Remote Sensing of Surface and Subsurface Soil Organic Carbon in Tidal Wetlands: A Review and Ideas for Future Research

Round 1

Reviewer 1 Report

The overall idea is very good but I was quite confused while reading the paper because the authors provided a review paper with a small research inside it (the case of 4.2 section. In my opinion, 4.2 section can be summarized and added in 4.1 and the detailed text should be moved to a supplementary material.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

 

 

Review of Remote Sensing manuscript 1726108 – “Remote Sensing of Surface and Belowground Soil Organic Carbon in Tidal Wetlands: A Review and Ideas for Future Research

 

Manuscript Recommendation

This manuscript presents a valuable literature review and proposes some key research questions concerning remote sensing of soil organic carbon (SOC) in tidal wetlands. The authors also summarize use of covariates for SOC remote sensed SOC predictions and suggest use of novel covariates that have potential for improving remote sensed SOC. The data and discussion are scientifically valid, and results receive appropriate and adequate statistical development and evaluation. The data are well-presented, and conclusions are consistent with the results obtained. The manuscript needs minor revision to address a few key details that could cause confusion across the scientific disciplines that Remote Sensing serves. Detailed review comments follow that outline suggested additions and and/or edits. The manuscript is appropriate for publication in Remote Sensing following revision. The abstract is rephrased to suggest use of the term subsurface as a replacement of belowground.

 

 

Detailed review comments.

 

Use of the term belowground

 

The term “belowground” has an established and widely understood definition- “Situated or occurring below the surface of the ground.” For example, belowground biomass includes all live roots > 2 mm that are beneath the soil surface.

 

This manuscript for important reasons (remote sensed surface layer data) excludes the surface soil layer from what is commonly referred to as belowground SOC. To avoid potential cross discipline confusion and conflation of the term belowground, the authors should consider use of an alternate term when referring to materials beneath the surface layer.

 

This reviewer suggests the term subsurface, which is defined as earth material near but not exposed at the surface of the ground. Technically the term subsurface excludes the surface soil layer, which is consistent with the use followed in this manuscript.

 

If the authors retain the term belowground, they should define it near the beginning and state that the definition varies from the widely known definition.

 

The manuscript title and keywords should include the term subsurface

 

The abstract is rephrased incorporated subsurface in lieu of belowground.

 

Title suggestion

Remote Sensing of Surface and Subsurface Soil Organic Carbon in Tidal Wetlands: A Review and Ideas for Future Research

 

Rephrased Abstract

Abstract: Tidal wetlands, a vital reservoir of soil organic carbon (SOC) can benefit from remote-sensing studies that enable improved spatio-temporal estimation and mapping of SOC stock. We found that a majority of remote sensing SOC mapping efforts have focused on upland ecosystems, not on tidal wetlands. We present a comprehensive review detailing the types of remote sensing models and methods, standard input variables, results, and limitations used in the handful of studies on tidal wetland SOC. Based on that synthesis, we pose several unexplored research questions and methods critical to advancing tidal wetland SOC science. Among those, the applicability of machine learning and deep learning models in predicting surface SOC and modeling requirements for subsurface SOC are the most important ones. We did not find any remote sensing study aimed at modeling subsurface SOC in tidal wetlands. Since tidal wetlands store a significant amount of SOC in subsurface layers, we hypothesize that surface layer SOC to be an important covariable along with other biophysical and climate variables in predicting subsurface SOC. Our preliminary results using field data from tidal wetlands in the southeastern United States and machine learning model output from a mangrove ecosystem in India revealed a strong non-linear but significant relationship (r2 = 0.68 and 0.20, respectively, p < 2.2*10-16) between surface and subsurface SOC at various depths. We investigated the applicability of the Gridded Soil Survey Geographic Database (gSSURGO) database for tidal wetland SOC by comparing SOC stocks calculated from Smithsonian's blue carbon network core-data (collected mainly in 2011) to SOC stocks calculated from SSURGO data extracted in 2018. SSURGO consistently yielded greater SOC stocks for surface and most subsurface layers in tidal wetlands. We conclude that a novel machine learning framework that utilizes remote sensing data and derived products, standard covariables reported in the limited literature, and more importantly, other novel and potentially informative covariables specific to tidal wetlands, such as tidal inundation frequency and height, vegetation species, and soil algal biomass could improve remote sensing-based tidal wetland SOC studies

 

Keywords

Add the term subsurface

 

 

Introduction

 

Rephrase lines 38-52

Wetlands are important transitional ecosystems that remove substantial nitrogen and phosphorus from upland runoff [1,2]. They play a critical role in recharging water tables, retaining nutrients, absorbing pollutants from runoff, filtering sediment, and sequestering atmospheric carbon [3–6]. Wetlands also support diverse wildlife, fisheries, and wet-field agriculture around the world. Historically, wetlands were considered unproductive agricultural lands. They were often drained and manipulated to conduct out different agricultural activities, but in the late 20th century the importance of wetlands and ecosystem services was recognized, triggering wide-ranging mapping efforts [1,2]. The early mapping efforts started with identifying the boundaries of wetlands, vegetation mapping, and periodic surveillance and change detection [7–10]. Due to the rising concentrations of carbon dioxide in the atmosphere, wetlands have become the focus of a wide range of research studies because of their ability to sequester carbon in large amounts [11]. Wetlands are effective in carbon sequestration due to anaerobic conditions; scarcity of oxygen requires use of less efficient (e.g., Mn, Fe) electron acceptors for microbially mediated carbon oxidation, and thus organic matter accumulates and is retained for longer durations [12,13].

 

 

Page 8

Line 328-329

Rephrase

The main limitation for the SOC mapping in tidal wetlands has been the inability to predict SOC below the surface layers because of limited SOC data at depth.

 

Page 9

Lines 339-340

Rephrase

The unique saturated conditions in wetlands enable soils to capture and retain SOC in surface and subsurface soil layers for a long period

 

Page 12

Lines 475-477

Rephrase

SSURGO data were mostly extracted in 2018 with some data from 2007??, whereas coastal carbon core data for the southeastern wetlands derive mainly from 2011 with some points from 2017-2018.

 

The authors in the text specifically state gSSURGO (raster spatial data) as the extract source utilized. The year 2007 for a gSSURGO extract would be incorrect as. gSSURGO first became available about 2012 or later.  Was a SSURGO (polygon-based data) extract done in 2007?

 

NRCS has a recommended citation for gSSURGO or SSURGO that includes month, date, and year.

 

Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Soil Survey Geographic (SSURGO) Database. Available online at https://sdmdataaccess.sc.egov.usda.gov. Accessed [month/day/year].

The SSURGO database is updated annually so the citation should include the time of the data extraction.

Data extractions at different times need separate citations.

 

The 2011 and 2017-2018 years for the CBCN cores are inconsistent with the caption for Figure 3 that states 2007-2018.

 

Lines 477-478

Rephrase

SSURGO data consistently gave higher SOC stocks compared to CBCN data for different depth ranges, as reported in Table 2.

 

The 2011 EPA National Wetland Condition Assessment (NWCA) or wetland inventory data is included in CBCN data. The CBCN data, however, includes cores other than NWCA. Does this study include just NWCA cores from CBCN?  if so, this should be stated.

 

 

Results of CBCN and SSURGO comparison

 

The calculated carbon stocks in Table 2 and require two data parameters - organic carbon (OC) and bulk density (BD). The larger SSURGO SOC stocks may be a function of both higher OC and higher OC values compared to CBCN data. Most BD measurements derive from well drained (water table well below the surface), mineral soils.

Thus, the BD values listed in SSURGO may be high for organic-rich soil horizons with high water tables (i.e., tidal wetlands)

Obtaining bulk density (intact) samples from tidal areas is difficult as water tables are at or near the surface and organic-rich soils common in tidal areas have low particle and bulk density. Lack of accurate BD values at depth may be a larger limitation to modeling SOC than measured OC values.

 

 

References.

Reference 2 as given is citing a book review not the original publication.

The original publication is:

Mitsch, William J., and James G. Gosselink. Wetlands. John Wiley & Sons, 2015.

 

Consider incorporating the following reference into the discussion and citations as it is pertinent to the topic and study area.

 

Osland, Michael J., et al. "Climate and plant controls on soil organic matter in coastal wetlands." Global change biology 24.11 (2018): 5361-5379.

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Overall, I see the value of the study and the literature review by the authors. There are a few issues that I would recommend addressing.

The references to point spectroscopy and point spectroscopy as the study is focused on remote sensing. There is a very extensive soil spectroscopy literature, that really is beyond the scope of this paper. Point spectroscopy serves a valuable and related, but still somewhat separate, objective compared to remote sensing. 

The authors also need to explain what the error by depth is with the modelled mangrove data used in their study. As the data is modelled, there is more uncertainty about what error might be in the dataset affecting the results. 

The final main issue is that the authors needs to more clearly explain what additional wetland variables they are proposing. They also need to present variables that are more direct measurements, as the types of modelled inputs (i.e. vegetation maps derived from Sentinel) present modelling with modelling results, which may magnify errors.

Specific comments: 

Line 81: Remove spectroradiometer discussion

Line 92: Remove spectroradiometer discussion

Table 1: Consider adding PRISMA as a hyperspectral data source

Line 169: Define NDMI

Line 389: Provide the model accuracy by depth wit the mangrove SOC data, not sure the overall accuracy.

Line 418: The correlation of 0.2 is quite weak and doesn't support the hypotheses, even though it is significant. This is especially the case when looking at Figure 4, where the correlation values at 60-100 and 100 - 200 are 0.09, and 0.04. Which while significant, indicate essentially no relationship. This could be a result of inaccuracy in the dataset (as modelled data), and the model not working well at depth. It doesn't invalidate the conclusions from the other dataset, but it does not support the author's hypotheses.

Line 470: Reword: therefore, paired Wilcoxon matched signed pairs exact test was used. 

Line 530: Suggesting vegetation species map as an input is not a clear suggestion. Would this really be necessary or simply ensuring training data collected from different vegetation communities? Would then using the input covariates from vegetation type mapping be sufficient without then needing to model with other model outputs?

Line 538: Remote sensing of AMF based on canopy characteristics again suggests using model outputs. Why not just use the canopy covariates? I would suggest the authors propose more direct remote sensing solutions. 

Line 582: This is misleading the correlations may be been statistically significant, but they were not environmentally meaningful in the mangrove dataset. 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

I do not agree with the authors using the mangrove data to support their hypothesis. The relationships between surface and subsurface carbon is very weak in their data. Yes the correlation is statistically significant, but that does not mean the relationship is strong or useful for prediction. Just that the correlation coefficient characterizes the strength of the overall relationship with certainty. Relying on p-values to conclude that the relationship can be used for subsurface SOC modelling is a misuse of p-values. 

The US wetland data supports their hypothesis, but the mangrove data only weakly does and there is no relationship at the 60-100 and 100-200 cm depths. Correlation coefficients of 0.09 and 0.04 indicate essentially no relationship. Significant or not. 

These specific issues need to be addressed:

Line 422 – 424: I don’t agree with using significance irrespective of the coefficient. Significance is just to determine if you can rely on the difference being reflective of the larger population. Correlations of 0.09 and 0.04 really indicate no relationship between surface and SOC at these depths.

Line 590 – 591: The data from the mangrove site does not support this statement. Correlations of 0.09 and 0.04 are very weak. Indicate essentially no relationship. 

 

 

 

Author Response

Thanks for you comments, Please see the attachment.

Author Response File: Author Response.docx