Review of Remote Sensing-Based Methods for Forest Aboveground Biomass Estimation: Progress, Challenges, and Prospects

Round 1

Reviewer 1 Report

Dear authors!

I think good work was conducted during review preparation. Many aspects of remotes sensing of forest cover to estimate AGB were considered in your work.

I general, I satisfied with your work. But I suggest some improvements of your work.

1. It is necessary to give some information of AGB estimation within Wildland-Urban Interface (WUI). It seems to me, this is important part of forest-covered territories.

2. Table 1. I suggest to link this information with supported references within this table.

3. Table 2. The same situation. I suggest to add references too.

Page 13. Please, support AI methods like ANN, KNN, SVM, RF, GB, ME with references.

Conclusion

I suggest to mark out a set of key findings with supported conclusions.

References

I satisfied with presented references list. 

It is all right.

Author Response

Thanks for your kind comments concerning our manuscript entitled “Review of Remote Sensing-based Methods for Forest Aboveground Biomass Estimation: Progress, Challenges, and Prospects” (Manuscript ID: forests-2383107). Those comments are all valuable and very helpful for improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made corrections which we hope meet with approval. Revised portion are marked in the manuscript. The response letter is attached.

Author Response File: Author Response.docx

Reviewer 2 Report

General comments:

This paper provides a review of remote sensing for biomass estimation. Overall, I did not find any major issues with the manuscript: it was well organized and well written, and it covers a wide range of literature on the subject area. However, I did find the novelty of this review somewhat lacking. There have been a number of review papers focusing on the main ideas presented here. As such, the authors should make a better effort to put a more substantial spin on the topic to make it stand out from the rest of the published literature.

For example, Lister et al. (2020) “Use of remote sensing data to improve the efficiency of national forest inventories: a case study from the United States national forest inventory” published in this same journal covers many of the same topics described here. Similarly, a paper I led (Knott et al. 2023 “Effects of outliers on remote sensing-assisted biomass estimation: a case study from the United States national forest inventory” in Methods in Ecology and Evolution) also talks about many similar topics, especially comparing ground-based data to remote sensing. Works by Erik Naesset, Terje Gobakken, Erkki Tomppo, and Ronald McRoberts have covered many of these topics as well (Naesset and Gobakken are cited a few times, but Tomppo and McRoberts are not). Examples from these authors include McRoberts et al (2010) “Using remotely sensed data to construct and assess forest attribute maps and related spatial products” published in Scandinavian Journal of Forest Research; McRoberts and Tomppo (2007) “Remote sensing support for national forest inventories” published in Remote Sensing of Environment; and Stahl et al (2016) “Use of models in large-area forest surveys: comparing model-assisted, model-based, and hybrid estimation” published in Forest Ecosystems; among others.

Specific comments

L 36: consider changing “at the three poles” to “at high latitudes and elevations” because glacial melting is not occurring only at the north/south pole/highest elevation, but rather more generally at higher latitudes/elevations
L55-61: This is a very long sentence, with multiple parts. Consider breaking into a few smaller statements.

L69: Also, the actual amount of land inventoried tends to be quite small; for example, the USDA Forest Service Forest Inventory and Analysis program uses 1 plot per 2400 ha of land.

L71 and throughout: A few colleagues have pointed out to me over the years that the term “remote sensing data” should instead by “remotely sensed data” because “remote sensing” is a verb (action) or noun (field of study), rather than an adjective.

L131: What do you mean “AGB level of the community”? This seems to be a bit of a circular argument here. AGB is the forest characteristic you are trying to predict based on structural information.

L132: Which studies? Add citations here. Also, what are the units here? Please spell out t/hm²m.

L132-137: These few sentences seem to be a summary of a study that assessed how different dimensions of AGB scaled relative to each other. Whereas other studies you have included state just the main points, this one goes into significantly more detail and is hard to follow the logic of why it was included.

L133, 138, 157: what is meant by “depressed forests”? I have not seen this term used before. Do you mean depression swamp forests? Or lowland forests?

L167: Again, if you use the phrase “Previous studies…” please make sure to cite those studies.

L207-211: these sentences are essentially a repeat of earlier parts of the manuscript and don’t add much additional information. Suggest changing the opening paragraph of this section to a more focused introduction.

L215: It would be useful for a reader new to the field of remote sensing to have a short description of what is meant by passive vs. active remote sensing.

L238: I would hesitate to say “the best tool” because it all depends on the goal of the project. E.g., for small area estimation, where you might only have a few dozen pixels from some of the passive sensors, or in areas where you might get a saturation effect for some commonly used indices like NDVI, passive sensors don’t give you much information relative to things like aerial or terrestrial lidar. This shows up in L322 when you state that “…optical remote sensing data limits its application…”

L261: seems to be an unnecessary paragraph break or a misplaced one.

L332: not sure what this sentence is trying to say? What is meant by “single-wood”?

L376: this has already been done, e.g., see Dubayah et al. (2022). GEDI launches a new era of biomass inference from space. Environmental Research Letters. https://doi.org/10.1088/1748-9326/ac8694 and associated GEDI data products (https://daac.ornl.gov/GEDI/guides/GEDI_L4A_AGB_Density_V2_1.html and https://daac.ornl.gov/GEDI/guides/GEDI_L4B_Gridded_Biomass.html)

L456: This is a very roundabout way to say “A statistical model is constructed between variables from remote sensing and field samples, and the model is used to predict AGB for areas

L464: LR, linear dummy variable, and linear mixed-effects are all parametric models, and therefore this study does not necessarily support the statement that “estimation accuracy is typically not very high”. Instead, you should look to papers that compare parametric to nonparametric models for AGB estimation, such as:

Tanase et al. (2014) Sensitivity of L-band radar backscatter to forest biomass in semiarid environments: a comparative analysis of parametric and nonparametric models. IEEE Transactions on Geoscience and Remote Sensing. https://doi.org/10.1109/TGRS.2013.2283521

Safari et al. (2018) Comparison of satellite-based estimates of aboveground biomass in coppice oak forests using parametric, semiparametric, and nonparametric modeling methods. Journal of Applied Remote Sensing. https://doi.org/10.1117/1.JRS.12.046026

Li and Mao (2020) Comparison of canopy closure estimation of plantations using parametric, semi-parametric, and non-parametric models based on GF-1 remote sensing images. Forests. https://doi.org/10.3390/f11050597

L513-517: Seems like there should be some citations supporting the description of RF, especially since there are phrases in quotations.

L587: “estimation results are more reliable” is then followed by “limiting their applicability”… I think this is an important point to make, that mechanistic models suffer from the “put bad in, get bad out” problem. In well parameterized mechanistic models, results might be superior to empirical approaches, but rarely are mechanistic models able to be fully characterized, and the assumptions and/or lack of data make them nearly impossible to be useful at scale.

L592: how are comprehensive models different from mechanistic models? To my understanding, FAREAST, LANDIS/LANDIS-II, FVS, SORTIE, and other similar models that simulate forest growth are generally still considered to be a type of mechanistic model. At a minimum, if you are wanting to separate comprehensive models from mechanistic models, you should include information about other commonly used simulation models such as the ones listed above.

Paragraph beginning on L623: There seems to be a lack of citations in this paragraph. See Lister et al (2020) and Knott et al (2023) (mentioned above) for more information about this topic. These citations and references cited within these publications could serve as references about many of the issues brought up in this paragraph. This also brings up a point about this paper that seems to be missing: there is very little discussion about what sources of field inventory data are being used. For example, in the United States, the USDA Forest Service, Forest Inventory and Analysis program provides national forest inventory (NFI) data collected as a probability-based (semi-systematic) sample to be used for the purpose of, among others, calibration of forest biomass estimation and mapping. Similarly, ecological research networks such as the National Ecological Observatory Network (NEON), Forest Global Ecosystem Observatory (ForestGEO), and others provide fewer sampling locations but a wealth of data (including commissioning their own remote sensing).

L775: “has” should be “have”

Figures and tables:

L109: Figure 1 shows a variety of sensors in and around a forest. However, I don’t think it’s reasonable to claim that the figure is showing “data and methods.” Rather, figure 3 shows different data and methods. How useful is figure 1? Consider including more information in the caption and main text supporting why figure 1 should be included here. Likewise, Figure 4 contains a similar but slightly different view of remote sensing of forests. Perhaps there would be a way to integrate the information contained in Figures 1 and 4? Figure 4 caption should do a better job of explaining what is being shown.

Figure 3: check journal specifications whether all abbreviations need to be listed in the caption. E.g., missing “VIs”, “LAI”, “SIF”, and “GPP"

A few minor suggestions above

Author Response

Thanks for your kind comments concerning our manuscript entitled “Review of Remote Sensing-based Methods for Forest Aboveground Biomass Estimation: Progress, Challenges, and Prospects” (Manuscript ID: forests-2383107). Those comments are all valuable and very helpful for improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made corrections which we hope meet with approval. Revised portion are marked in the manuscript. The response letter is attached.

Author Response File: Author Response.docx

Reviewer 3 Report

This review focuses on estimating above-ground forest biomass (AGB) in the context of global climate change, evaluating various remote sensing methods and data. The aim is to understand the responses of forest ecosystems to greenhouse gas emissions and promote sustainable forest resource management. Methods such as passive optical remote sensing, microwave remote sensing, LiDAR, and multi-source data integration are examined, along with empirical, physical, mechanistic, and comprehensive models. The review also analyzes the uncertainty in AGB estimation and suggests strategies for improving accuracy and reducing uncertainties. Lastly, prospects are discussed, such as using LiDAR technology for large-scale sampling and integrating various data sources to obtain accurate global AGB estimates.

The review is clear and linear; however, in the AGB estimation methods, it cites articles that employ techniques such as Random Forest, support vector machine, and ANN but does not mention more modern approaches using convolutional networks or UNet networks. The authors should really include such approaches, as literature is quickly growing on it (e.g "Above-ground Biomass Wheat Estimation: Deep Learning with UAV-based RGB Images" or "ReUse: REgressive Unet for Carbon Storage and Above-Ground Biomass Estimation.")

In conclusion, the review provides a clear and comprehensive overview of AGB estimation methods and will prove to be a valuable resource for researchers in the field if authors make it cover as many recent papers as possible 

Author Response

Thanks for your kind comments concerning our manuscript entitled “Review of Remote Sensing-based Methods for Forest Aboveground Biomass Estimation: Progress, Challenges, and Prospects” (Manuscript ID: forests-2383107). Those comments are all valuable and very helpful for improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made corrections which we hope meet with approval. Revised portion are marked in the manuscript. The response letter is attached.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Most of the changes sufficiently address my previous concerns. However, one major edit that needs to be changed: my suggestion to the authors was to use "remotely sensed data" instead of "remote sensing data", with the former being preferred and the latter not preferred but still acceptable. However, the authors changed to "remotely sensing data" which is now completely incorrect. 

Most of the changes sufficiently address my previous concerns. However, one major edit that needs to be changed: my suggestion to the authors was to use "remotely sensed data" instead of "remote sensing data", with the former being preferred and the latter not preferred but still acceptable. However, the authors changed to "remotely sensing data" which is now completely incorrect. 

Author Response

Please see attachment for response comments.

Author Response File: Author Response.docx