Next Article in Journal
High-Resolution Imagery Classification Based on Different Levels of Information
Previous Article in Journal
Accurate Building Extraction from Fused DSM and UAV Images Using a Chain Fully Convolutional Neural Network
 
 
Article
Peer-Review Record

Quick Aboveground Carbon Stock Estimation of Densely Planted Shrubs by Using Point Cloud Derived from Unmanned Aerial Vehicle

Remote Sens. 2019, 11(24), 2914; https://doi.org/10.3390/rs11242914
by Xueyan Zhang 1,2
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Remote Sens. 2019, 11(24), 2914; https://doi.org/10.3390/rs11242914
Submission received: 9 October 2019 / Revised: 28 November 2019 / Accepted: 4 December 2019 / Published: 5 December 2019

Round 1

Reviewer 1 Report

Well done!! it is a good piece of research work demonstrating image analysis techniques to extract canopy features and apply if for calculating carbon sink in shrubs. I like the way you have accepted the limitations in the discussion. That way you admit it needs further testing. Good stuff. 

Please go through my comments that may improve the quality of your article.

line 92: wouldn't be better if can include North direction in the map (figure 1 and 2)?

line 119: suggestion>> W is the aboveground biomass (rather just telling biomass)

line 124: how did you determined carbon content is 0.5? is it from literature? then cite the reff.

line 130: is it Figure 2? or should be a different Figure for the UAV?

line 142: What was the orthomosaic procedure conducted to stitch the image tiles to a single image? 

Line 150: in this section please give further details how did you seperated the crown area or crown feature from the background? table 1 gives results to each shrub, how did your RGB analysis identified those each shrubs? I see you have described how obtained the missing DTM points by interpolation, is crown area manually selected at that step? if so please specify it clearly.

Line 161-169: Can you support the procedure of using kriging interpolation with similar past work to estimate the missing data points? As a better method to obtain DTM points sheltered by the shrub canopy.

Line 170-172: How can you justify the exact spatial resolution of the image pixels? How much meter or centimeter is covered by a single pixel? Don't you need to do a small check by measuring a length of a known object in the image? or if you have another justification for that please include.

Line 214: Please include the Pearson correlation and RMSE in each scatter plot.

 

--end---

 

Author Response

Response to reviewer #1

 

Point1: Well done!! it is a good piece of research work demonstrating image analysis techniques to extract canopy features and apply if for calculating carbon sink in shrubs. I like the way you have accepted the limitations in the discussion. That way you admit it needs further testing. Good stuff. 

Please go through my comments that may improve the quality of your article.

Response1: Thank you for your positive comment.

 

Point2: line 92: wouldn't be better if can include North direction in the map (figure 1 and 2)?

Response2: As suggested, I have added the North direction in Figures 1 and 2.

 

Point3:line 119: suggestion>> W is the aboveground biomass (rather just telling biomass)

Response3: As suggested, the word “aboveground“ has been added.

 

Point4: line 124: how did you determined carbon content is 0.5? is it from literature? then cite the reff.

Response4: Yes, this value was obtained from the following reference, which has been added in the revised manuscript:

Wei W.; Zhao, J.; Qing H.; Zeng J. Study on the measurement method of carbon sink in Kubuqi Desert shrub—taking Caragana korshinskii and salix as example. Journal of Chifeng University. 2016, 32, 118-121.

 

Point5: line 130: is it Figure 2? or should be a different Figure for the UAV?

Response5: Here, I mean to imply Figure 3, which for the UAV is added.

 

Point 6: line 142: What was the orthomosaic procedure conducted to stitch the image tiles to a single image? 

Response6: The orthomosaic procedure was conducted using Agrisoft Photoscan (Agisoft LLC, Russian). The sentence beginning with “the resulting image size was 4000 ´ 3000 pixels” has been moved to Line 155 in the “Shrub volume accounting using photogrammetry” section.

 

Point 7: Line 150: in this section please give further details how did you seperated the crown area or crown feature from the background? table 1 gives results to each shrub, how did your RGB analysis identified those each shrubs? I see you have described how obtained the missing DTM points by interpolation, is crown area manually selected at that step? if so please specify it clearly.

Response 7: The crown areas were separated or shrub belt boundary was extracted by visual interpretation (Line 163). In addition, the average crown and height of each shrub was identified through a field survey (Lines 100–102). The following descriptions have been added to the revised manuscript:

Line 163: “The shrub belt boundary and crown area were extracted using manual visual interpretation in each plot”

Line 203: “The crown and height of 209 shrubs were measured through field survey in three plots.”.

Point 8: Line 161-169: Can you support the procedure of using kriging interpolation with similar past work to estimate the missing data points? As a better method to obtain DTM points sheltered by the shrub canopy.

Response 8: I believe that the kriging interpolation is a better method to obtain DTM sheltered by the shrub canopy. It is acceptable in such a small sample area.

 

Point 9: Line 170-172: How can you justify the exact spatial resolution of the image pixels? How much meter or centimeter is covered by a single pixel? Don't you need to do a small check by measuring a length of a known object in the image? or if you have another justification for that please include.

Response 9: To be consistent with the spatial resolution of DTM, DSM and CHM were resampled into 0.1 m ´ 0.1 m by using ArcGIS.

 

Point 10: Line 214: Please include the Pearson correlation and RMSE in each scatter plot.

Response 10: I have revised the scatter plots.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Xueyan Zhang,

Thank you very much for your very interesting article. I read it with great interest. My expertise is definitely stronger in the UAV remote sensing than in the carbon accounting topic and therefore I will focus my review mainly on these parts of your work.

Major comments:

I have an overall good impression of the draft even if the number of sample plots and their area is quite small and you have left out various very relevant pieces of research, which are highly relevant for your work. The discussion as well as the method description need some mayor improvements in detail and depth. Please see detailed comments below.

L 58-64: This paragraph mixes in its comparison sensors and platforms. There are LiDAR UAVs out there therefore you have to compare SfM with LiDAR. See for example Wallace et al 2016 DOI: 10.3390/f7030062

L 136 – 143:  The camera focal length and acquisition angle is missing (probably nadir), as well as a more decent description of the ‘GPS equipment’ (with or without RTK, which precision). Please also justify your choice for the acquisition scenario a little better please see e.g. Torres-Sanchez et al. 2017 DOI: 10.1007/s11119-017-9502-0, or Frey et al. 2018 DOI: 10.3390/rs10060912 This should be also part in your discussion.

L 282: Please discuss what impacts the small sample size has on your results, which parts are reliable and which parts need further research to become reliable. Please note more possible influencing factors like acquisition season, drought effects, etc. which could bias your calibration.

 

Minor comments:

L 37: The abbreviation tCO2e has not been introduced beforehand

L 92: The quality of the figure is week, brackets are touching outlines, the coordinates are not readable and there is quite a grey moiré. Please enhance it.

L 100: I am not a native speaker but I couldn’t find splines a s a Word for planted rows. Please check if this is the correct term.

 L 102: The readability of the figure is weak especially the black text on the dark green ground

L 138: I have not heard about the term ‘heading overlap’. I think ‘forward overlap’ is more common

L 157 / L 159: I have not controlled all of your citations but [30] and [31] are definitely incorrect numbered. For the determination of the DTM from UAV-SfM data I would recommend to check Wallace 2019 DOI: 10.3390/f10030284 and Zhang 2016 DOI: 10.3390/rs8060501

L 214: I expect to be able to differentiate at least some points in a scatterplot. Se comments on graphics quality above. Numbers should not touch outlines and should be in a readable size. The grey backgrounds are irritating

L 220: avoid over precise numbers

 L 273: [39] is missing in your references

L 395. The Reference list is incomplete

Author Response

Response to reviewer #2

 

Point1: Thank you very much for your very interesting article. I read it with great interest. My expertise is definitely stronger in the UAV remote sensing than in the carbon accounting topic and therefore I will focus my review mainly on these parts of your work.

Response1: Thank you for your positive comment.

Point2: I have an overall good impression of the draft even if the number of sample plots and their area is quite small and you have left out various very relevant pieces of research, which are highly relevant for your work. The discussion as well as the method description need some mayor improvements in detail and depth. Please see detailed comments below.

L 58-64: This paragraph mixes in its comparison sensors and platforms. There are LiDAR UAVs out there therefore you have to compare SfM with LiDAR. See for example Wallace et al 2016 DOI: 10.3390/f7030062

Response2: Thank you for your constructive suggestion. The comparison between the sensor and platforms was conducted separately. The following text about SFM was added in the revised manuscript:

“There are two remote sensing techniques that are suitable for application on a UAV platform: airborne laser scanning (ALS) and structure from motion (SfM). The SfM photogrammetric technique underperforms in terms of accuracy, whereas ALS is capable of providing more accurate estimates of the vertical structure of plants. However, the SfM is more accessible than ALS, which is too expensive for shrub owners in developing countries. Therefore, the use of UAVs with SfM technique allows for the detection of surface data at acceptable spatial and temporal resolutions, and it is a more cost-effective solution.”

Point3: L 136 – 143:  The camera focal length and acquisition angle is missing (probably nadir), as well as a more decent description of the ‘GPS equipment’ (with or without RTK, which precision). Please also justify your choice for the acquisition scenario a little better please see e.g. Torres-Sanchez et al. 2017 DOI: 10.1007/s11119-017-9502-0, or Frey et al. 2018 DOI: 10.3390/rs10060912 This should be also part in your discussion.

Response3: I agree that some key parameters were missing and have added the same. The camera focal length was 35 mm, and the effective pixel in 1-inch CMOS was 20 million. The camera was set to F-stop of 2.8, shutter speed of 1/2000 s, and ISO-values between 100 and 1600 that were automatically adapted to the given lighting situations. The images were obtained from a height of 60 m with 70% side overlap, 90% forward overlap, and an optical GSD of 3 cm. During the flight, the UAV automatically triggered the camera every 13 m, simultaneously recording its position using an internal GPS/GLONASS dual-mode satellite positioning system. Fifty-eight ground control points were marked and measured using real time kinematic (RTK) equipment in field plots. The safe mode and camera angel of −90° were preferred during the UAV image acquisition.

All this information is added in the data-acquisition section.

Point4: L 282: Please discuss what impacts the small sample size has on your results, which parts are reliable and which parts need further research to become reliable. Please note more possible influencing factors like acquisition season, drought effects, etc. which could bias your calibration.

 Response4: 4 I agree that more factors that would influence the survey and model estimated should be discussed. The added discussions are as follows:

“The number of plots was relatively small, which limited the generality of the proposed estimation model. The model was reliable when applied locally but it lacked generalization ability. Furthermore, in August 2018, the study area received abundant rainfall and luxuriant shrub growth. In case of drought years or seasons, the number of shrub leaves is small, and this may lead to large changes in volume estimation and have an important impact on the model parameters. Therefore, more plots in different locations and times should be developed in subsequent experiments, after which the model results could be more reliable.”

Point5: Minor comments:

L 37: The abbreviation tCO2e has not been introduced beforehand

Response5: The abbreviation of tCO2e has been introduced as required :ton CO2 equivalent.

Point6: L 92: The quality of the figure is week, brackets are touching outlines, the coordinates are not readable and there is quite a grey moiré. Please enhance it.

Response6: Figure 1 has been revised.

Point7: L 100: I am not a native speaker but I couldn’t find splines a s a Word for planted rows. Please check if this is the correct term.

Response7: The word “spline” has been replaced with “belt”.

Point8:  L 102: The readability of the figure is weak especially the black text on the dark green ground

Response8: Figure 2 has been revised for readability.

Point9: L 138: I have not heard about the term ‘heading overlap’. I think ‘forward overlap’ is more common

Response9: I agree that the term should be “forward overlap,” and have revised it accordingly.

Point10: L 157 / L 159: I have not controlled all of your citations but [30] and [31] are definitely incorrect numbered. For the determination of the DTM from UAV-SfM data I would recommend to check Wallace 2019 DOI: 10.3390/f10030284 and Zhang 2016 DOI: 10.3390/rs8060501

Response10: I apologize for this oversight and have revised it accordingly.

Point11: L 214: I expect to be able to differentiate at least some points in a scatterplot. Se comments on graphics quality above. Numbers should not touch outlines and should be in a readable size. The grey backgrounds are irritating

Response11: I have tried to reduce the size of points to improve resolution of scatterplot.

Point12: L 220: avoid over precise numbers

Response12: These have been revised.

Point13:  L 273: [39] is missing in your references

Response13: The reference has been checked.

Point14:L 395. The Reference list is incomplete

Response14: The reference has been checked and revised.

 

Author Response File: Author Response.docx

Reviewer 3 Report

General Comments

 

Dear author,

The overall idea, carbon sink estimation with UAV data, is very good and can potentially find a good practical use.

However, my main concern is that you are not estimating/investigating carbon sink. In order to be able to discuss about carbon sink, you must analyse a time series of data, which will then give a picture of the carbon cycle. By analysing the carbon cycle, you can then infer when an ecosystem is acting as either sink or source of carbon.

Because your work has collected data during ONE date only, you cannot say you are estimating carbon sink. You may claim that you are estimating current above ground carbon content.

Therefore, I would suggest you correct this aspect throughout your manuscript.

Yu can still discuss that your methodology has the potential for carbon sink estimations by acquiring time series of UAV data, for example.

 

 

Specific Comments

L13-14 : “ However, the carbon sink of individual densely planted shrubs cannot be accounted for using UAVs.”

So, why are you using UAVs? You said that you CANNOT above.

 

L19 “model of the aboveground carbon sink of the densely planted shrub belt”

Provide a measure of accuracy. E.g. R2, RMSE

 

L71 “Therefore, we”

There is only ONE author.

 

Equation 2: “?????”

What about the annual variation of carbon in these shrub species? Depending on the time of the year, the plants could well be acting as carbon source. This is well known in temperate forests, for ex. in the work of “Mizunuma, T., Wilkinson, M., L. Eaton, E., Mencuccini, M., IL Morison, J., & Grace, J. (2013). The relationship between carbon dioxide uptake and canopy colour from two camera systems in a deciduous forest in southern E ngland. Functional Ecology27(1), 196-207.”

 

Figure 4: Add colour legend to a) and add graphical scale in both a) and b).

 

Table 1: Carbon sink> kg CO2 per ha? (Same for Table 2)

 

Figure 6: Please add a legend to the colours.

 

 

Author Response

Response to reviewer #3

 

Point1: The overall idea, carbon sink estimation with UAV data, is very good and can potentially find a good practical use.

Response1: Thank you for your positive comment.

Point2: However, my main concern is that you are not estimating/investigating carbon sink. In order to be able to discuss about carbon sink, you must analyse a time series of data, which will then give a picture of the carbon cycle. By analysing the carbon cycle, you can then infer when an ecosystem is acting as either sink or source of carbon.

Because your work has collected data during ONE date only, you cannot say you are estimating carbon sink. You may claim that you are estimating current above ground carbon content.

Therefore, I would suggest you correct this aspect throughout your manuscript.

Yu can still discuss that your methodology has the potential for carbon sink estimations by acquiring time series of UAV data, for example.

 Response2: Thank you for your suggestion. The carbon sink calculation requires a series of UAV data to calculate the difference between carbon storages of a baseline scenario and a scenario showing increase in carbon storage. Actually, the shrub in study area was planted using seeds. The aboveground carbon stock is theoretically equal to aboveground carbon sink. To avoid confusion to readers, I have replaced the sink with stock.

Point3: Specific Comments

L13-14 : “ However, the carbon sink of individual densely planted shrubs cannot be accounted for using UAVs.”

So, why are you using UAVs? You said that you CANNOT above.

 Response3: I have revised this expression as follows:

“However, the UAV-based estimation of the aboveground carbon stock of densely planted shrubs still faces certain challenges.”

Point4: L19 “model of the aboveground carbon sink of the densely planted shrub belt”

Provide a measure of accuracy. E.g. R2, RMSE

 Response4: the measure of accuracy was added.

Point5: L71 “Therefore, we”

There is only ONE author.

 Response5: I have corrected this clerical error

Point 6: Equation 2: “?????”

What about the annual variation of carbon in these shrub species? Depending on the time of the year, the plants could well be acting as carbon source. This is well known in temperate forests, for ex. in the work of “Mizunuma, T., Wilkinson, M., L. Eaton, E., Mencuccini, M., IL Morison, J., & Grace, J. (2013). The relationship between carbon dioxide uptake and canopy colour from two camera systems in a deciduous forest in southern E ngland. Functional Ecology27(1), 196-207.”

 Response6: I have revised ????? to ??tock, which represents aboveground carbon storage of shrubs; the model developed in this study focused on carbon stock rather than carbon sink. Therefore, the carbon stock of C. intermedia increases every year until it reaches a mature forest.

Point 7: Figure 4: Add colour legend to a) and add graphical scale in both a) and b).

 Response7: The color legend and graphical scale are added.

Point8: Table 1: Carbon sink> kg CO2 per ha? (Same for Table 2)

 Response8: The unit was revised to kg CO2e , which represents the carbon storage in one belt.

Point9: Figure 6: Please add a legend to the colours.

Response9: Legend explaining the colors has been added.

Author Response File: Author Response.docx

Reviewer 4 Report

Review - remotesensing-624963

Quick carbon sink estimation of densely planted shrubs using RGB images and an unmanned aerial vehicle

 

Really interesting read surrounding an important and often overlooked part of Carbon accounting and targets. Unfortunately, there is a major issue with the training and validation data sample size which needs to be addressed.

 

I have a mix of general and specific comments below.

 

General text comments:

-the manuscript throughout is highly repetitive

-inconsistent with unit measurement abbreviations

-issues with the majority of figures in manuscript

 

Title and Abstract:

-this manuscript is about aboveground carbon sink estimation, this should be clearer in the title – i.e. Quick aboveground carbon sink estimation of densely planted shrubs using RGB images and an unmanned aerial vehicle

-also the methods are more dependent on the point cloud derived from UAV collected data not the RGB images

-the abstract should include results and introduce the two hypotheses

 

Introduction:

-sentence structure is poor throughout the introduction, requires further work

-line: 47 and 48: missing full-stop

-line 53: suggest removal of ‘On the other hand’

-lines 61-63: false information on satellite remote sensing. Sensors on board satellites can have a cm resolution (WorldView-3, Geo-Eye-1, etc.) and can be used in cloudy conditions (RADAR). Instead you could compare with a commonly used optical sensor satellite, for example the Landsat mission.

-lines 67-69: would ‘conventional methods’ use a UAV? You describe them later as a manual field-based method (i.e. hand measured in the field).

-lines 71-75: these hypotheses seem a bit redundant, you only call back to the 1st in lines 238-239 in the results section. You do not address the hypotheses in the discussion or conclusion section.

 

Methods:

-Figure 1: in general a very low quality Figure - letters not lined up correctly, no north arrow, no scale bar, figure coordinates unreadable, no explanation of inset in 1.a, grid lines should be removed from 1.b, and add labels to identify field plots (G1, G2, G3).

-Figure 2: bottom of plot names have been cut off

-lines 120-121: highly repetitive, remove.

-lines 123-124: requires reference and explanation of method similar to biomass allometric equation

-line 126: use abbreviation – UAV

-Figure 3: reposition letters

-line 152: how did you derive the point cloud? Requires further explanation.

-lines 154-69: please restructure – very repetitive, needs to be clearer and more direct

-Figure 4: frame outline has been cut-off bottom and right-hand side

-line 185: make clear you are using the training data here (i.e. 6 splines)

-line 189: major issue – sample size too small, especially for validation data. Potential option for model validation is to adopt a leave-one-out-cross validation (LOOCV) approach.

 

Results:

-lines 200-201: it is m not cm in equation 1. keep it consistent in the text

-Table 1: Add biomass column and change average crown and height units to m

-line 206: use abbreviation – UAV

-Figure 5: Very low resolution plots, scales unreadable in some cases

-line 220: Table 2. Caption – use abbreviation – DSM

-Figure 6: reposition letters

-line 229: Table 2. Caption – use abbreviation – UAV. Make clear in the caption that G1 and G2 are training data and G3 validation data.

-Figure 7: reposition letters

 

Discussion:

-lines 249-265: highly repetitive of previous sections, adds very little

-line 282: no justification of small sample size, needs further explanation and references for ‘good statistical characteristics’ to back up point

 

Conclusion:

-need to refer back to hypotheses

 

Thanks for the opportunity to read this interesting manuscript.

I hope my comments are constructive and help strengthen the research. 

Comments for author File: Comments.docx

Author Response

Response to reviewer #4

 

Point1: Really interesting read surrounding an important and often overlooked part of Carbon accounting and targets. Unfortunately, there is a major issue with the training and validation data sample size which needs to be addressed.

Response1: Thank you for your positive comment. I have revised the manuscript according to your comments.

 Point2: I have a mix of general and specific comments below.

General text comments:

-the manuscript throughout is highly repetitive

-inconsistent with unit measurement abbreviations

-issues with the majority of figures in manuscript

Title and Abstract:

-this manuscript is about aboveground carbon sink estimation, this should be clearer in the title – i.e. Quick aboveground carbon sink estimation of densely planted shrubs using RGB images and an unmanned aerial vehicle

-also the methods are more dependent on the point cloud derived from UAV collected data not the RGB images

-the abstract should include results and introduce the two hypotheses

 Response2:

-The title of paper has been revised to the following. “Quick Aboveground Carbon Storage Estimation of Densely Planted Shrubs By Using Point Cloud derived from Unmanned Aerial Vehicle”.

-Results and two hypotheses (revised) are added into the abstract section.

“The specific objectives of this research are as follows: 1) to test the statistical relationship between aboveground carbon stock and volume of a densely planted shrub belt, and 2) to develop a model to estimate aboveground carbon stock by monitoring the volume of the densely planted shrub belt using a UAV. The results showed that â…°) the aboveground carbon stock would increase with the increase in the volume of the shrub belt, â…±) an estimation model of the aboveground carbon stock of the densely planted shrub belt was developed (), and iii) the validation assessment to estimate aboveground carbon stock by using the UAV-based estimation model produced a coefficient of determination of R2 = 0.74 and an overall root mean square error of 18.79-kg CO2e..”

 

Point3:

Introduction:

-sentence structure is poor throughout the introduction, requires further work

-line: 47 and 48: missing full-stop

-line 53: suggest removal of ‘On the other hand’

-lines 61-63: false information on satellite remote sensing. Sensors on board satellites can have a cm resolution (WorldView-3, Geo-Eye-1, etc.) and can be used in cloudy conditions (RADAR). Instead you could compare with a commonly used optical sensor satellite, for example the Landsat mission.

-lines 67-69: would ‘conventional methods’ use a UAV? You describe them later as a manual field-based method (i.e. hand measured in the field).

-lines 71-75: these hypotheses seem a bit redundant, you only call back to the 1st in lines 238-239 in the results section. You do not address the hypotheses in the discussion or conclusion section.

 Response3:

- I have improved the sentence structure of the introduction section.

-lines 47 and 48: periods have been added.

-line 53: “On the other hand” has been removed.

-lines 61–63: The comparison between UAV and satellite has been revised as follow:

“A UAV can acquire real-time high-resolution images as well as offset the fixed acquisition cycle of optical sensor satellite remote sensing and sensitivity to weather conditions (e.g., cloudiness)”

-lines 67–69: the conventional method includes manual field measurements. The “use a UAV” has been deleted.

-line 71–75, I agree with your comment. The word “hypotheses” has been replaced by “the specific objectives of this research”. Some echoes have been added in the discussion and conclusion sections.

The details are as follow:

“The specific objectives of this research are as follows: 1) to test the statistical relationship between aboveground carbon stock and the volume of the densely planted shrub belt and 2) to develop an estimation model of aboveground carbon stock by monitoring the volume of the densely planted shrub belt using a UAV.”

 

Point4:

Methods:

-Figure 1: in general a very low quality Figure - letters not lined up correctly, no north arrow, no scale bar, figure coordinates unreadable, no explanation of inset in 1.a, grid lines should be removed from 1.b, and add labels to identify field plots (G1, G2, G3).

-Figure 2: bottom of plot names have been cut off

-lines 120-121: highly repetitive, remove.

-lines 123-124: requires reference and explanation of method similar to biomass allometric equation

-line 126: use abbreviation – UAV

-Figure 3: reposition letters

-line 152: how did you derive the point cloud? Requires further explanation.

-lines 154-69: please restructure – very repetitive, needs to be clearer and more direct

-Figure 4: frame outline has been cut-off bottom and right-hand side

-line 185: make clear you are using the training data here (i.e. 6 splines)

-line 189: major issue – sample size too small, especially for validation data. Potential option for model validation is to adopt a leave-one-out-cross validation (LOOCV) approach.

 Response4:

Figure 1: The issues of letters, north arrow, scale bar, figure coordinates, grid lines, and labels were revised.

Figure 2: the plot names have been corrected.

-lines 120–121: The following sentences have been deleted. “According to Eq. (1), we could calculate the biomass of the plants, which was used as the estimated biomass based on field data.”

 

-lines 123–124: the reference and explanation have been added.

-line 126: it has been replaced by UAV.

-Figure 3: the letters a) and b) have been relocated.

-line 152: The following explanation for 3D-cloud-point establishment has been added.

“The 3D point cloud was generated using Agrisoft Photoscan Professional Edition software (Agisoft LLC, Russia) [31]. First, the camera position for each image and common points in the images were located and matched; this facilitated the fitting of calibration parameters of the camera. Second, the point cloud was built based on the estimated camera positions and images themselves [32]. A digital surface model (DSM), digital terrain model (DTM), and an orthomosaic were also generated [16]. The 3D point cloud DSM and orthophoto are shown in Figure 5a.”

-lines 154–169: the whole paragraph has been revised as follows:

“To estimate the volume of the densely planted shrubs, each pixel of the canopy height model (CHM)was treated as a quadratic prism, where the volume equals the base area (10 cm ´ 10 cm) multiplied by the height (canopy height). In theory, a CHM could be obtained by subtracting the DTM from DSM, as recommended by González-Jaramillo et al. [30]. However, it is difficult to extract a DTM from the UAV-derived DSM because the camera could not survey the terrain surface through the canopy [34]. First, to obtain a complete DTM, the shrub-belt boundary area was extracted using manual visual interpretation in each plot (Figure 5b). Second, kriging interpolation was used to estimate the missing surface of the DTM sheltered by the shrub canopy by using ArcGIS (ESRI, US) with a pixel size of 10 cm ´ 10 cm. The control points of the kriging interpolation were selected at the 1-m buffer outside the boundary of the shrub belt (Figure 5b). To avoid boundary effects of the interpolation surface, the outer border of the interpolation surface was considered as the 5-m buffer of the shrub plot (Figure 5b). It is vital to ensure the fitting accuracy between the kriging surface and DSM. Third, the points outside the 1-m shrub buffer in the plot were selected to verify the fitting accuracy of the kriging surface and DTM. Finally, the CHM was extracted by the kriging surface from the DSM, displaying the difference between the kriging surface and DSM.”

 

-Figure 4: the frame outline has been corrected.

-line 185: The volume and field-surveyed carbon stock of shrub belts in plots G1 and G2 were treated as training data.

-line 189: I agree that the too small sample size is a major limitation in this study, which would influence the survey and model estimated. The LOOCV is useful tools for prediction accuracy assessment. The methods of LOOCV added as follow:

In methods section: “In order to avoid the model instability caused by small samples, leave-one-out cross-validation (LOOCV) was used to evaluate model performance. One data was used for testing (testing data), and the aboveground carbon stocks model was fit on the remaining data (training group) the fitted aboveground carbon stocks was then used to predict the testing data and repeated until predictions for all data were generated. Accuracy and prediction ability of the model was assessed using root-squared error of LOOCV (), mean-squared error (MSE), and based R2 ().

                       (7)

                                                      (8)

                                                    (9)

Where is the number of data; denotes the mean-squared error of data; is the date of observation, is the data of prediction, and is the mean of observation. The is the measure of fit to the 1:1 line. ”

In result section: “The LOOCV results for the aboveground carbon stock model performed well. The is 17.26 kg CO2e and is 0.74, indicating that the model performance was good for making aboveground carbon stock prediction.”

   In discussion section: “The number of plots was relatively small, which limited the generality of the proposed estimation model. Although prediction ability of the model was determined using LOOCV. The model was reliable when applied locally but it lacked generalization ability. Furthermore, in August 2018, the study area received abundant rainfall and luxuriant shrub growth. In case of drought years or seasons, the number of shrub leaves is small, and this may lead to large changes in volume estimation and have an important impact on the model parameters. Therefore, more plots in different locations and times should be developed in subsequent experiments, after which the model results could be more reliable.”

 

Point5:

Results:

-lines 200-201: it is m not cm in equation 1. keep it consistent in the text

-Table 1: Add biomass column and change average crown and height units to m

-line 206: use abbreviation – UAV

-Figure 5: Very low resolution plots, scales unreadable in some cases

-line 220: Table 2. Caption – use abbreviation – DSM

-Figure 6: reposition letters

-line 229: Table 2. Caption – use abbreviation – UAV. Make clear in the caption that G1 and G2 are training data and G3 validation data.

-Figure 7: reposition letters

 Response5:

-lines 200-201: I have revised “cm” to “m”.

-Table 1: The biomass column and change average crown and height units have been corrected.

-line 206: As suggested, the abbreviation was used.

-Figure 5 has been revised.

-line 220: The abbreviation od DSM was used.

-Figure 6: The letters have been relocated.

-line 229: it has been replaced by UAV. I added new columns of training and validation data in the table.

-Figure7: the letters have been relocated.

Point6:

Discussion:

-lines 249-265: highly repetitive of previous sections, adds very little

-line 282: no justification of small sample size, needs further explanation and references for ‘good statistical characteristics’ to back up point

 Response6:

-lines 249-265: I have revised the two paragraphs.

“To date, shrub owners have not found the carbon trade particularly attractive, and the CCER has not encouraged much new shrub. One important reason is that shrub projects of potential commercial value commonly require at least 10 000 ha with no less than 200 sampling plots (20m ´ 20 m) [36], and every 5 years will repeat the survey during the 20-years project. Therefore, shrub owners experience significant challenge in completing the sampling work. In this study, a quick aboveground carbon stock estimation method was proposed for densely planted shrubs using 3D point cloud and a UAV. The core philosophy was to estimate the aboveground carbon stock according to the volume of the densely planted shrubs. Therefore, the quick aboveground carbon-stock-estimation method using UAVs is a cheap and efficient method for introducing densely planted shrubs to the carbon market in China.

Accurate information about aboveground biomass are critical parameters for carbon stock accounting. Individual UAV-based tree detection for estimating aboveground biomass has been widely applied, where the height and crown are measured to derive the allometric equation [33]. Figure 2 shows that it is difficult to distinguish individual densely planted shrubs from the shrub belts, let alone to measure the height and crown of a single shrub. Therefore, in this study, a method was developed to estimate the aboveground carbon stock by measuring the shrub belt volume by using a UAV. According to the results shown in Figure 8a, a significant correlation exists between the aboveground carbon stock and shrub belt volume. Thus, it could be concluded that the shrub belt volume, which is easy to measure using UAVs, is suitable for estimating the aboveground carbon stock of densely planted shrubs.”

 

-line 282: I have added the model assessment using LOOCV. The discussion about the issue of small sample size has been revised as follow.

“The number of plots was relatively small, which limited the generality of the proposed estimation model. Although prediction ability of the model was determined using LOOCV. The model was reliable when applied locally but it lacked generalization ability. Furthermore, in August 2018, the study area received abundant rainfall and luxuriant shrub growth. In case of drought years or seasons, the number of shrub leaves is small, and this may lead to large changes in volume estimation and have an important impact on the model parameters. Therefore, more plots in different locations and times should be developed in subsequent experiments, after which the model results could be more reliable.”

 

Point7:

Conclusion:

-need to refer back to hypotheses

 Response7:

I have replaced the term “hypotheses” with “specific objectives” in the introduction section, and summarized these objectives in the conclusion section.

Point8:

Thanks for the opportunity to read this interesting manuscript.

I hope my comments are constructive and help strengthen the research. 

Response8: Thank you for your constructive and helpful suggestions.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Dear Xueyan Zhang,

thank you for your extensive revisions. The article has clearly enhanced according to clarity, reproducebility and the discussion now includes some of the mayor weeknesses of the design which have to be discussed. I'm still not compleatly convinced about the quality of the figures. The maps in figure 1 have still poor quality with visible pixels, hardly readale text and compression artefacts and I would suggest to enhance this agian. But since this is not really a comment on the content, I leave this up to the editor, if he/she accepts this in the current form. Congratulations to this nice article.

Best wishes

Author Response

Point1: Thank you for your extensive revisions. The article has clearly enhanced according to clarity, reproducebility and the discussion now includes some of the mayor weeknesses of the design which have to be discussed. I'm still not compleatly convinced about the quality of the figures. The maps in figure 1 have still poor quality with visible pixels, hardly readale text and compression artefacts and I would suggest to enhance this agian. But since this is not really a comment on the content, I leave this up to the editor, if he/she accepts this in the current form. Congratulations to this nice article.

Response1: Thanks for your suggestions. The visible pixels of figure 1 has been further improved to 600 dpi.

Author Response File: Author Response.docx

Reviewer 3 Report

Dear Author.

I am happy with your responses and actions taken.

Well done.

Author Response

Reviewer#3:

Point1:I am happy with your responses and actions taken. Well done.

Response1: Thank you for your positive comment.

Author Response File: Author Response.docx

Reviewer 4 Report

Great work with the revisions! Only comment is it would be nice to see the LOOCV outputs plotted up in the results, also you said you used R2 to asses the predictive power but didn’t report it (a table with predictive results might be a nice addition).

Author Response

Reviewer#4:

Point1:Great work with the revisions! Only comment is it would be nice to see the LOOCV outputs plotted up in the results, also you said you used R2 to asses the predictive power but didn’t report it (a table with predictive results might be a nice addition).

Response1: Thank you for your positive comment and further suggestion. I added the table 4 to report the performance of the prediction model.

Table 4. Performance of the prediction model for aboveground carbon stock.

Prediction model

RMSE

(kg CO2e)

R2

(kg CO2e)

 

Aboveground carbon stock

18.79

0.74

17.26

0.74

Author Response File: Author Response.docx

Back to TopTop