Improvement of FAPAR Estimation Under the Presence of Non-Green Vegetation Considering Fractional Vegetation Coverage

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper entitled “Improvement of FAPAR estimation under presence of non-green vegetation considering fractional vegetation coverage” proposed an improved method for estimating photosynthetically active radiation absorption fraction (FAPAR), which improves the estimation accuracy in the presence of non-green vegetation by considering vegetation cover fraction (FVC). It is reasonable in structure and standard in writing. However, there are still some problems to be clarified in this paper. I present below some comments to be considered for author before accepting the paper.

1. “In this paper, to address the problem of FAPAR overestimation due to the non-green vegetation canopy not considered, we propose an improved method for FAPAR estimation considering the effect of fractional vegetation coverage (FVC) to reduce the overestimation of FAPAR calculated based on the Beer-Lambert’s law using MODIS LAI data.“ The assumption of Beer Lambert's law is a homogeneous turbid medium. Although introducing FVC can partially solve the problem of non-green vegetation canopy, this assumption may still not be fully applicable in some complex vegetation structures, such as forest or shrubs. Because forests have three-dimensional features. How does the author consider this?

2. The author uses NDVI to calculate FVC, but in tropical high vegetation coverage areas, NDVI may reach saturation. How did the author consider the limitations of using NDVI to calculate vegetation coverage?

3. When discussing the accuracy of FAPAR estimation for different types of biological communities, it is suggested that the author further explore why certain types of biological communities (such as tropical broadleaf evergreen forest and boreal forest) tend to underestimate and why some tend to overestimate? I guess this may be related to errors in MODIS LAI products, inaccurate FVC estimation, or the specificity of vegetation structure?

Author Response

Comments 1: “In this paper, to address the problem of FAPAR overestimation due to the non-green vegetation canopy not considered, we propose an improved method for FAPAR estimation considering the effect of fractional vegetation coverage (FVC) to reduce the overestimation of FAPAR calculated based on the Beer-Lambert’s law using MODIS LAI data.” The assumption of Beer Lambert's law is a homogeneous turbid medium. Although introducing FVC can partially solve the problem of non-green vegetation canopy, this assumption may still not be fully applicable in some complex vegetation structures, such as forest or shrubs. Because forests have three-dimensional features. How does the author consider this?

Response 1: Since the Beer-Lambert’s law requires the satisfaction of the homogeneous turbid medium assumption, it tends to yield more accurate results in relatively uniform vegetation types such as croplands and grasslands. When considering FVC, the impact of the non-homogeneous turbid canopy of forests or shrubs on the calculation results is primarily reflected in two aspects: the presence of non-green canopy areas, such as bare ground between trees or shrubs, can lead to the failure of the homogeneous turbid medium assumption at the pixel scale [1]. Additionally, in vegetation types with pronounced seasonal growth variations, non-green canopy areas may arise when the peak growing season has not been reached, as not all leaves have fully developed. This can result in the canopy distribution at the pixel scale not satisfying the homogeneous turbid medium assumption. Therefore, incorporating FVC has shown improved results in temperate mixed forests.

However, this method essentially assumes that the distribution of leaves of each individual tree or shrub can be approximated as satisfying the homogeneous turbid medium assumption. For forests or shrubs with three-dimensional features, the canopy of individual trees may also fail to meet the homogeneous turbid medium assumption. The current methodology is unable to address this limitation, and further in-depth research is required to improve upon this shortcoming. Consequently, it is explicitly stated in the discussion that “This assumption may still not be fully applicable in some complex vegetation structures, such as forests or shrubs.”

Revision 1: “The non-homogeneous turbid canopy of forests or shrubs may also influence the results. Although introducing FVC can partially solve the problem of non-green vegetation canopy, this method essentially assumes that the distribution of leaves of each individual tree or shrub can be approximated as satisfying the homogeneous turbid medium assumption. Thus, this assumption is still not fully applicable in the complex vegetation structures of forests or shrubs since they exhibit three-dimensional features. The current methodology is unable to address this limitation, and further in-depth research is required to improve upon this shortcoming.” was added, see Section 4.4, lines 574-581 in the manuscript.

 

Comments 2: The author uses NDVI to calculate FVC, but in tropical high vegetation coverage areas, NDVI may reach saturation. How did the author consider the limitations of using NDVI to calculate vegetation coverage?

Response 2: According to the description of the MODIS LAI/FPAR backup algorithm, the NDVI for various vegetation types tends to saturate at an average value of around 0.8, at which point both LAI and FAPAR are also at relatively high levels (with an average LAI of 5.25 and an average FAPAR of 0.88) [2]. Under the condition, FVC is also at a high level, nearly approaching 100%. This means that NDVI saturation should not affect the FVC estimation greatly. In addition, as illustrated in Figure 7, when FVC is high, its influence on the calculation of the FAPAR_FVC model is relatively minor. When NDVI approaches saturation, FVC itself is already high. Thus, even if NDVI saturation induces some uncertainties in FVC estimation, its effect on the FAPAR_FVC model should be limited.

Revision 2: “The NDVI is used to calculate FVC; however, NDVI may reach saturation, particularly in tropical regions with high FVC, where the NDVI typically saturates at an average value of approximately 0.8 [2]. Under the condition, both LAI and FAPAR are also at relatively high levels, and FVC is also at a high level, nearly approaching 100%. This means that NDVI saturation should not affect the FVC estimation greatly. In addition, as illustrated in Figure 7, when FVC is high, its influence on the calculation of the FAPAR_FVC model is relatively minor. When NDVI approaches saturation, FVC itself is already high. Thus, even if NDVI saturation induces some uncertainties in FVC estimation, its effect on the FAPAR_FVC should be limited. However, under conditions of low FVC, NDVI may be influenced by soil background effects. Vegetation indices less affected by soil background, such as EVI and SAVI, could be considered for estimating FVC.” was added, see Section 4.4, lines 546-556 in the manuscript.

 

Comments 3: When discussing the accuracy of FAPAR estimation for different types of biological communities, it is suggested that the author further explore why certain types of biological communities (such as tropical broadleaf evergreen forest and boreal forest) tend to underestimate and why some tend to overestimate? I guess this may be related to errors in MODIS LAI products, inaccurate FVC estimation, or the specificity of vegetation structure?

Response 3: That's true. The overestimation and underestimation of FAPAR_FVC are primarily associated with errors in MODIS LAI. As shown in Figure 8, the underestimation samples are derived from tropical broadleaf evergreen forests and boreal forests, where MODIS LAI is consistently underestimated across all samples. This underestimation of LAI consequently leads to the underestimation of FAPAR in these sites. Conversely, at other sites where overestimation occurs, the number of samples with overestimated LAI clearly exceeds those with underestimated LAI, resulting in an overestimation of FAPAR. Additionally, uncertainties in FVC estimation mainly arise from the influence of clouds and fog on NDVI. For tropical broadleaf evergreen forests, cloud and fog effects from January to April in this study led to an underestimation of NDVI, thereby causing a more pronounced underestimation of FAPAR_FVC for this vegetation type. The impact of forest canopy structure is not yet evident in the results. This study includes four forest biomes, among which tropical broadleaf evergreen forests and boreal forests at two sites exhibit underestimation, while the other two biomes at three sites show overestimation, without demonstrating a consistent pattern.

Revision 3: “The overestimation and underestimation of FAPAR_FVC are primarily associated with errors in MODIS LAI. As shown in Figure 8, the underestimation samples are derived from tropical broadleaf evergreen forests and boreal forests, where MODIS LAI is consistently underestimated across all samples. This underestimation of LAI consequently led to the underestimation of FAPAR in these sites. Conversely, at other sites where overestimation occurs, the number of samples with overestimated LAI clearly exceeds those with underestimated LAI, resulting in an overestimation of FAPAR.” was added, see Section 4.4, lines 525-531 in the manuscript.

 

1. Shabanov, N.; Gastellu-Etchegorry, J.-P. The stochastic Beer–Lambert–Bouguer law for discontinuous vegetation canopies. Journal of Quantitative Spectroscopy and Radiative Transfer 2018, 214, 18–32, doi:https://doi.org/10.1016/j.jqsrt.2018.04.021.
2. Knyazikhin, Y.; Glassy, J.; Privette, J.L.; Tian, Y.; Lotsch, A.; Zhang, Y.; Wang, Y.; Morisette, J.T.; Votava, P.; Myneni, R.B.; et al. MODIS leaf area index (LAI) and fraction of photosynthetically active radiation absorbed by vegetation (FPAR) product (MOD15) algorithm theoretical basis document. 1999. Available online: https://modis.gsfc.nasa.gov/data/atbd/atbd_mod15.pdf (accessed on 29 January 2024).

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Dear,

 

After reading the article entitled "Improvement of FAPAR estimation under presence of non-green vegetation considering fractional vegetation coverage", I have the following comments:

 

1- The article is well written and has an adequate approach to the problem of overestimation of MODIS LAI/FPAR. The text cites quality and relatively recent articles to address the subject, the problems involved and the research that is being developed on the subject.

 

2- Methodologically, I agree with the methods used and their formulations, but I have reservations regarding the database. Data sources for the data used in this study are based on 1-year series of data, which raises the question: Why not larger series? You need to be aware of whether adding a larger series of data could have any impact on the results.

 

3- The results can be better presented through a Taylor diagram, to evaluate the efficiency of the estimates. You can keep the current figures, and add a new one with a Taylor diagram.

 

4- It is confusing to note the inclusion of figures 5 and 6 and table 4 in the discussion section. All possible research results, in the form of tables, graphs and any other form of data expression, must be presented in the results section. The discussion section should be dedicated to explaining the phenomena identified in the results, the support in the literature that is available to defend the point of view presented and how the results are positioned in the literature, corroborating other theses or starting a new one. Including results in the discussion section is inappropriate.

Author Response

Comments 1: The article is well written and has an adequate approach to the problem of overestimation of MODIS LAI/FPAR. The text cites quality and relatively recent articles to address the subject, the problems involved and the research that is being developed on the subject.

Response 1: Thank you for your patient review and constructive comments.

 

Comments 2: Methodologically, I agree with the methods used and their formulations, but I have reservations regarding the database. Data sources for the data used in this study are based on 1-year series of data, which raises the question: Why not larger series? You need to be aware of whether adding a larger series of data could have any impact on the results.

Response 2: Since ground validation data require in situ measured LAI and FAPAR data, it is essential to consider the issue of sample site heterogeneity: specifically, whether the representativeness and quantity of sampling data are sufficient to reflect the average conditions of the sample site. Therefore, collecting validation data at the site scale requires substantial investment in human and material resources. Only in the case of the BigFoot Project and the grass savanna, the number of samples within the sample sites were adequate to represent the average conditions of the respective sample plots according to our review. Unfortunately, the BigFoot Project did not provide data with a longer time series. However, the Dahra data from the grass savanna consist of two years of observational data (2001 and 2002) collected at the same site. Despite the notable disparity in FAPAR between 2001 and 2002, the improvement in the presented method over the two years exhibited the consistent results.

Revision 2: “The Dahra data from the grass savanna represented two years of observational data (2001 and 2002) collected at the same site, and the improvement in the presented method over the two years exhibited the consistent pattern. Despite the notable disparity in FAPAR between 2001 and 2002, the accuracy of FAPAR_FVC was clearly higher than that of FAPAR_LAI and FAPAR_MOD in both years. Specifically, the MAPE decreased by 9.1% and 50.2% in 2001, and by 18.9% and 56.7% in 2002, respectively.” was added, see Section 3.3, lines 326-331 in the manuscript.

 

Comments 3: The results can be better presented through a Taylor diagram, to evaluate the efficiency of the estimates. You can keep the current figures, and add a new one with a Taylor diagram.

Response 3: Taylor diagrams and associated descriptions were added, see Section 3.2, lines 295-310 in the manuscript.

 

Comments 4: It is confusing to note the inclusion of figures 5 and 6 and table 4 in the discussion section. All possible research results, in the form of tables, graphs and any other form of data expression, must be presented in the results section. The discussion section should be dedicated to explaining the phenomena identified in the results, the support in the literature that is available to defend the point of view presented and how the results are positioned in the literature, corroborating other theses or starting a new one. Including results in the discussion section is inappropriate.

Response 4: The relevant content was reorganized into the Section Results, lines 362-380, 382-385, 389-390, 397-427.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript entitled "Improvement of FAPAR estimation under presence of non-green vegetation considering fractional vegetation coverage",The manuscript addresses an important issue related to improving the estimation of Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) by incorporating fractional vegetation coverage (FVC). This enhancement has implications for Gross Primary Production (GPP) modeling and remote sensing applications. The study is well-motivated, and the methodology is generally sound. However, several areas require clarification, additional explanation, and revision to improve the manuscript's quality and readability.These problems are listed as below:

1.The methodology section, while detailed, could benefit from a schematic flowchart that visually outlines the steps of the FAPARFVC calculation.

2.While the manuscript includes comparisons with FAPARLAI and FAPARMOD, the statistical validation could be enhanced by including additional metrics or alternative validation datasets.

3.Figures and tables are informative, but their captions could provide more detailed descriptions. For example, the caption for Figure 2 should elaborate on what each model comparison reveals.

4.The discussion would benefit from a deeper exploration of the limitations of the proposed approach, particularly under conditions of low FVC or in biome types where MODIS LAI errors are significant.

5.The legends for Figures 3 and 4 should indicate what the error bars or shaded regions represent (e.g., standard deviation, confidence intervals).

6.Consider adding a table summarizing the performance of FAPARFVC, FAPARLAI, and FAPARMOD across different biomes for quick reference.

7.Ensure all references are complete and formatted according to journal guidelines. Some references in the discussion section lack DOI links.

8.Insufficient introduction of references in the past five years

9.Line 143: In Equation (1), the placement of NDVI variables should be checked for typographical alignment.

10.In Figure 1, add labels to clarify the relationship between LAIpixel and LAIcanopy.

Author Response

Comments 1: The methodology section, while detailed, could benefit from a schematic flowchart that visually outlines the steps of the FAPAR_FVC calculation.

Response 1: The schematic flowchart was added, see Section 2.2.1, Figure 2, lines 174-180 in the manuscript.

 

Comments 2: While the manuscript includes comparisons with FAPAR_LAI and FAPAR_MOD, the statistical validation could be enhanced by including additional metrics or alternative validation datasets.

Response 2: Since ground validation data require in situ measured LAI and FAPAR data, it is essential to consider the issue of sample site heterogeneity: specifically, whether the representativeness and quantity of sampling data are sufficient to reflect the average conditions of the sample site. Acquiring the necessary data entails a substantial amount of effort, making the expansion of validation data unfeasible in the short term. Therefore, we have introduced a new statistical metric, RPIQ, as suggested and updated the relevant data. The validation results of RPIQ are generally consistent with those of MAPE, with the exception of minor differences observed in the validation outcomes for the full dataset and the tallgrass prairie samples.

Revision 2: For further details, please refer to Section 2.2.3, lines 209-212; Section 3.1, lines 222-223; and Section 3.2, lines 274-279, 281-285 in the manuscript.

 

Comments 3: Figures and tables are informative, but their captions could provide more detailed descriptions. For example, the caption for Figure 2 should elaborate on what each model comparison reveals.

Response 3: The relevant description has been updated in the figures and tables.

Revision 3: “The accuracy of FAPAR_LAI is higher than that of FAPAR_MOD, and the accuracy of FAPAR_FVC is superior to that of FAPAR_LAI.” was added, see Section 3.1, Figure 3, lines 235-237 in the manuscript.

“For biome types overestimated by FAPAR_MOD, FAPAR_LAI and FAPAR_FVC estimated FAPAR with clearly lower overestimates compared to FAPAR_MOD, with FAPAR_FVC being more clearly lower. For the biome types underestimated by FAPAR_MOD, the underestimation of FAPAR was more severe for FAPAR_LAI and FAPAR_FVC, especially for FAPAR_FVC.” was added, see Section 3.2, Figure 4, lines 266-269 in the manuscript.

“For biome types overestimated by FAPAR_MOD, FAPAR_LAI and FAPAR_FVC estimated FAPAR with clearly lower overestimates compared to FAPAR_MOD, with FAPAR_FVC being more clearly lower. For the biome types underestimated by FAPAR_MOD, the underestimation of FAPAR was more severe for FAPAR_LAI and FAPAR_FVC, especially for FAPAR_FVC.” was added, see Section 3.2, Table 2, lines 282-285 in the manuscript.

“Most biome types demonstrated overestimation of FAPAR by each model during off-peak vegetation growth period.” was added, see Section 3.3, Figure 6, lines 323-325 in the manuscript.

“FAPAR_FVC showed clearly higher improvement in the overestimation of FAPAR compared to FAPAR_LAI and FAPAR_MOD during the off-peak growth period than during the peak growth period.” was added, see Section 3.3, Table 3, lines 345-346 in the manuscript.

“The accuracy of all three models in simulating FAPAR increased with increasing FVC.” was added, see Section 3.3, Table 4, lines 364-365 in the manuscript.

“The proportion of FAPAR decrease gradually increased with decreasing FVC.” was added, see Section 3.4, Figure 7, lines 379-380 in the manuscript.

“There was a significantly positive correlation between LAI error and FAPAR error.” was added, see Section 3.5, Figure 8, lines 418-419 in the manuscript.

 

Comments 4: The discussion would benefit from a deeper exploration of the limitations of the proposed approach, particularly under conditions of low FVC or in biome types where MODIS LAI errors are significant.

Response 4: The limitations of this study are also evident under conditions of low FVC and in biome types where MODIS LAI errors are significant. Under conditions of low FVC, NDVI may be influenced by soil background effects. Vegetation indices less affected by soil background, such as EVI and SAVI, could be considered for estimating FVC.

In section 4.4, we have discussed the limitations of the proposed approach from the point of view of MODIS LAI error. However, we did not emphasize in biome types where MODIS LAI errors are significant. The revised manuscript highlighted the regions with significant underestimation of LAI, such as tropical broadleaf evergreen forest and boreal forest.

Revision 4: “However, under conditions of low FVC, NDVI may be influenced by soil background effects. Vegetation indices less affected by soil background, such as EVI and SAVI, could be considered for estimating FVC.” was added, see Section 4.4, lines 554-556 in the manuscript.

“Especially in biomes with significant underestimation of LAI, such as tropical broadleaf evergreen forest and boreal forest.” was added, see Section 4.4, lines 589-590 in the manuscript.

 

Comments 5: The legends for Figures 3 and 4 should indicate what the error bars or shaded regions represent (e.g., standard deviation, confidence intervals).

Response 5: The relevant legends with shaded regions representing 95% confidence intervals were updated and shown in Figures 3, 4, and 6.

 

Comments 6: Consider adding a table summarizing the performance of FAPAR_FVC, FAPAR_LAI, and FAPAR_MOD across different biomes for quick reference.

Response 6: The relevant table was added, see Section 3.2, Table 2, lines 280-285 in the manuscript.

 

Comments 7: Ensure all references are complete and formatted according to journal guidelines. Some references in the discussion section lack DOI links.

Response 7: DOI links were added.

 

Comments 8: Insufficient introduction of references in the past five years.

Response 8: Several references for the past five years were added.

[4]        Li, Z.H., Zhang, Q., Li, J., Yang, X., Wu, Y.F., Zhang, Z.Y., Wang, S.H., Wang, H.Z., Zhang, Y.G., 2020. Solar-induced chlorophyll fluorescence and its link to canopy photosynthesis in maize from continuous ground measurements. Remote Sensing of Environment, 236(1), 111420.

[16]      Sun, Y.K., Yuan, D.K., Zheng, X., Yang, S.S., Zhang, S., Zhang, J.H., Bai, Y., 2024. Estimation of radiation scalar using deep learning for improved gross primary productivity estimation based on a light-use efficiency model. International Journal of Digital Earth, 17(1), 2402421.

[17]      Huang, L.X., Yuan, W.P., Zheng, Y., Zhou, Y.L., He, M.Z., Jin, J.X., Huang, X.J., Chen, S.Y., Liu, M., Guan, X.B., Jiang, S.Z., Lin, X.F., Li, Z.L., Tang, R.L., 2024. A dynamic-leaf light use efficiency model for improving gross primary production estimation. Environmental Research Letters, 19(1), 014066.

[25]      Zheng, Y.J., Xiao, Z.Q., Li, J., Yang, H., Song, J.L., 2022. Evaluation of global fraction of absorbed photosynthetically active radiation (FAPAR) products at 500 m spatial resolution, Remote Sensing, pp. 3304.

[26]      Lai, P.Y., Marshall, M., Darvishzadeh, R., Nelson, A., 2025. Crop productivity under heat stress: a structural analysis of light use efficiency models. Agricultural and Forest Meteorology, 362(2025), 110376.

 

Comments 9: Line 143: In Equation (1), the placement of NDVI variables should be checked for typographical alignment.

Response 9: The relevant typographical alignment was revised.

 

Comments 10: In Figure 1, add labels to clarify the relationship between LAI_pixel and LAI_canopy.

Response 10: The relevant relationship description was added, see Section 2.1, Figure 1, lines 147-158 in the manuscript.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have made acceptable revisions to address the concerns raised in the previous review.