Next Article in Journal
Tracking the Land Use/Land Cover Change in an Area with Underground Mining and Reforestation via Continuous Landsat Classification
Next Article in Special Issue
Estimating Stand Age from Airborne Laser Scanning Data to Improve Models of Black Spruce Wood Density in the Boreal Forest of Ontario
Previous Article in Journal
Remote Sensing of Lake Ice Phenology across a Range of Lakes Sizes, ME, USA
Previous Article in Special Issue
Benchmarking the Retrieval of Biomass in Boreal Forests Using P-Band SAR Backscatter with Multi-Temporal C- and L-Band Observations
Open AccessArticle
Peer-Review Record

Reflectance Properties of Hemiboreal Mixed Forest Canopies with Focus on Red Edge and Near Infrared Spectral Regions

Remote Sens. 2019, 11(14), 1717; https://doi.org/10.3390/rs11141717
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Remote Sens. 2019, 11(14), 1717; https://doi.org/10.3390/rs11141717
Received: 18 May 2019 / Revised: 12 July 2019 / Accepted: 18 July 2019 / Published: 20 July 2019
(This article belongs to the Special Issue Remote Sensing of Boreal Forests)

Round 1

Reviewer 1 Report

The authors present a paper on the reflectance properties of hemiboreal forest canopies.

The paper is unusual in several points. Data has been collected with a point VNIR spectrometer from a helicopter while most comparable studies nowadays use an imaging spectrometer. 

The analyses do not go very deep, but many correlations are reported. The strangest result is that only two distinct wavelength are found as REIP using the first derivative method. Maybe some spectral smoothing might help here.

I did not find serious mistakes in the paper. If the editors and the other reviewers find it fit for publication I will not object.

There are a lot of missing articles ("the") in the text. Language editing will improve legibility.

Author Response

Thank you for reviewing our manuscript!


Reviewer 1: “The paper is unusual in several points. Data has been collected with a point VNIR spectrometer from a helicopter while most comparable studies nowadays use an imaging spectrometer.”


Reply: We have added more detailed explanation about the advantages of our measurement setup, correction methods (e.g. simultaneous irradiance measurements) and comparison with imaging systems such as APEX and CASI.


Reviewer 1: “The strangest result is that only two distinct wavelength are found as REIP using the first derivative method. Maybe some spectral smoothing might help here.”


Reply: This effect would disappear with lowered spectral resolution (i.e. re-sampling to lower resolution or using linear four-point interpolation) but it is the actual feature of higher resolution spectra not noise.


Reviewer 1: “There are a lot of missing articles ("the") in the text. Language editing will improve legibility.”


Reply: We have edited language.


Reviewer 1: “I did not find serious mistakes in the paper. If the editors and the other reviewers find it fit for publication I will not object.”


Reply: Thank you!



Author Response File: Author Response.pdf

Reviewer 2 Report

In this manuscript, a general evaluation of spectral characteristics of an specific arboreal vegetation (hemiboreal mixed forest) is performed. The authors focus their efforts in describing changes in the red edge spectral region and in a proxy for Sun Induced Fluorescence (Farea). These factors were compared to vegetation properties at the stand level and to spectral response in other wavelengths in the visible and near infrared region. It was mainly observed that: (i) spectral response in individual bands was negatively correlated with biomass related traits: (ii) correlation between vegetation indices (VIs) was high; (iii) correlation between VIs / Farea and individual bands indicated that most of these indicators were strongly positively correlated with reflectance in the NIR, except PRI. From these results the authors concluded that, the unexpected relationship between vegetation development and individual bands reflection could be related to shadow effects; the simple ratio formulation proposed by the authors could replace more complex formulations to describe changes related to red edge inflection point; PRI was apparently better correlated with actual photosynthetic status than Farea.

Despite presenting some interesting results, some points have not been sufficiently explored by the authors. For example, a better explanation for the negative correlation between reflectance and traits related to biomass should be provided. Also, an attempt to decouple the effects of canopy structure  and leaf pigments on the results observed should be made. An alternative process to explain these observations could be the reduction of chlorophyll contents followed by increase in canopy structure (similar behaviour of reflectance in all wavelengths), however this would not explain a negative correlation between biomass related traits and spectral response. Introducing specific indices for the evaluation of canopy traits (e.g., MCARI2, WDVI, etc.) could help with the interpretation of the data. Also, a broader exploration of red edge related formulations could better demonstrate the advantage of the simple ratio formulation proposed in this study. Besides that, some specific observations / concerns are indicated below:

Line 9: What do you mean by medium high resolution? Maybe better to indicate a ‘relative high resolution’ and include in parentheses the width of the spectral bands and the spectral sampling interval (number of bands).

Line 30: Platforms cited with completely different spatial resolution and potential application. Maybe include examples of future solutions for SIF related studies (e.g.: FLEX mission, etc.) with more comparable outputs.

Line 33: It is not clear what scale is the target of the current study. It appears to be the large scale from this paragraph, however in the previous phrase sentinel data is cited (medium resolution platform). Better to focus on the scale targeted or to include examples of different scales in all cases, if the object is to describe recent advancements in remote sensing-based vegetation monitoring at medium and large scales.

Line 34: Again, in this paragraph a total different scale is introduced (leaf level). Maybe better to indicate that here.

Line 38: Rather than facilitate the use, it may be pointed out that in the absence of direct fluorescence measurements non-photochemical quenching might be a good proxy to actual vegetation photosynthetic capacity. Therefore, the relationship between PRI and  SIF in this context is that both can provide assessment of current status of vegetation photosynthetic activity and relating them might be more meaningful than just observing the relationship between pigment content and plant structure (through reflectance measurements in other wavelengths) with current vegetation photosynthetic capacity (represented through SIF or PRI).

Line 43 to 47: Another factor affecting PRI values is the canopy structure itself. It might be useful to add this to the list of factors cited. A paper describing this can be cited (e.g.: 10.1016/j.rse.2016.12.015, or some specific publication targeting the characterization of arboreal vegetation).

Line 53 to 58: It might be useful to indicate the spatial resolution for possible applications of the findings (at least the scale in which the measurements were made could already be indicated to better situate the reader).

Line 80 to 81: It might be of interest to indicate that each spectral measurement corresponded to the average of 8 measurements taken in 17 x 3 meters strips (assuming that one spectra was registered by second). This might help the reader to visualize the support of the measurements without the necessity to calculate it.

Figure 1: It might be useful to represent the measurements as points to better illustrate the density of the information acquired in each stand. Were all stands included in the analysis? Some of them seem to be outside the flight path and spectral mixing with nearby areas might have occurred. Better to consider / indicate / discuss that.

Line 99: How the stray light correction is made? Might be of interest to the reader.

Line 104 to 105: Were the areas corresponding to the sensor footprint extracted/located in the field for acquiring the ground truth or the whole stand was considered as the ‘measured’ area? Variation between sampling density and location might have had considerable impact in the outputs. How the authors have tackled this aspect during the analysis?

Figure 2: It would be clearer if points were added to the curves indicating the location of the actual spectral bands in all sub-figures. This would allow the reader to better appreciate the resolution of the data acquired.

Line 111: The fact that simple ratio and normalized formulations resulted in a near linear relationship might be related to the scale / method of spectral measurement. Since the main advantage of a normalized index is the compensation for changes in illumination conditions if the measurements with which the indices are calculate correspond to a relatively large surface and if several observations are averaged the difference between normalized and simple ratio tend to decrease, I suppose. Better to indicate / check that since this observation might not hold in other situations.

Line 113: Why these specific formulations were chosen? For example a formulation with 705 and 750 nm instead of 705 and 783 nm is more frequently used in other articles. Please indicate the articles or  empirical evidence for the use of these indices and band combinations.

Line 116: Maximal change or inflection point? Please clarify.

Line 124: Please correct the spelling of ‘exact’.

Line 124 to 129: The authors describe the method used to approximate the fluorescence signal from relatively coarse spectral information. Is this methodology describe anywhere else (please provide source)? If this is not the case, there has been a laboratory-based or proximal sensing evaluation to verify if this hypothesis hold?

Line 133: It is important to point out the fact that other methods for red-edge inflection point calculation exists (e.g.: 10.1016/j.rse.2005.12.011). It is also extensively reported in the literature that a double peak effect is observed for the reflectance in the red-edge region, which can be attributed to changes in fluoresce emissions that follows alterations in vegetation photosynthetic capacity, as far I understand (10.1109/IGARSS.2003.1293856). Therefore, these results are already expected and no mention to previous literature about this topic has been made so far (in particular in the introduction). I would recommend to take this factor into account since one of the main objectives of the article are to describe the relationship between the red-edge inflection point, SIF and vegetation properties.

Line 139 to 141: If this discontinuity has been well reported in previous articles, should this evaluation be included in the analysis? Other methods with better performance have been proposed in the literature. It might be of interest to include a justification for the use of such approach, and why other methods were not evaluated.

Line 143 to 144: Might this be an indication that the spectral information acquired lacked resolution for the implementation of such analysis? Was any smoothing applied before the derivative calculation? It is important to explain why this happened.

Line 147 to 150: Since this information is already present in the literature is this evaluation necessary?

Line 158: Table 1 should be placed near this citation to make reference easier.

Line 159 to 164: Tests with the same or very close spectral wavelengths have been performed previously in other studies (e.g.: 10.1016/1011-1344(93)06963-4, Figure 4). Please indicate the added value of presenting such comparison. Rather focus on aspects that have not been tackled before. Please make these aspects more explicit while explaining the motivation of the study.

Line 174 to 176: Maybe it is important to indicate that the correlation is negative between vegetation indices (VIs) and reflectance in the red region. This signals that while values of the VIs increased chlorophyll content probably decreased and canopy structure increased. This is also in agreement with SIF values which are higher for higher canopy structure levels (higher values of reflectance in the NIR) since with larger canopy structure the photosynthetic capacity potentially increased.

Line 186 to 187: Please provide references for this affirmation.

Line 187 to 189: This affirmation might hold only for the factors varying in this specific dataset. Since measurements were made in a single date, factors affecting the variability of the VIs values might have been restricted to a narrow process (e.g.: increase in canopy structure followed by decrease in leaf chlorophyll content). For a more in depth evaluation of such hypothesis analysis based on radiative transfer modelling could be used to evaluate the behaviour of these formulations under a broader conditions variation. Otherwise, the authors should search in the literature for similar cases in which this behaviour is observed.

Line 194 to 197: Were these parameters described in the material and methods? Please provide reference for them and move this part to the material and methods if not included there yet.

Line 197 to 199: This might be related to the changes observed in this specific dataset. Apparently as vegetation structure increases chlorophyll content decreases which makes reflectance in the visible (related to plant pigments) and NIR wavelengths (related to vegetation structural traits) change in a similar pattern (e.g.: higher reflectance values in the NIR region are followed by higher values in the red or green wavelengths due to lower content of leaf chlorophyll in the leaves, according to Table 3; maybe correlation between single bands should be added to better illustrate this point). It might be of interest to point it out.

Line 202 to 203: The negative correlation between LAI and reflectance in the red region indicates that structure gains were potentially inversely related to leaf chlorophyll contents. Better to indicate this to the reader. Also, including the correlation between traits might help the interpretation.

Figure 5. Simplify the y-axis title. Include regressions coefficient (R2) and/or other accuracy indicators (RMSE, linear equation between x and y variables).

Table 3. The title can be simplified, there is no need to list all factors since they are already described in each column and row.

Figure 6. Same as Figure 5. Include accuracy / performance indicators (R2, RMSE, etc.).

Line 212 to 213: Although this might be true, it has already been observed in other articles that this measure is not suitable for vegetation monitoring due to the double peak effect in the red-edge region. Therefore, the inclusion of this parameter in the analysis is questionable.

Line 223 to 226: This is expected and is well described in the literature. Also, the spectrometer used might lack resolution (i.e., band width of approximately 4 nm) for detailed analysis of this aspect. The authors could focus instead on other factors not well described so far or they should make some effort to evaluate LAI based on the spectral information, which could help the interpretation of the data (i.e.: decoupling effects of canopy structure and leaf pigments concentration).

Line 227 to 229: The Double peak effect should be related to SIF in the far red region. As observed for SIF, the increase in canopy structure increased values of fluorescence emitted (positive correlation with NIR wavelengths) and this might have been reflected in the maximum value of the first derivative as well. Besides, canopy structure itself affects largely the reflectance in the red edge region and therefore is expected that growth form is related to the double peak effect. It might be of interest to point this out. Please include citations about this topic, which is extensively reported in the literature.

Line 229 to 233: It might be an indicative that the spectral resolution of the spectroradiometer used is not adequate for this analysis.

Line 233 to 234: These studies already exists as far as I understand. Please include literature about the topic, previous studies cannot be ignored.

Line 238 to 245: This is an adequate observation, computation of the first derivative might also have been deeply affected by the limited resolution of the spectrometer used.

Line 248: remove ‘these’ on ‘these vegetation indices’.

Line 251 to 253: Any particular reason / explanation for that? Very important for the interpretation of the results. Can it be that PRI expressed better the current vegetation photosynthetic capacity while Farea was also affected by canopy structure components and was less effective for this purpose? Please try to explain.

Line 272 to 273: Was this component (shadow) representative in the study sites during the measurements? Was this effect estimated for the areas measured? It cannot be that chlorophyll content and canopy structure presented such a pattern that increase in biomass resulted in negative correlation with indices due to reduction on chlorophyll content at leaf level? Analysis of VIs designed for canopy structure estimation (MCARI2, WDVI, etc.) might help with the interpretation of the data. The analysis of the NIR region alone (single bands) might be more susceptible to other factors than canopy structure.

Line 279 to 283: Was this fact related to shadow effects according to the authors cited? This counterintuitive result (negative relationship between biomass and NIR reflectance) must be better explained / discussed taking into account all factors affecting the reflectance in this region.

Line 298 to 301: Evaluation of VIs related to canopy structure may provide a better insight in the entangled effect of canopy structure, pigments concentration and current vegetation photosynthetic capacity.

Line 303 to 306: Despite the narrow band implementation using bands located in a very sensitive spectral region, simple ratio indices are generally very sensitive to illumination conditions, amongst other factors which are considered in more complex formulations. Has this or other factors affected the outputs obtained? How would this formulation behave in a more broad dataset (multi-temporal, etc.)?

Line 328 to 335: This aspect has been already discussed in previous articles, as far as I understand. I would recommend to focus rather on explaining the negative correlation between reflectance values in individual bands and biomass, for example, or other aspects not well discussed so far (indicated in the previous comments).

Line 336 to 348: It would be interesting to see some considerations regarding future applications of the results obtained in this study, as well recommendations for further exploration of aspects not well explained using the current methodological framework / dataset.

Author Response


Thank you for a very helpful and thorough review!


Reviewer 2: “A better explanation for the negative correlation between reflectance and traits related to biomass should be provided. Also, an attempt to decouple the effects of canopy structure and leaf pigments on the results observed should be made. An alternative process to explain these observations could be the reduction of chlorophyll contents followed by increase in canopy structure (similar behaviour of reflectance in all wavelengths), however this would not explain a negative correlation between biomass related traits and spectral response.”


Reply: In agricultural crops or sparse vegetation the reflectance in NIR region increases as the amount of biomass increases above soil. In hemiboreal forests the canopy brightness in NIR starts to decrease due to the increase in the roughness of the forest canopy as a surface as forest grows higher and biomass increases. The amount of shade will increase with stand age. At the early age before canopy closure the proportion of sunlit ground vegetation decreases as trees grow bigger. After the maximum canopy closure is attained LAI will not change much but stem-wood volume and tree height can continue growing. Forest canopy consists of multiple layers of vegetation with different pigments content and structure (e.g. shade tolerant understory species and more light demanding upper-story species).



Reviewer 2: “Introducing specific indices for the evaluation of canopy traits (e.g., MCARI2, WDVI, etc.) could help with the interpretation of the data.”


Reply: Vast amount of remote sensing studies deal with crops and sparse vegetation where the change in relative contribution of reflectance signal originating from soil is associated with the change in the amount of biomass. MCARI2 (Modified Chlorophyll Absorption Reflectance Index 2) and WDVI (Weighted Difference Vegetation Index) are both specifically designed for those vegetation types where soil background optical properties have crucial influence on reflectance signal measured above the vegetation and therefore soil adjustment would be needed (e.g. crops). We have deliberately tried to limit the number of different vegetation indices in the current study since we made recently a review of the performance of 152 previously published vegetation indices (Hallik et al 2017).


We have added explanation to introduction “Vast amount of remote sensing studies deal with crops and sparse vegetation where the change in the amount of biomass simultaneously changes the relative contribution of reflectance signal originating from soil. Soil background optical properties have crucial influence on reflectance signal measured above the vegetation in such canopies and therefore soil adjustment would be needed to extract the information about vegetation characteristics (Baret et al 1993). Soil line concept has been used in formulation of many vegetation indices (e.g. SAVI, MCARI2, WDVI). However, such approach would not be appropriate in the case of multi-layered closed forest canopies, where background signal can originate from lower layers of trees, herbaceous understory or moss-layer.”


Reviewer 2: “Line 9: What do you mean by medium high resolution? Maybe better to indicate a ‘relative high resolution’ and include in parentheses the width of the spectral bands and the spectral sampling interval (number of bands).”


Reply: We have changed the wording “With relative high spectral resolution top-of-canopy measurements (bandwidth 10 nm, spectral step 3.3 nm) we found that both estimates of red edge

inflection point were well related to ...”


Reviewer 2: “Line 30: Platforms cited with completely different spatial resolution and potential application. Maybe include examples of future solutions for SIF related studies (e.g.: FLEX mission, etc.) with more comparable outputs.”


Reply: We added “and the future ESA FLEX mission (planned for launch by 2023) will improve the spatial resolution of space borne SIF estimates to about 300 x 300 m2.”


Reviewer 2: “Line 33: It is not clear what scale is the target of the current study. It appears to be the large scale from this paragraph, however in the previous phrase sentinel data is cited (medium resolution platform). Better to focus on the scale targeted or to include examples of different scales in all cases, if the object is to describe recent advancements in remote sensing-based vegetation monitoring at medium and large scales.”


Reply: Our results are from airborne top-of-canopy (TOC) measurements. The overall purpose for conducting the reflectance measurements at intermediate TOC scale is to bridge together the different scales from leaf level observation to satellite borne data. Our study was conducted at stand level and we have now added more examples of APEX and CASI measurements at similar scale. We added: “Airborne hyperspectral imagers CASI [Chen et al 1999] and APEX [Schaepman et al 2015] provide high quality spectral at-sensor radiance data. Such data or even non-calibrated digital numbers by new lightweight hyperspectral imagers [Resonon Inc.] can provide vegetation indices or perform classification of targets. For top-of-canopy spectral reflectance simultaneous measurements of irradiance spectra and in case of high-flying airborne measurements [Raczko et al 2017] atmospheric correction are needed. Our study is focused on intermediate scale of so called bottom-of-atmosphere (BOA) / top-of-canopy (TOC) measurements above mixed forest canopy. Low-level airborne spectral data are supported by simultaneous recording of irradiance spectra to convert airborne data to TOC spectral directional reflectance, and by forestry data provided by the National forestry database.


Reviewer 2: “Line 34: Again, in this paragraph a total different scale is introduced (leaf level). Maybe better to indicate that here.”


Reply: Ground validation of physiological responses for satellite-borne observations begins at leaf level so all these different scales must be finally combined. We have made strong efforts to improve the wording and explanation of different scales in introduction section.


Reviewer 2: “Line 38: Rather than facilitate the use, it may be pointed out that in the absence of direct fluorescence measurements non-photochemical quenching might be a good proxy to actual vegetation photosynthetic capacity. Therefore, the relationship between PRI and SIF in this context is that both can provide assessment of current status of vegetation photosynthetic activity and relating them might be more meaningful than just observing the relationship between pigment content and plant structure (through reflectance measurements in other wavelengths) with current vegetation photosynthetic capacity (represented through SIF or PRI).”


Reply: We have added more thorough explanation about the origin of PRI and its relation to NPQ at leaf level. We added to the introduction:

Absorbed light energy can be: (1)used for photosynthesis by photochemical energy conversion, (2) dissipated as heat or (3) re-emitted as chlorophyll fluorescence. The first two processes are called photochemicaland non-photochemical quenching as both mechanisms reduce the chlorophyll fluorescence emissions. If non-photochemical quenching (NPQ) is known or not changing then chlorophyll fluorescence can be used to estimate photosynthesis. PAM chlorophyll a fluorometry allows to calculate parameters such as NPQ but not passive fluorescence measurement systems. Xanthophyll cycle is the protective mechanism that regulates NPQ to dissipate excess energy safely as heat and it causes also a subtle changes at leaf level absorption spectra (Bilger et al 1989). Photochemical reflectance index (PRI) was originally constructed to assess the state of xanthophyll cycle via those spectral changes.”


Reviewer 2: “Line 43 to 47: Another factor affecting PRI values is the canopy structure itself. It might be useful to add this to the list of factors cited. A paper describing this can be cited (e.g.: 10.1016/j.rse.2016.12.015, or some specific publication targeting the characterization of arboreal vegetation).”


Reply: We have added the suggested citation (Gitelson et al 2017) to emphasise that stand-level PRI should not be interpreted as proxy of same physiological processes it is responding at leaf level. Two citations describing canopy structure effects (e.g. non-physiological structural shadowing effects) in boreal forest to PRI have been included already (Mõttus et al 2015; Takala et al 2016) but we improved the wording.



Reviewer 2: “Line 53 to 58: It might be useful to indicate the spatial resolution for possible applications of the findings (at least the scale in which the measurements were made could already be indicated to better situate the reader).”


Reply: Spectra were sampled with the spatial step fo 2.1 m on the flight transect. The footprint of the field-of-view of the UAVSpec at the flight height of 80 m is a circle of the diameter of 3 m. We have also added: “For top-of canopy spectral reflectance simultaneous measurements of irradiance spectra and in case of high-flying airborne measurements atmospheric correction are needed. Our study is focused on intermediate scale of so called bottom-of-atmosphere (BOA) / top-of-canopy (TOC) measurements above mixed forest canopy. Low-level airborne spectral data are supported by simultaneous recording of irradiance spectra to convert airborne data to TOC spectral directional reflectance, and by forestry data provided by the National forestry database.”



Reviewer 2: “Line 80 to 81: It might be of interest to indicate that each spectral measurement corresponded to the average of 8 measurements taken in 17 x 3 meters strips (assuming that one spectra was registered by second). This might help the reader to visualize the support of the measurements without the necessity to calculate it. Figure 1: It might be useful to represent the measurements as points to better illustrate the density of the information acquired in each stand. Were all stands included in the analysis? Some of them seem to be outside the flight path and spectral mixing with nearby areas might have occurred. Better to consider / indicate / discuss that.”


Reply: 8 spectra were registered per second. Considering the flight speed 17 m/s the spectral recordings would be by 2.1 m step. Field-of-view (FOV) on ground was about 2.5 - 3 m. Averaging was made later at stand level. We used 8 m buffer zone to avoid spectral mixing with nearby area. Spectral recordings closer than 8 m to the stand border were rejected and stands with less than 10 spectral recordings (after buffering) were rejected from the analysis. All spectral recordings within one stand were averaged. Stand is the standard unit of forest inventories. One stand is one record in forest inventory database and it means a homogeneous forest patch. We have added more thorough explanation of the meaning of “stand” in forest inventory database.

We added into the section of Material and Methods: “A forest stand is a patch of homogeneous forest considering its species composition, age, tree height, tree density, site type. A forest management inventory database including 1:10,000 map of stands is available for Järvselja Training and Experimental Forest district. The database is updated regularly. Several forest parameters such as species composition, age, breast-height diameter (cm), tree height H (m), basal area for the dominant and secondary layers G1 and G2 (m 2 /ha), stem volume for the dominant and secondary layer M1 and M2 (m 3 /ha), stem volume increment Z v (m 3 /ha/yr), site type, etc. have been recorded for every stand. The minimal area of a stand is 0.1 ha according to the requirements by national forest inventory.”



Reviewer 2: “Line 99: How the stray light correction is made? Might be of interest to the reader.”


Reply: In the section ' Material and Methods', we wrote "Stray light was corrected with deconvolution method proposed by Kostkowski [16] for spectral instruments. Instrument function of the spectral sensor was characterised using a double monochromator as described in Kuusk et al. [17]."



Reviewer 2: “Line 104 to 105: Were the areas corresponding to the sensor footprint extracted/located in the field for acquiring the ground truth or the whole stand was considered as the ‘measured’ area? Variation between sampling density and location might have had considerable impact in the outputs. How the authors have tackled this aspect during the analysis?”


Reply: We have added morethorough explanation of the meaning of “stand” in forest inventory database in the section of ‘Material and Methods'. All spectral recordings within one stand were averaged. Stand is the standard unit of forest inventories. One stand is one record in national forest inventory database and it means a homogeneous forest patch.



Reviewer 2: “Figure 2: It would be clearer if points were added to the curves indicating the location of the actual spectral bands in all sub-figures. This would allow the reader to better appreciate the resolution of the data acquired.”


Reply: We added to the figure 2 caption that spectral resolution (FWHM) was 10 nm and spectral sampling interval 3.3 nm



Reviewer 2: “Line 111: The fact that simple ratio and normalized formulations resulted in a near linear relationship might be related to the scale / method of spectral measurement. Since the main advantage of a normalized index is the compensation for changes in illumination conditions if the measurements with which the indices are calculate correspond to a relatively large surface and if several observations are averaged the difference between normalized and simple ratio tend to decrease, I suppose. Better to indicate / check that since this observation might not hold in other situations.”


Reply: Radiance depends on illumination but we are using reflectance. We pointed out both SR and NDVI formulation so that other researchers could check if the near-linear relationship at those wavelengths holds in other situations as well. We have seen that this relationship holds both at leaf level (multiple species) and forest canopy level.



Reviewer 2: “Line 113: Why these specific formulations were chosen? For example a formulation with 705 and 750 nm instead of 705 and 783 nm is more frequently used in other articles. Please indicate the articles or  empirical evidence for the use of these indices and band combinations.”


Reply: Based on the mechanistic background the reference wavelength should be at NIR plateau in such spectral region where the reflectance is not changing with wavelengths (i.e. scattering not absorption features). NDVI705 has only one sensitive band at 705 nm (for tracking pigment absorption) and the stable region at NIR plateau should be used as reference band based on the physical principles. 750 nm is still very close to the red edge of NIR plateau as strong correlation with pigments content can be observed up to 735 nm. For example in the case of cheaper or older sensors there can be practical need to use as short wavelengths as possible (750 nm) for reference at NIR plateau due to the increasing noise and problems with sensitivity at bigger wavelengths. If there are no technical limitations then longer wavelengths in the middle of NIR plateau should be preferred. For the sake of uniformity we used in the current study the same waveband (783 nm) which represented NIR plateau in S2REP also in NDVI665 and NDVI705 as NIR reference band. We have added more thorough explanation about vegetation indices to the introduction section.



Reviewer 2: “Line 116: Maximal change or inflection point? Please clarify.”

Reply: Red-edge inflection point is mathematically defined as the location of maximal change i.e. the location of the maximum of the first derivative.



Reviewer 2: “Line 124: Please correct the spelling of ‘exact’.”

Reply: Done



Reviewer 2: “Line 124 to 129: The authors describe the method used to approximate the fluorescence signal from relatively coarse spectral information. Is this methodology describe anywhere else (please provide source)? If this is not the case, there has been a laboratory-based or proximal sensing evaluation to verify if this hypothesis hold?”

Reply: We have included more thorough explanation. Methodology is based on common Fraunhofer line discrimination principles with small adjustments described (fitting the curve instead of straight reference line and calculating area).



Reviewer 2: “Line 133: It is important to point out the fact that other methods for red-edge inflection point calculation exists (e.g.: 10.1016/j.rse.2005.12.011). It is also extensively reported in the literature that a double peak effect is observed for the reflectance in the red-edge region, which can be attributed to changes in fluoresce emissions that follows alterations in vegetation photosynthetic capacity, as far I understand (10.1109/IGARSS.2003.1293856). Therefore, these results are already expected and no mention to previous literature about this topic has been made so far (in particular in the introduction). I would recommend to take this factor into account since one of the main objectives of the article are to describe the relationship between the red-edge inflection point, SIF and vegetation properties.”


Reply: We have added the suggested citations and explanation about the fluorescence double-peak into the section of introduction. The wavelength regions of peak jump in forest canopy in our study do not match with this fluorescence double-peak feature. The shoulder visible at 700 nm in fig. 2B could be caused by the first fluorescence peak but this peak was higher only in non-forested patches which were removed from our analyses in the early stage. We had 3 peaks in total and the jump from the first peak (700 nm) which occurred only in non-forest (herbaceous) patches to the other peaks (717 nm and 727 nm) could match with the wavelengths of the double-peak effect presented in 10.1109/IGARSS.2003.1293856 but not the jump between the second and third peak at 717 nm and 727 nm. The structural and physiological mechanisms which cause the precise shape of red-edge spectral features should be definitely further studied. Although we removed non-forested patches from the analyses as the current study deals only with forest canopy we have felt necessary to point out the existence of 700 nm peak to provide the correct context for the reported peaks at 717 nm and 727 nm.



Reviewer 2: “Line 139 to 141: If this discontinuity has been well reported in previous articles, should this evaluation be included in the analysis? Other methods with better performance have been proposed in the literature. It might be of interest to include a justification for the use of such approach, and why other methods were not evaluated.”

Reply: We have explained that we selected linear four point interpolation method (S2REP) because of this method can be also used for calculations at satellite level from Sentinel-2 MSI data and we compared S2REP with the original mathematical definition of red-edge inflection point (location of the maximum of the first derivative). We tried also different parameters such as derivative peak amplitudes and peak ratios but these did not add new information to S2REP in our dataset and hence were not reported.



Reviewer 2: “Line 143 to 144: Might this be an indication that the spectral information acquired lacked resolution for the implementation of such analysis? Was any smoothing applied before the derivative calculation? It is important to explain why this happened.”

Reply:We can explain what does not cause this phenomenon. UAVSpec3 has 3.3 nm sampling interval, thus, there are two pixels between 717 nm and 727 nm and the two-modal result can not be explained only by the low spectral sampling interval of the spectrometer. It can not be explained with noise either. We use in this study averaged reflectance spectra per stand. For example over RAMI stands the number of recorded spectra reaches 1300 and we see the same effect as in other stands with 10 to 130 spectral recordings averaged. Noise levels were reduced with averagingmultiple spectral recordings not smoothing and therefore smoothing has not affected the shape of actual small spectral features.



Reviewer 2: “Line 147 to 150: Since this information is already present in the literature is this evaluation necessary?”


Reply: The precise shape of the red-edge region from higher resolution spectral data could give much more information than simply one location of the red-edge inflection point and deserves to be studied much more intensively in future.



Reviewer 2: “Line 158: Table 1 should be placed near this citation to make reference easier.”


Reply: Done



Reviewer 2: “Line 159 to 164: Tests with the same or very close spectral wavelengths have been performed previously in other studies (e.g.: 10.1016/1011-1344(93)06963-4, Figure 4). Please indicate the added value of presenting such comparison. Rather focus on aspects that have not been tackled before. Please make these aspects more explicit while explaining the motivation of the study.”

Reply: We selected two most classical wavelengths (665 nm and 705 nm) to represent the comparison of NDVI formulations near chlorophyll absorption maximum and away from chlorophyll absorption maximum. We have improved the wording of the introduction to explain the motivations of the study better.



Reviewer 2: “Line 174 to 176: Maybe it is important to indicate that the correlation is negative between vegetation indices (VIs) and reflectance in the red region. This signals that while values of the VIs increased chlorophyll content probably decreased and canopy structure increased. This is also in agreement with SIF values which are higher for higher canopy structure levels (higher values of reflectance in the NIR) since with larger canopy structure the photosynthetic capacity potentially increased.”

Reply: Forest height (H), stemwood volume (M1) and basal area (G1) show the amount of structural elements in the forest canopy. Actually we see that the reflectance in the NIR region is reduced when the amount of structural elements is increasing. The relationship between structural parameters (G1, M1, H) and NIR reflectance is always negative in Table 3. The most likely physical explanation is the increased shadow fraction in FoV of the spectrometer. After canopy closure of young forest its brightness in NIR will start to decline as trees grow bigger and older. Hemiboreal forest canopy becomes less uniform as trees grow bigger and older. This process includes self-thinning for example. As forest ages and grows higher the changes in shadow fraction are most likely governing all the relationships we see between spectral variables in the current study. (Our datset contains 300 forest stands with different ages between 15 to 230 years.)


In our dataset, vegetation indices (except PRI) have positive correlation with NIR reflectance and negative with visible reflectance (Table 1) but the parameters of canopy structure from forest inventory have negative correlation with both NIR and visible reflectance (Table 3).



Reviewer 2: “Line 186 to 187: Please provide references for this affirmation.”

Reply: Done



Reviewer 2: “Line 187 to 189: This affirmation might hold only for the factors varying in this specific dataset. Since measurements were made in a single date, factors affecting the variability of the VIs values might have been restricted to a narrow process (e.g.: increase in canopy structure followed by decrease in leaf chlorophyll content). For a more in depth evaluation of such hypothesis analysis based on radiative transfer modelling could be used to evaluate the behaviour of these formulations under a broader conditions variation. Otherwise, the authors should search in the literature for similar cases in which this behaviour is observed.”

Reply: We have improved the wording. Our aim was to explain why we did not use NDVI form in the current study (not to make generalization for all other situations).



Reviewer 2: “Line 194 to 197: Were these parameters described in the material and methods? Please provide reference for them and move this part to the material and methods if not included there yet.”

Reply: Done



Reviewer 2: “Line 197 to 199: This might be related to the changes observed in this specific dataset. Apparently as vegetation structure increases chlorophyll content decreases which makes reflectance in the visible (related to plant pigments) and NIR wavelengths (related to vegetation structural traits) change in a similar pattern (e.g.: higher reflectance values in the NIR region are followed by higher values in the red or green wavelengths due to lower content of leaf chlorophyll in the leaves, according to Table 3; maybe correlation between single bands should be added to better illustrate this point). It might be of interest to point it out.”

Reply: Forest canopy functions somewhat differently. In agricultural crops or sparse vegetation reflectance in NIR region increases as the amount of biomass increases above soil. In hemiboreal forests with canopy closure forest brightness in NIR starts to decrease as forest grows higher and biomass increases. We have added more discussion about it.

Reviewer 2: “Line 202 to 203: The negative correlation between LAI and reflectance in the red region indicates that structure gains were potentially inversely related to leaf chlorophyll contents. Better to indicate this to the reader. Also, including the correlation between traits might help the interpretation.”

Reply: Forest inventory data does not include leaf chlorophyll and forest canopy itself includes multiple different layers of vegetation from trees to shade tolerant understory herbs and mosses. Trait correlation network in forest canopy is extremely complex.



Reviewer 2: “Figure 5. Simplify the y-axis title. Include regressions coefficient (R2) and/or other accuracy indicators (RMSE, linear equation between x and y variables).”

Reply: Done



Reviewer 2: “Table 3. The title can be simplified”

Reply: We have simplified the title.



Reviewer 2: “Figure 6. Same as Figure 5. Include accuracy / performance indicators (R2, RMSE, etc.).”

Reply: Done



Reviewer 2: “Line 212 to 213: Although this might be true, it has already been observed in other articles that this measure is not suitable for vegetation monitoring due to the double peak effect in the red-edge region. Therefore, the inclusion of this parameter in the analysis is questionable.”

Reply: It is included as the mathematical definition of red edge inflection point.



Reviewer 2: “Line 223 to 226: This is expected and is well described in the literature. Also, the spectrometer used might lack resolution (i.e., band width of approximately 4 nm) for detailed analysis of this aspect. The authors could focus instead on other factors not well described so far or they should make some effort to evaluate LAI based on the spectral information, which could help the interpretation of the data (i.e.: decoupling effects of canopy structure and leaf pigments concentration).”

Reply: The precise shape of the red-edge region from higher resolution spectral data could give much more information than simply one location of the red-edge inflection point and deserves to be studied much more intensively in future. We have added more explanation about LAI and canopy structure in hemiboreal forests of different ages. We do not have corresponding leaf pigments data for these 300 stand.



Reviewer 2: “Line 227 to 229: The Double peak effect should be related to SIF in the far red region. As observed for SIF, the increase in canopy structure increased values of fluorescence emitted (positive correlation with NIR wavelengths) and this might have been reflected in the maximum value of the first derivative as well. Besides, canopy structure itself affects largely the reflectance in the red edge region and therefore is expected that growth form is related to the double peak effect. It might be of interest to point this out. Please include citations about this topic, which is extensively reported in the literature.”

Reply: The wavelength regions of peak jump in forest canopy in our study do not match with this fluorescence double-peak feature. The shoulder visible at 700 nm in fig. 2B could be caused by the first fluorescence peak but this peak was higher only in non-forested patches which were removed from our analyses in the early stage. We had 3 peaks in total and the jump from the first peak (700 nm) which occurred only in non-forest (herbaceous) patches to the other peaks (717 nm and 727 nm) could match with the wavelengths of the double-peak effect presented in 10.1109/IGARSS.2003.1293856 but not the jump between the second and the third peak at 717 nm and 727 nm. The structural and physiological mechanisms which cause the precise shape of red-edge spectral features should be definitely further studied. Although we removed non-forested patches from the analyses as the current study deals only with forest canopy we have felt necessary to point out the existence of 700 nm peak to provide the correct context for the reported peaks at 717 nm and 727 nm.



Reviewer 2: “Line 229 to 233: It might be an indicative that the spectral resolution of the spectroradiometer used is not adequate for this analysis.”

Reply: UAVSpec3 has 3.3 nm sampling interval, thus, there are two pixels between 717 nm and 727 nm and the two-modal result can not be explained only by the low spectral sampling interval of the spectrometer.



Reviewer 2: “Line 233 to 234: These studies already exists as far as I understand. Please include literature about the topic, previous studies cannot be ignored.”


Reply: We have added the citation.



Reviewer 2: “Line 238 to 245: This is an adequate observation, computation of the first derivative might also have been deeply affected by the limited resolution of the spectrometer used.”

Reply: Higher spectral resolution would allow even better characterisation of the precise shape of red-edge spectral region.



Reviewer 2: “Line 248: remove ‘these’ on ‘these vegetation indices’.”


Reply: Done



Reviewer 2: “Line 251 to 253: Any particular reason / explanation for that? Very important for the interpretation of the results. Can it be that PRI expressed better the current vegetation photosynthetic capacity while Farea was also affected by canopy structure components and was less effective for this purpose? Please try to explain.”

Reply: Both canopy level PRI and Farea must be affected by canopy structure and the effect of structural arrangement of canopy elements is most likely different for PRI and Farea. We have added more explanation.



Reviewer 2: “Line 272 to 273: Was this component (shadow) representative in the study sites during the measurements? Was this effect estimated for the areas measured? It cannot be that chlorophyll content and canopy structure presented such a pattern that increase in biomass resulted in negative correlation with indices due to reduction on chlorophyll content at leaf level? Analysis of VIs designed for canopy structure estimation (MCARI2, WDVI, etc.) might help with the interpretation of the data. The analysis of the NIR region alone (single bands) might be more susceptible to other factors than canopy structure.”

Reply: MCARI2 (Modified Chlorophyll Absorption Reflectance Index 2) and WDVI (Weighted Difference Vegetation Index) are both specifically designed for those vegetation types where soil background optical properties have crucial influence on reflectance signal measured above the vegetation and therefore soil adjustment would be needed (e.g. crops). However, such approach would not be appropriate in the case of multi-layered closed forest canopies, where background signal can originate from lower layers of trees, herbaceous understory or moss-layer. We have added more thorough explanation about the relationship between shadows and canopy structure in forest.



Reviewer 2: “Line 279 to 283: Was this fact related to shadow effects according to the authors cited? This counterintuitive result (negative relationship between biomass and NIR reflectance) must be better explained / discussed taking into account all factors affecting the reflectance in this region.”

Reply: Yes, the negative relationship between biomass/height and NIR reflectance has been previously related to shadow effects by other authors as well including also model calculations. We have added more thorough discussion about the shadow effect on forest age and NIR reflectance relations.



Reviewer 2: “Line 298 to 301: Evaluation of VIs related to canopy structure may provide a better insight in the entangled effect of canopy structure, pigments concentration and current vegetation photosynthetic capacity.”

Reply: VIs developed to account for canopy structure in agricultural crops based on soil line are not suitable for characterising forest canopy structure. Forest canopy is very heterogeneous with multiple layers of canopy with different species and strong vertical gradient in leaf traits due to the light acclimation of foliar traits along vertical light availability gradient within the canopy.



Reviewer 2: “Line 303 to 306: Despite the narrow band implementation using bands located in a very sensitive spectral region, simple ratio indices are generally very sensitive to illumination conditions, amongst other factors which are considered in more complex formulations. Has this or other factors affected the outputs obtained? How would this formulation behave in a more broad dataset (multi-temporal, etc.)?”

Reply: Radiance depends on illumination but we are using reflectance. We pointed out both SR and NDVI formulation so that other researchers could check if the near-linear relationship at those wavelengths holds in other situations as well. We have seen that this relationship holds both at leaf level (multiple species) and forest canopy level. We have added more thorough explanation.

Reviewer 2: “Line 328 to 335: This aspect has been already discussed in previous articles, as far as I understand. I would recommend to focus rather on explaining the negative correlation between reflectance values in individual bands and biomass, for example, or other aspects not well discussed so far (indicated in the previous comments).”

Reply: We have added more thorough explanation about the forest height, age and shadows effects on forest canopy optical properties.

Reviewer 2: “Line 336 to 348: It would be interesting to see some considerations regarding future applications of the results obtained in this study, as well recommendations for further exploration of aspects not well explained using the current methodological framework / dataset.”

Reply:We would recommend further exploration of the precise shape of the red-edge region with higher spectral resolution. Current trend is towards the increase of the availability of very-high-resolution spectral data collected for SIF. Such information will be soon available even from the satellite level from ESA FLEX mission (scheduled launch in 2023). The number of peaks and the mechanism of peak jumps in red-edge derivative spectrum (particularly the interaction between canopy structure and physiological processes)should be elucidated considering the possible space-borne application in future.



Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors,

thank you very much for your manuscript. The topic is very interesting, but the quality of the present version is low. You should:

the Introduction is significantly too general, you have to add much more details oriented on theoretical background of your topic.

the Methods aren't properly presented. I tried to follow your research algorithm, but it isn't possible. Please, add a research schema, then please describe all steps. You should add a part oriented on statistical analyses, verification of model, accuracy assessment.

results aren't linked together, there are sets of different procedures, which you conducted.

Discussion should present a real comparison of your results and achievements acquired by other researchers. You should indicate positive elements of your methods.

Much more details you can find in the attached manuscript.

All the best for you

Reviewer

Comments for author File: Comments.pdf

Author Response

Thank you for reviewing our manuscript!


Reviewer 3: “The Introduction is significantly too general, you should present much more details on current solutions of your topic. How your topic is soleved out by other researchers?”


Reply: We have improved the introduction.


Reviewer 3: “the Methods aren't properly presented. I tried to follow your research algorithm, but it isn't possible. Please, add a research schema, then please describe all  steps. Please, add much more details how you acquired and processed data? How many did you have research polygons?”


Reply: We have added schema of data processing steps. Final dataset contained 300 research polygons. Each research polygon corresponds to one record in National Forest Inventory database.



8 spectra were registered per second. Considering the flight speed 17 m/s the spectral recordings would be by 2.1 m step. Field-of-view (FOV) on ground was about 2.5 - 3 m. Averaging was made later at stand level. We used 8 m buffer zone to avoid spectral mixing with nearby area. Spectral recordings closer than 8 m to the stand border were rejected and stands with less than 10 spectral recordings (after buffering) were rejected from the analysis. All spectral recordings within one stand were averaged. Stand is the standard unit of forest inventories. One stand is one record in forest inventory database and it means a homogeneous forest patch. We have also added more thorough explanation of the meaning of “stand” in forest inventory database.



Reviewer 3: “You need to add a subchapter oriented on accuracy assessment”


Reply:We have added more thorough explanation into Methods section. Measurements were carried out in stable illumination conditions, the calibration errors of spectrometers were reduced by using ratio of spectrometer signals in Eq. (1). The uncertainty of the reference calibration r_lambda is less than 0.005 for the spectral range 300-2200 nm.


Reviewer 3: “Statistical analyses should be presented in Methods. Why did you select the Spearman rank order correlation?”


Reply: Spearman rank order correlation was used for assessing the strength and direction of association between variables as some pairwise relationships were non-linear. The significance level on 0.05 was used.


Reviewer 3: “You should indicate positive elements of your methods.”


Reply: For top-of canopy spectral reflectance simultaneous measurements of irradiance spectra and in case of high-flying airborne measurements atmospheric correction are needed. Our study is focused on intermediate scale of so called bottom-of-atmosphere (BOA) / top-of-canopy (TOC) measurements above mixed forest canopy. Such measurements are rarely conducted above forests due to the technical difficulties. Most of other research has been conducted on agricultural crops. Simultaneous recording of irradiance spectra to convert airborne data to TOC spectral directional reflectance is the strength of our data. Airborne imagers may provide at-sensor radiance data or even non-calibrated digital numbers. In our study low-flying airborne spectral data are supported by simultaneous recording of irradiance spectra to convert airborne data to TOC spectral directional reflectance, and by forestry data provided by the National forestry database.


Reviewer 3: “results aren't linked together, there are sets of different procedures, which you conducted. Discussion should present a real comparison of your results and achievements  acquired by other researchers. You should prepare a table to compare your results and achievements of other researchers.”


Reply:We have added the requested table, expanded discussion and linked results together in conclusions.


Reviewer 3: “You can add some more details of references, to show an impact of cited papers to the text”


Reply:Done


Reviewer 3: “references of all indices are needed”


Reply: Done.


Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear Author(s),


You have made alterations in the manuscript considering the suggestions given in the previous round of the review process. In my opinion these changes contributed to improve the quality of the material presented. However some points would require further adjustments as indicated the file attached. In particular, the double peak effect described by the authors in not the same conventionally observed in the red-edge region. This fact might cause confusion and this should be clearly indicated throughout the text. Besides, a possible explanation for the mechanism behind this observation should be provided. Have other studied focusing on canopy level data presented similar results? This is not completely answered by the authors and citations should be provided indicating if (a) this is a new observation (and what are the possible caused for that) or (b) if this effect has been observed before (and what were the causes indicated by the authors in this case). Please check the document attached for more details on the corrections / alterations still required in my opinion.


Best regards,

Comments for author File: Comments.pdf

Author Response

Thank you for reviewing and correcting our manuscript!


>chlorophyll-a?


Reply: We added also an additional reference (Porcar-Castell et al 2014)


> Change the order of this and the previous phrases. Adding 'In addition,

> variations...' might help to connect the text.


Reply: Done


> use instead '...is to compare...'


Reply: Done


> replace by '...in the...'?


Reply: Done


> replace by: '...with reflectance in..'


Reply: Done


> Add reference to these vegetation indices if possible.


Reply: Done


> Better examples of UAV-based sensing application in forestry can be

> found I suppose.


Reply: Most of UAV-based practical application in forestry are just for visual inspection by human eye not for numerical calculations.


> And the forest traits? Are those not part of the main objectives?


Reply: We improved the wording "and forest inventory variables"


> replace '...inter-relationships of...' by 'inter-relationships between..'


Reply: Done


> replace by '...using measurements with...'


Reply: Done


> replace by: ' , '


Reply: Done


> '... of the number of recorded spectra...'


Reply: Done


> replace by: '... varied in general between ...'


Reply: Done


> were


Reply: Done


> replace by: '...a single formulation in each case:'


Reply: Edited


> Figure 3: Lines indicating bands position in (a) and (b) are not

> matching. Difficult to see regular graph divisions and lines indicating

> specific wavelengths (change color / line width for better

> visualization).


Reply: We have edited the image for better visualization. Vertical lines denote in (a) spectral bands selected for further analysis in current study as single band reflectance factors and in (b) the location of peak maximum for red edge inflection point around 700, 717 and 727 nm.


> If this approach is proposed by the authors (first time implemented)

> please indicate, otherwise provide citations of articles containing the

> original idea / articles in which the current implementation is based.


Reply: We have improved the wording. Difference between apparent reflectance and continuum is in the interval of 752-765 nm due to the spectral resolution of our spectrometers. Integration over bigger interval would not change resulting values.


> Add citation of works demonstrating these aspects.


Reply: Done


> replace by: '... the apparent reflectance factor attributed to a

> spectral band center...'

> replace by: '... correspond to the...'


Reply: Due to low spectral resolution of the spectrometer the peak is spread over a wider area and therefore we use peak area instead of peak height. We have edited the wording to improve the clarity.


> '... of the...'


Reply: Done


> replace by 'slightly'


Reply: Done


> replace by: '...two spectral measurements / bands...


Reply: Done


> This effect is the same double peak described in the paper by Alonso et

> al. [14].


Reply:Yes, we tell here that similar effect has been also observed at leaf level.


> In general, the effects described as 'double peak' by the authors are in

> fact concentrated in a very narrow region of the spectra and are not

> related to the conventional 'double peak' effect observed due to the

> interaction between pigments reflectance and fluorescence. As indicated

> by the authors it might be a residual effect of canopy structural

> properties / forest understory. Also characteristics of the sensor used

> could result in effects of this type. Please try to provide a possible

> mechanism that could result in the current observations. Was that

> observed before?


Reply: We have cited several earlier works. Contradicting opinions exist about the possible mechanism as it has been explained either simply by the effect of the amount of chlorophyll (le Maire et al 2004) or as the result of photochemistry and chlorophyll fluorescence (Alonso et al 2003). We can show confidently that the phenomenon exists but we could only add more speculations about the mechanism.


> Are you sure it was never observed before? Are you not able to provide a

> possible explanation?


Reply:For example Kochubey & Kazantsev 2007 have found at leaf level measurements of agricultural crop also three peaks at similar wavelength as we observed at canopy level over forests. We would be able to provide speculation about the possible mechanism rather than explanation. Laboratory analysis of fine-scale spectral features of different pigment-protein complexes could maybe help to provide explanation.


> And?...Were not worth to consider?


Reply: Comparison of all different curve fitting and extrapolation methods would be a topic for a whole independent review article. Different approaches for producing a single location of REIP have been used and developed over a very long time period. We considered S2REP currently the most influential approach due to Sentinel-2.


> corresponds to the


Reply: Done


> discrete

> locations


Reply: Done


> Are you sure that this has never been observed before? Please make sure

> the literature was thoroughly searched.


Reply:We have shown that there are some previous studies which confirm the existence of multiple peaks but more research to elucidate the mechanism would be needed in order to make proper use of the information that multiple peaks could give. Maybe the relative share of some core and light-harvesting complexes could also influence this region but it would be just a speculation. Pigment-protein complexes govern the precise absorption spectra of leaves. More detailed laboratory studies would be needed.


We are very grateful for the reviewer for the useful comments and suggestions which have helped to improve our manuscript significantly!

p { margin-bottom: 0.1in; line-height: 120%; }


Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors,

thank you very much for your proper revision of the manuscript, it looks much better, but there are few small parts which could be better presented.

I marked them in the text, so please, look at the attached manuscript.

Best wishes

Reviewer

Comments for author File: Comments.pdf

Author Response

Thank you for reviewing our manuscript!


>  It is a citation train, please indicate an impact of references.


Reply: Done


> Please, add full names when you write first time a name.


Reply: Done (we added full names and citations)


> Please add details of the manufacturer.


Reply:We added the citation to website


> Please, add more details, e.g. why Spearman? statistical tests? Please add more details on validation and an accuracy assessment of the models


Reply:All statistical tests were performed with R version 3.4.4 (2018-03-15). Spearman rank order correlation was used because of some pairwise relationships were (in various ways) non-linear.

p { margin-bottom: 0.1in; line-height: 120%; }


Author Response File: Author Response.pdf

Back to TopTop