbon [24]. However, detailed mechanism for soil carbon turnover is not explained only by surface layer, but defined as deeper layers [26]. Furthermore, our ability to investigate ecosystem carbon relies on total soil carbon which partially found in surface soil layer [27]. But Jobbagy and Jackson (2000) predicted that 56% of total soil C can be found 1 m beneath the of surface. According to current studies, the impact of soil depth on soil bulk density simulation or the estimation of stratified soil bulk density has seldom been considered in the Pedotransfer function model using soil properties[25,30]. The grasslands of northern China are in the central part of East Asia. Diverse land use accompanies complex climatic conditions from west to east across the East Asian continent, which makes the construction of soil Pedotransfer function model more challenging, with considering that few reports on stratified soil bulk density in such a large scale. Accordingly, the objectives of the present study are as follows:

(i) By employing the existing stratified data (a part of soil bulk density stratification data), a new SPTF was constructed for stratified missing data of soil bulk density in a large scale.

(ii)The spatial pattern of soil bulk density stratification data was analyzed, in addition to vertical soil profile estimation based on the stratified soil bulk density data.

(iii) The relationship between soil bulk density, grassland type and organic carbon content was discussed."

 

Line 90: Section 2. Materials and Methods: Add a flowchart that shows the different steps of the method adopted and comment in detail the steps containing novelty compared to the current literature.

Response: We added the flowchart based on the reviewer’s suggestion. We placed the flowchart in the material in line 234. Furthermore, we rewrote the methodology to make it clear in line 236-248 and 254-262 respectively as below:

Line 236-248:

" According to the Figure 2, the soil bulk density dataset of 143 grassland sample sites from “Strategy Pilot Project of Chinese Academy of Sciences” was used for stratified model after K_S normality test (p < 0.01) which was performed using SPSS 20.0 software, and 117 verified sample sites were eventually obtained for statistical analyses after excluding sites with obvious anomalies. Based on the stratified data of soil bulk density in 117 verified sample sites, the variability coefficients of vertical layer for soil bulk density and soil organic carbon content in each sample site, named K_SBD and K_SOC respectively, could be obtained by using linear regression analysis (equation 1). Furthermore, 397 grassland sample sites from “Investigation on the Degradation and Causes of Grassland Resources in Key Pastoral Areas of Temperate Grassland in China” were used to constructed spatial model (equation 2). Based on the SPTF equation (1) and (2), soil bulk density in different soil layers (0–10cm, 10–20cm, 20–30cm, 30–50cm) was estimated"

Line 254-262:

"The MWRM model [35] was established using the surface soil bulk density data obtained from the grassland survey, including 6 geographical factors of elevation, annual mean temperature, annual mean rainfall, accumulated temperature (≥ 10°C), humidity, and NDVI. Here, we used MWRM        model to interpolate the spatial stratified pattern of soil bulk density (Figure 2). We got the spatial grid data (1 km × 1 km) of surface soil bulk density (0-50 cm) according to the SPTF equation (1) and (2). After that, the stratified spatial pattern of soil bulk density (0–10cm, 10–20cm, 20–30cm, 30–50cm) was estimated by MWRM model with auxiliary data and spatial grid data. "

The added flowchart is presented as below:

 

Figure 1: Please indicate in different colors the sample points used for the implementation of the model from those used for the validation phase.

Response: We revised Fig.1 based on the reviewer’s suggestion.

 

Line 122: Section 2.2 Soil bulk density survey and soil sampling: To better individuate the characteristics of the two soil sampling used and the finality of their use, it is better to separate the two descriptions (eg., with a bullet list)

Response: Thank you for your professional and important suggestions. We add the Table 1 in line 137 based on the reviewer’s suggestion to describe the individual characteristics of the datasets. We also rewrote the methodology to make it clear in line 240-243, and add related description in line 142 and 181 separately.

 

Lines 180-183: Please justify the selection of NDVI MODIS data from mid-August. Why don’t select the standard years' cycle Jan-Dec?

Response: Thanks very much for your suggestion. The growing season of grassland is in mid-August. Therefore, NDVI data from mid-August instead of whole year's cycle could better reflect the real growing condition of the grassland. Many previous studies also proved it more accurate to use the growing season data (Gonsamo et al., 2018; Zhang and Zhang, 2019).

Referred publication

Gonsamo A, Chen J M, Ooi Y W. Peak season plant activity shift towards spring is reflected by increasing carbon uptake by extratropical ecosystems. Global Change Biology, 2018, 24:2117-2128.

Zhang X, Zhang B. The responses of natural vegetation dynamics to drought during the growing season across China. Journal of Hydrology, 2019, 574:706-714.

 

Line 235: Model validation and accuracy evaluation

First of all, the equations are not entirely corrected. The sum on i=1 to n refers to the i-th sample point, therefore, both the predicted and measured values have to refer to i-th sample point. Hence, the term Tj has no meaning; there is no sum on j. It is better to change the notation (e.g., Pi instead of Ti and Mi instead of Tj with the respective mean).

Moreover, in this section, it is useful to clarify which samples are used for validation (clearly distinct from the samples used for model implementation).

Response: Thanks very much for your professional suggestion. We corrected the equations in line 258-259. Furthermore, we clarified the data source and application in Table 1 according to the reviewer’s suggestion.

 

Table 1 and Table 2: Please specify the depth of the shown data.

Response: Thanks very much for your suggestion. We added the depth of soil data in line 315 and 329 separately.

 

Table 3: This table shows the comparison of the apparent soil density data from different studies. The study on Yili, Xinjiang (ref 33) provides information for the layer 0-30 cm; why do you compare the layer 0-10 cm from the present study? Your results also include 10-20 and 20-30 cm. Perhaps it is better to compare the mean of your three layers.

Response: Thanks very much for your constructive suggestion. We added the depth information of different layers in Table 3 in line 363-364.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors,

Thank you for revision of your manuscript, and the explanations. The changes you made in the manuscript improved it. Now, I may accept the paper for publication in IJGI.
I appreciate mostly practical value of the paper.

Sincerely Yours,

Reviewer

Author Response

Revision notes for “Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China”

We have revised the manuscript according to the comments/suggestions from the reviewers. The following revision notes record these changes (Underlined are comments/suggestions by the reviewers, and the text immediately followed is our response).

Response to Reviewer#1

 General comments

Thank you for revision of your manuscript, and the explanations. The changes you made in the manuscript improved it. Now, I may accept the paper for publication in IJGI.
I appreciate mostly practical value of the paper.

Response: We would like to thank reviewer #1 for your positive remarks. Thank you for helping us to better understand bulk density prediction and to improve the quality of our manuscript.

Author Response File: Author Response.pdf

Reviewer 3 Report

Stratified Data Reconstruction and Spatial Pattern  Analyses of Soil Bulk Density in the Northern  Grasslands of China

Review: ijgi-963725-peer-review-v2

 

General comments

Thanks to the authors for their effort to improve the manuscript. The changes made have greatly improved the intelligibility of the study, in particular the clarity of the method section.

Some minor changes are required to finalize the manuscript and make it publishable in IJGI.

 

Specific comments

1. Introduction and 2.1 Study area: Please specify in these sections that the analysis refers to “natural grasslands”. The extrapolation for cultivated grassland should contain also information on land management as the use of different machineries (at least tyred vs tracked) differently affects the compaction of soils (see e.g., Imbrenda et al.,2014 doi: 10.1111/ejss.12184). Repeated passes of heavy agricultural machinery induce the formation of the plough ‘pan’ by compressing soil aggregates.

2.4.4 Model Validation: Please add information on how the layers are considered in the validation step. By taking into account the results shown in Figure 4, it seems that you aggregate all the layers (0-10, 10-20, 20-30, 30-40, and 40-50), it has to be clarified in the method section.

Minor changes

Line 65: “models” instead of “model”

Line 79: “by” instead of “as”; “defined by deeper layers”

Line 80: eliminate “But”

Lines 98-99: move “Figure 1” from the end of the sentence to after the word “China”

Line 126: Move Table 1 at the end of the section 2.2.

Line 131: Please specify which model is the subject of validation.

Table 1: Please specify “Model validation” to which model refers to. Being present a unique validation set, I understand that it refers to the comprehensive model (i.e., Stratified (SPFT) plus Spatial MWRM) models). Anyway, the authors need to detail this issue.

Line 137: if you have additional info, please add details on GPS position errors and the type of the camera used.

Figure 2: The model validation step is missing.

Verify the order of references (e.g. [33] is before [32])

Author Response

Revision notes for “Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China”

We have revised the manuscript according to the comments/suggestions from the reviewers. The following revision notes record these changes (Underlined are comments/suggestions by the reviewers, and the text immediately followed is our response).

 

Response to Reviewer#3

 General comments

Thanks to the authors for their effort to improve the manuscript. The changes made have greatly improved the intelligibility of the study, in particular the clarity of the method section.

Some minor changes are required to finalize the manuscript and make it publishable in IJGI.

Response: Thank you for your professional and careful review. You precious review work have helped us in better understanding bulk density prediction and largely improved the quality of our manuscript. Your pertinent and insightful points/suggestions are greatly appreciated. We have incorporated almost all of the suggested changes into the new version of our manuscript.

 

Specific comments

1.Introduction and 2.1 Study area: Please specify in these sections that the analysis refers to “natural grasslands”. The extrapolation for cultivated grassland should contain also information on land management as the use of different machineries (at least tyred vs tracked) differently affects the compaction of soils (see e.g., Imbrenda et al.,2014 doi: 10.1111/ejss.12184). Repeated passes of heavy agricultural machinery induce the formation of the plough ‘pan’ by compressing soil aggregates.

Response: Thank you very much for your professional comments. As reviewer commented, irrational land management influences the compaction of soils and increases incidence of soil degradation, which would influence the accurate prediction of soil bulk density. Therefore, in our research sampling and prediction only involved natural grasslands. We added relevant introduction and clarified our manuscript (Marked in red) as below:

Modification in 1. introduction section in Line 80-83

“Furthermore, soil compaction was also induced by intense human activities, which affected the soil bulk density and organic carbon storage [32]. In case of that, most research focused on natural grasslands because the prediction accuracy would be heavily affected if containing the extrapolation for cultivated grassland.”
Modification in 2.1 study area section in Line 139-140

“In this study, the sample points and model prediction only refer to natural grasslands; non-grassland and cultivated grasslands were masked by boundaries.”

Reference :

Imbrenda V, D'Emilio M, Lanfredi M, et al. Indicators for the estimation of vulnerability to land degradation derived from soil compaction and vegetation cover. European Journal of Soil science, 2014, 65(6):907–923.

 

2.4.4 Model Validation: Please add information on how the layers are considered in the validation step. By taking into account the results shown in Figure 4, it seems that you aggregate all the layers (0-10, 10-20, 20-30, 30-40, and 40-50), it has to be clarified in the method section.

Response: Thanks very much for your professional suggestion and constructive comments, according to which, we added the relevant information (Marked in red below) in method section (2.4.4 Model validation and accuracy evaluation) in Line 278-279 as below:

“A total of 190 sample points were used to model validation, to compare with average value of different layers of soil bulk density (0–10 cm, 10–20 cm, 20–30 cm, 30–50 cm).”

 

Minor changes

Line 65: “models” instead of “model”

Response: Thanks for your close attention. Modified (Line 67 in new version).

Line 79: “by” instead of “as”; “defined by deeper layers”

Response: Thanks for your close attention. Modified (Line 88 in new version).

Line 80: eliminate “But”

Response: Thanks for your close attention. Modified (Line 90 in new version).

Lines 98-99: move “Figure 1” from the end of the sentence to after the word “China”

Response: Thanks for your professional concern and close attention. “Figure 1” was moved from the end of the sentence to after the word “China” in Line 112.

Line 126: Move Table 1 at the end of the section 2.2.

Response: Thanks for your professional suggestion. Table 1 was moved to the end of the section 2.2 in Line 200.

Line 131: Please specify which model is the subject of validation.

Response: Thanks for your professional suggestion. We furthur specified the subject of validation in Line 145-148 as below:

“Part of the soil bulk density data in dataset was applied in SPTF model validation. Axillary data (environment data, remote sensing data and vegetation data) was applied in spatial interpolation.”

 

Table 1: Please specify “Model validation” to which model refers to. Being present a unique validation set, I understand that it refers to the comprehensive model (i.e., Stratified (SPFT) plus Spatial MWRM) models). Anyway, the authors need to detail this issue.

Response: Thanks for your professional concern and close attention. Table 1 was corrected in Line 200-201 as below:

 Table 1. Different datasets source and application.

Dataset	Investigation on the Degradation and Causes of Grassland Resources in Key Pastoral Areas of Temperate Grassland in China	Pilot Project of the Chinese Academy of Sciences		Other datasets
Sample Numbers	587		143	587
Application	For spatial model construction and validation, spatial interpolation		For stratified model construction	For spatial interpolation
Model construction	397		117	397
Model validation	190 (Applied in SPTF model validation)		None	190 (Applied in SPTF model validation)

 

Line 137: if you have additional info, please add details on GPS position errors and the type of the camera used.

Response: Thanks for your professional concern and close attention. The GPS (American MAGELLAN, eXplorist 610, position errors less than 15 m) and camera (Canon, EOS 2000D) information was added in Line 156-157.

 

Figure 2: The model validation step is missing.

Response: Thanks for your professional concern and close attention. Figure 2 was revised in Line 243 as below:

Verify the order of references (e.g. [33] is before [32])

Response: Thanks for your close attention. We checked and modified the sequence of references.

Author Response File: Author Response.pdf