the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Jan 2024
Monitoring cropland daily carbon dioxide exchange at field scales with Sentinel-2 satellite imagery
Abstract. Improving the accuracy of monitoring cropland CO2 exchange at heterogeneous spatial scales is of high importance to limit spatial and temporal uncertainty of terrestrial carbon (C) dynamic estimates. A combination of field scale eddy covariance (EC) CO2 flux data and spatially matched Sentinel-2 derived vegetation indices (VIs) was tested as an approach to upscale agricultural C flux. The ability of different VIs to estimate daily net ecosystem exchange (NEE) and gross primary productivity (GPP) based on linear regression models was assessed. Most VIs showed high (>0.9) and statistically significant (p<0.001) correlations with GPP and NEE although some VIs deviated from the seasonal pattern of CO2 exchange. In contrast, correlations between ecosystem respiration (Reco) and VIs were weak and statistically not significant, and no attempt was made to estimate Reco from VIs. Linear regression models explained generally more than 80 % and 70 % of the variability of NEE and GPP, respectively, with high variability amongst the individual VIs. The performance in estimating daily C fluxes varied amongst VIs depending on C flux component (NEE or GPP) and observation period. RMSE values ranged from 1.35 g C m-2 d-1 using the Green Normilized Difference Vegetation Index (GNDVI) for NEE to 5 g C m-2 d-1 using the Simple Ratio (SR) for GPP. This equated to an underestimated net C uptake of only 41 g C m-2 (18 %) and an overestimation of gross C uptake of 854 g C m-2 (73 %). Differences between measured and estimated C fluxes were mainly explained by the diversion of the C flux and VI signal during winter, when C uptake stayed low while VI values indicated an increased C uptake due to relatively high crop leaf area. Overall, results exhibited similar error margins as mechanistic crop models. Thus, they indicated suitability and developability of the proposed approach to monitor cropland C exchange with satellite derived VIs.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2021 KB)
-
Supplement
(17 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2021 KB) - Metadata XML
-
Supplement
(17 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2988', Anonymous Referee #1, 16 Feb 2024
Overview
This manuscript combines field scale EC data and satellite derived VIs to estimate the carbon flux over a local cropland in Europe. Several kinds of VIs are used to estimate NEE and GPP based on linear regression models. Their method is reasonably straightforward, and assumption is clearly stated. The results are well investigated from multiple aspects, which makes their conclusions more concrete. Their estimated values including uncertainties are fairly compared to previous works. The authors explain the strengths and weaknesses of their approach very well. Some figures and tables may be a little hard to read/follow to readers, though.
Followings are my minor comments and suggestions.
In the abstract, please clarify the size/location of the study field.
L20: Please define “RMSE”
L79: “limits”
L86: Please define “NIR”
L114: Please define “a.s.l.”
L134: Data “were” measured…
L141: For date expression, please be consistent (“5th March” or “23 August”).
L164: Please define “SCL”.
L198: There is no “section 0” in this manuscript.
L222: It will help to put the total NEE, GPP and Reco for each growing season in Fig.2 as you did for the climate parameters.
L254: “apart from” Do you mean “except for”?
L277-278: The VI observations are sparse compared to NEE/GPP, so it might be hard to tell “these VIs lag about 18 days behind the NEE signal for the senescence period of WR” from the Fig. 3. Please specify how you verify it.
L308: Please define “lowess (should be LOWESS)”.
L287-288: Why the signs of correlations are opposite of yours?
L289: This sentence is hard to understand and seems not quite matching the results shown in Fig. 4.
L321 and L324: “nir”--> please define and change to ”NIR”.
L322: “swir” --> please define and change to “SWIR”.
L323: “amongst the highest” is not true. Do you mean “one of the highest”?
L383: Please define “E”. “RMSE” does NOT “increased” (but improved).
L405: It is not very clear that R2-values are calculated between what parameters.
L422: “τ” --> Do you mean “ρ”?
L434-439: These sentences are hard to understand. “overestimation” and “underestimation” with respect to what? Your estimated results?
L435: “204” and “217” are for cumulative C flux, right? Please mention that and specify the time period.
L440: “85%” --> “85.66%”.
L442: Please add “, respectively” after “(17.93%)”.
L443: “1.35 and 1.36” --> “1.36 and 1.35”. “(Table 6)” --> “(Table 5)”.
L451: Please add “, respectively” after “2021/2022,”.
L452-453: This sentence is a little hard to understand. I do not see how it is related to Table 3. Where the number “311 g C m-2” come from?
L467: Please add “,” after “harvest”.
In sections 3.5.2 and 3.5.3, more detailed explanations of the models are needed such as the model simulation years (is there any inter-annual variability)? Readers might understand better if section 3.5.3 starts with explanation of Table 6, and then give comparisons with references.
Figure 1: It is a little too hard to understand what the percentage of “cumulative FP area” means. How was it calculated? It is not clear the definitions of “main field” and “area of interest”.
Figure 2 (The top panel): Please specify the duration of “flowering of WR”. (the first 2 top panels): please explain the gray dots and black solid lines in the figure caption. Also, please differentiate the GPP and Reco by using different color or line type. I do not see the “urea spreading” in 2020/2021 listed in Table 1.
Please give x-axis title as “Time” and do not use the numbers such as 5.3, 1.5, 1.7… which are hard to understand. You could show the dates (i.e. 05 Mar 2020) only for the start and end of the growing season or so. Also, it may be helpful for readers to put the x-tick values at the top of the figure, too.
Figure 3: Y-title for VIs should be “(-1)*VIs”? There is no explanation for blue dots. Again, the x-axis’ thick name of “5.3, 1.6…” are hard to follow. Please give monthly tick marks and label them properly. Also, it would be easier for readers to compare the time series of NEE/GPP and VIs if they have a common zero (horizontal) line. Please add legends for the vertical broken- and dotted- lines.
Figure 5: Please put an x-title and change the x-tick name format.
Table 1: What is “ca.”?
Table 3: Please state the unit clearly.
Table 4: Please explain what the bold and italic numbers indicate either in the caption or main text. Why are ρ-values not shown here (just wonder) as in Table 5? It could be better to make the line thicker between different growing periods (in Table 5, also).
Table 5: Please explain what the bold numbers indicate.
Table 6: the top of the 3rd column should be “NEE and GPP estimated [g C m-2]”. You could add “mean” to Table 4 and 5 as well because it was mentioned in the main text. What does the “**” indicate? Again, please explain about the bolded numbers (it seems they are mentioned in the next, but not all of them are bolded).
S1 The third paragraph: “Reco … but showed no response the grubbing events”. Is this true? It seems the Reco increased between the first and second grubbing events. The last (fifth) paragraph, “GPP and Reco were both lower …”: because the signs of GPP and Reco are opposite, “lower” is a little confusing. Could be better to say “smaller”.
Citation: https://doi.org/10.5194/egusphere-2023-2988-RC1 -
AC1: 'Reply on RC1', Pia Gottschalk, 15 Mar 2024
Dear Reviewer 1,
I first like to apologise for only replying now! I was drowned in other committments. The more I like to thank you for your diligent review and constructive comments, for spotting erroneous details and giving helpful hints for improving the manuscript. I did now go through all the comments one by one and replied briefly to each. See attached. As I understand there are no big critical problems which we cannot address successfully but please check yourself.
Thank you very much again!
Best regards,
Pia Gottschalk (and on behalf of all authors)
-
AC2: 'Reply on RC1', Pia Gottschalk, 18 Mar 2024
Correction: The area of the 'main field' at our study site Heydenhof is about 112 ha. We will mention size and location in the abstract as suggested by RC1.
Citation: https://doi.org/10.5194/egusphere-2023-2988-AC2
-
AC1: 'Reply on RC1', Pia Gottschalk, 15 Mar 2024
-
RC2: 'Comment on egusphere-2023-2988', Anonymous Referee #2, 07 Mar 2024
Gottschalk et al. present a high-resolution investigation of the vegetation indices against the carbon fluxes from a tower in the cropland. They processed the flux observations thoroughly to ensure the data quality and estimated the fluxes based on linear models using VIs, which were then compared with crop models. They showed that observed fluxes correlate well with the field management actions, phenological development of crops, and some Sentinel-2-derived VIs (especially for GPP). They concluded with discussions of the limitations of their analyses (e.g., adopting a simple linear model).
Major comment:
In general, the data processing is thorough, and the results are well presented. Leveraging the high-resolution feature from Sentinel-2 data, their EC-based carbon fluxes and Vis are all matched to the plot scale, which addressed the common spatial mismatch issue in prior studies.
Despite the promising flux-VI relationship presented, my major concern lies in the novelty of this study and the representativeness of such flux-VI relationships across crop traits and climates (despite that the authors acknowledged till the very end). The authors suggested that their analyses indicate “the suitability and developability of the proposed approach to monitor cropland C exchange with satellite derived Vis” (L25-26) and acknowledged the current challenges in flux estimates being the low number of EC towers in croplands (L79-81). When considering croplands over different climates/regions, how robust is the use of a linear model to predict carbon fluxes? Would involving additional meteorological variables be more helpful (which is relative to statements on L490-491)?
The results/conclusions presented from this study did not seem to necessarily address the limitation of low data density of the flux observations, as the results were derived from one crop flux tower in Germany with seemingly low data density (compensating for better data quality). Moreover, regarding the high-resolution capability, are there any prior studies that utilized Vis from Landsat and Sentinel-2 for estimating carbon fluxes over croplands at meter scales?
In sum, the knowledge/technical gaps from prior studies and the novelty of this study can be better phrased. I would suggest conducting a more thorough literature review and rewording relevant text to highlight the advances of this study from prior studies (e.g., the examination of various VIs, the estimation of the carbon budget, the alignment of footprint between VIs and flux signals, and investigation of crop management signals from EC and VI data, from my understanding?).
Here are some minor comments:
Figures 2, 3: the time format on the x-axis is not super intuitive. Either change the date/time format or add a caption. What do the vertical lines in the dashed or dotted lines indicate in Figure 3? Add a few text labels to facilitate the comprehension of the discussion on page 18 (e.g., the timing for flowering on L273 and senescence on L276 - 278).
L294-296: The poorer correlation between Reco and VIs seems as expected. Does that suggest additional meteorological variables describing the air and soil columns are needed?
There are a lot of numbers going into the result sections and tables. There could be better ways to present the data more intuitively, e.g., replacing Table 6 with a bar plot.
Citation: https://doi.org/10.5194/egusphere-2023-2988-RC2 -
AC3: 'Reply on RC2', Pia Gottschalk, 24 Mar 2024
Dear reviewer 2,
we would like to thank you for your thoughtful review of our manuscript and critical remarks and helpful comments which we tried to address all in the attached document. I hope we were able to address your concerns sufficiently and found convincing reasoning to overcome them.
I do apologise for the delay in responding to your comments!
Kind regards,
Pia Gottschalk in behalf of all co-authors
-
AC3: 'Reply on RC2', Pia Gottschalk, 24 Mar 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-2988', Anonymous Referee #1, 16 Feb 2024
Overview
This manuscript combines field scale EC data and satellite derived VIs to estimate the carbon flux over a local cropland in Europe. Several kinds of VIs are used to estimate NEE and GPP based on linear regression models. Their method is reasonably straightforward, and assumption is clearly stated. The results are well investigated from multiple aspects, which makes their conclusions more concrete. Their estimated values including uncertainties are fairly compared to previous works. The authors explain the strengths and weaknesses of their approach very well. Some figures and tables may be a little hard to read/follow to readers, though.
Followings are my minor comments and suggestions.
In the abstract, please clarify the size/location of the study field.
L20: Please define “RMSE”
L79: “limits”
L86: Please define “NIR”
L114: Please define “a.s.l.”
L134: Data “were” measured…
L141: For date expression, please be consistent (“5th March” or “23 August”).
L164: Please define “SCL”.
L198: There is no “section 0” in this manuscript.
L222: It will help to put the total NEE, GPP and Reco for each growing season in Fig.2 as you did for the climate parameters.
L254: “apart from” Do you mean “except for”?
L277-278: The VI observations are sparse compared to NEE/GPP, so it might be hard to tell “these VIs lag about 18 days behind the NEE signal for the senescence period of WR” from the Fig. 3. Please specify how you verify it.
L308: Please define “lowess (should be LOWESS)”.
L287-288: Why the signs of correlations are opposite of yours?
L289: This sentence is hard to understand and seems not quite matching the results shown in Fig. 4.
L321 and L324: “nir”--> please define and change to ”NIR”.
L322: “swir” --> please define and change to “SWIR”.
L323: “amongst the highest” is not true. Do you mean “one of the highest”?
L383: Please define “E”. “RMSE” does NOT “increased” (but improved).
L405: It is not very clear that R2-values are calculated between what parameters.
L422: “τ” --> Do you mean “ρ”?
L434-439: These sentences are hard to understand. “overestimation” and “underestimation” with respect to what? Your estimated results?
L435: “204” and “217” are for cumulative C flux, right? Please mention that and specify the time period.
L440: “85%” --> “85.66%”.
L442: Please add “, respectively” after “(17.93%)”.
L443: “1.35 and 1.36” --> “1.36 and 1.35”. “(Table 6)” --> “(Table 5)”.
L451: Please add “, respectively” after “2021/2022,”.
L452-453: This sentence is a little hard to understand. I do not see how it is related to Table 3. Where the number “311 g C m-2” come from?
L467: Please add “,” after “harvest”.
In sections 3.5.2 and 3.5.3, more detailed explanations of the models are needed such as the model simulation years (is there any inter-annual variability)? Readers might understand better if section 3.5.3 starts with explanation of Table 6, and then give comparisons with references.
Figure 1: It is a little too hard to understand what the percentage of “cumulative FP area” means. How was it calculated? It is not clear the definitions of “main field” and “area of interest”.
Figure 2 (The top panel): Please specify the duration of “flowering of WR”. (the first 2 top panels): please explain the gray dots and black solid lines in the figure caption. Also, please differentiate the GPP and Reco by using different color or line type. I do not see the “urea spreading” in 2020/2021 listed in Table 1.
Please give x-axis title as “Time” and do not use the numbers such as 5.3, 1.5, 1.7… which are hard to understand. You could show the dates (i.e. 05 Mar 2020) only for the start and end of the growing season or so. Also, it may be helpful for readers to put the x-tick values at the top of the figure, too.
Figure 3: Y-title for VIs should be “(-1)*VIs”? There is no explanation for blue dots. Again, the x-axis’ thick name of “5.3, 1.6…” are hard to follow. Please give monthly tick marks and label them properly. Also, it would be easier for readers to compare the time series of NEE/GPP and VIs if they have a common zero (horizontal) line. Please add legends for the vertical broken- and dotted- lines.
Figure 5: Please put an x-title and change the x-tick name format.
Table 1: What is “ca.”?
Table 3: Please state the unit clearly.
Table 4: Please explain what the bold and italic numbers indicate either in the caption or main text. Why are ρ-values not shown here (just wonder) as in Table 5? It could be better to make the line thicker between different growing periods (in Table 5, also).
Table 5: Please explain what the bold numbers indicate.
Table 6: the top of the 3rd column should be “NEE and GPP estimated [g C m-2]”. You could add “mean” to Table 4 and 5 as well because it was mentioned in the main text. What does the “**” indicate? Again, please explain about the bolded numbers (it seems they are mentioned in the next, but not all of them are bolded).
S1 The third paragraph: “Reco … but showed no response the grubbing events”. Is this true? It seems the Reco increased between the first and second grubbing events. The last (fifth) paragraph, “GPP and Reco were both lower …”: because the signs of GPP and Reco are opposite, “lower” is a little confusing. Could be better to say “smaller”.
Citation: https://doi.org/10.5194/egusphere-2023-2988-RC1 -
AC1: 'Reply on RC1', Pia Gottschalk, 15 Mar 2024
Dear Reviewer 1,
I first like to apologise for only replying now! I was drowned in other committments. The more I like to thank you for your diligent review and constructive comments, for spotting erroneous details and giving helpful hints for improving the manuscript. I did now go through all the comments one by one and replied briefly to each. See attached. As I understand there are no big critical problems which we cannot address successfully but please check yourself.
Thank you very much again!
Best regards,
Pia Gottschalk (and on behalf of all authors)
-
AC2: 'Reply on RC1', Pia Gottschalk, 18 Mar 2024
Correction: The area of the 'main field' at our study site Heydenhof is about 112 ha. We will mention size and location in the abstract as suggested by RC1.
Citation: https://doi.org/10.5194/egusphere-2023-2988-AC2
-
AC1: 'Reply on RC1', Pia Gottschalk, 15 Mar 2024
-
RC2: 'Comment on egusphere-2023-2988', Anonymous Referee #2, 07 Mar 2024
Gottschalk et al. present a high-resolution investigation of the vegetation indices against the carbon fluxes from a tower in the cropland. They processed the flux observations thoroughly to ensure the data quality and estimated the fluxes based on linear models using VIs, which were then compared with crop models. They showed that observed fluxes correlate well with the field management actions, phenological development of crops, and some Sentinel-2-derived VIs (especially for GPP). They concluded with discussions of the limitations of their analyses (e.g., adopting a simple linear model).
Major comment:
In general, the data processing is thorough, and the results are well presented. Leveraging the high-resolution feature from Sentinel-2 data, their EC-based carbon fluxes and Vis are all matched to the plot scale, which addressed the common spatial mismatch issue in prior studies.
Despite the promising flux-VI relationship presented, my major concern lies in the novelty of this study and the representativeness of such flux-VI relationships across crop traits and climates (despite that the authors acknowledged till the very end). The authors suggested that their analyses indicate “the suitability and developability of the proposed approach to monitor cropland C exchange with satellite derived Vis” (L25-26) and acknowledged the current challenges in flux estimates being the low number of EC towers in croplands (L79-81). When considering croplands over different climates/regions, how robust is the use of a linear model to predict carbon fluxes? Would involving additional meteorological variables be more helpful (which is relative to statements on L490-491)?
The results/conclusions presented from this study did not seem to necessarily address the limitation of low data density of the flux observations, as the results were derived from one crop flux tower in Germany with seemingly low data density (compensating for better data quality). Moreover, regarding the high-resolution capability, are there any prior studies that utilized Vis from Landsat and Sentinel-2 for estimating carbon fluxes over croplands at meter scales?
In sum, the knowledge/technical gaps from prior studies and the novelty of this study can be better phrased. I would suggest conducting a more thorough literature review and rewording relevant text to highlight the advances of this study from prior studies (e.g., the examination of various VIs, the estimation of the carbon budget, the alignment of footprint between VIs and flux signals, and investigation of crop management signals from EC and VI data, from my understanding?).
Here are some minor comments:
Figures 2, 3: the time format on the x-axis is not super intuitive. Either change the date/time format or add a caption. What do the vertical lines in the dashed or dotted lines indicate in Figure 3? Add a few text labels to facilitate the comprehension of the discussion on page 18 (e.g., the timing for flowering on L273 and senescence on L276 - 278).
L294-296: The poorer correlation between Reco and VIs seems as expected. Does that suggest additional meteorological variables describing the air and soil columns are needed?
There are a lot of numbers going into the result sections and tables. There could be better ways to present the data more intuitively, e.g., replacing Table 6 with a bar plot.
Citation: https://doi.org/10.5194/egusphere-2023-2988-RC2 -
AC3: 'Reply on RC2', Pia Gottschalk, 24 Mar 2024
Dear reviewer 2,
we would like to thank you for your thoughtful review of our manuscript and critical remarks and helpful comments which we tried to address all in the attached document. I hope we were able to address your concerns sufficiently and found convincing reasoning to overcome them.
I do apologise for the delay in responding to your comments!
Kind regards,
Pia Gottschalk in behalf of all co-authors
-
AC3: 'Reply on RC2', Pia Gottschalk, 24 Mar 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 364 | 133 | 34 | 531 | 53 | 21 | 26 |
- HTML: 364
- PDF: 133
- XML: 34
- Total: 531
- Supplement: 53
- BibTeX: 21
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Pia Gottschalk
Aram Kalhori
Christian Wille
Torsten Sachs
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2021 KB) - Metadata XML
-
Supplement
(17 KB) - BibTeX
- EndNote
- Final revised paper