the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Apr 2025
Using GNSS-based vegetation optical depth, tree sway motion, and eddy-covariance to examine evaporation of canopy-intercepted rainfall in a subalpine forest
Abstract. Recent advances in the measurement of water content within a forest, have led to new possibilities to study canopy evaporation. We used a pair of Global Navigation Satellite System GNSS receivers (one above the canopy and one near the forest floor) to calculate the vegetation optical depth VOD during the warm season in a Colorado subalpine forest. One goal in our study was to compare VOD to the concurrent tree sway motion and subcanopy/above-canopy eddy-covariance evapotranspiration ET measurements. We found that VOD increased and tree sway frequency decreased during wet periods; furthermore, both measurements exhibited a linear relationship between each other and suggested that it took around 14 h after rainfall ceased for the intercepted rainwater to fully evaporate from the canopy. On dry days, we found that tree sway was more sensitive to diel changes in internal tree-water content than VOD. The ET measurements provided quantitative estimates of canopy evaporation (0.02 mm h−1 at night, to 0.08 mm h−1 during mid-day). Following rainfall, nighttime VOD, tree sway and ET all showed a steady (nearly constant) drying of the canopy. Variability in the VOD and tree sway measurements, comparisons with water content from the CLM4.5 land-surface model, and challenges with ET measurements, are also discussed.
- Preprint
(3080 KB) - Metadata XML
-
Supplement
(6408 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2025-1755', Anonymous Referee #1, 13 May 2025
Burns et al. measured warm season ET, VOD and tree sway in Colorado subalpine forest to was to track evaporation of intercepted water. They found that a canopy intercepted water can take a tremendous amount of time to evaporate and thus can contribute to ET at the site especially on dry days after a rain event. This work well describes VOD and tree sway as a method of measuring evaporation of canopy interception.Comments:To help distinguish these VOD measurements from more common measurements excluding, I suggest to specify in the acronym that this is VOD plus intercepted water. For example, VOD underscore "int" to distinguish from studies where VOD refers to water only within the tissues.line 66: Please clarify what the "small pockets" are referring to.line 75: Parentheses missing around GNSSline 95: The inclusion of the CLM analysis while exciting comes up abruptly here and the importance of this analysis could be integrated sooner. Suggest to reference Burns et al 2018 or specific the concepts from that paper that will be expanded upon here.line 163: Suggest to replace "use the concept" with "assume"Figure 3, 8: It would be helpful to please clarify in the results why there is not diurnal cycle of VOD where VOD peaks in the morning and decreases into the afternoon even on the dry days. This is common finding in Holtzman et al. 2021 Biogeosciences and Yao et al. 2024 Geophysical Research Letters. Later on VOD does increase due to rain but do the trees not dehydrate when transpiring through the day?336-337: Please describe the B0, G1, A1, and F2 cases here or in the methods sections for reference to the reader.419: This is does not seem like a novel finding given your reference to the common method of removing these data in the introduction. Please clarify the novelty here.477: suggest to replace leaf with needles for clarity.Citation: https://doi.org/
10.5194/egusphere-2025-1755-RC1 - AC1: 'Reply on RC1', Sean P. Burns, 16 May 2025
-
RC2: 'Comment on egusphere-2025-1755', Anonymous Referee #2, 22 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1755/egusphere-2025-1755-RC2-supplement.pdf
- AC2: 'Reply on RC2', Sean P. Burns, 30 May 2025
-
RC3: 'Comment on egusphere-2025-1755', Anonymous Referee #3, 27 May 2025
This paper compiles VOD, tree sway frequency, and flux tower data for a subalpine forest. Additional data include output from a land-surface model, and field/sensor data related to air temperature, bole temperature, bole apparent dielectric permittivity, wetness, and precipitation. Together, these data are used to suggest that VOD and tree sway frequency capture signals related to canopy evaporation. These measurements offer unique insight into canopy storage dynamics from completely independent observations. I don’t have any major suggestions/comments for the authors. The following suggestions are meant to help the reader understand additional context for the tree sway measurements, interpretation of results, and to facilitate reproducibility/follow up studies.
Main Comments
- I agree with Referee #2 in that parts of the tree sway methodology can be expanded, even briefly. Suggested expansions:
- A) Line 145-146. If possible, include a brief description of the accelerometer methods, even though they are included in Raleigh 2022. Can go in the appendix, along with the other measurement expanded details if authors see fit.
- B) Line 147. Why was a 30-minute window chosen? Were other windows tried but this was the best one for the analysis?
- C) Line 148. What interpolation and smoothing function(s) were used?
- D) Line 156-157. How was the value chosen to be 1? Was this done by offsetting the low frequency trend such that the average frequency over the time series was 1? If so, please include this.
- Similarly, Line 138, can there be a brief inclusion of the data processing for VOD? Even if it is a brief summary, it would be helpful. This can be in the appendix.
- Table 1. Is it possible to add the original/raw frequency of measurements for each observation?
- Figure 2. Is the accelerometer associated with Pine 3 or Pine 4? Or are the dielectric permittivity sensor on two different trees? Please clarify.
- Line 347-349. Is it possible to expand on this (how these empirical relationships might facilitate changes in the land-surface model)? This seems like an important consideration, however there is nuance to it. In particular for tree sway frequency, the tree sway frequency is proportional to the inverse square root of tree mass, so while the frequency is observed to be linear, this does not mean that the tree mass (interception) is changing linearly. However, there is no mention of how tree sway frequency relates to mass in a mathematical form throughout the manuscript. I suggest adding this here and/or earlier on in the manuscript so if improvements on the land-surface model can begin there’s the additional context of how physically sway frequency relates to changing tree mass due to interception.
- This is not exceptionally important, and the authors can ignore: I was trying to determine any sort of relative magnitude differences in responses for the precipitation events presented in table 3 and figure 8 (and 9). I tried to find the lowest precipitation event (3.6 mm, solid cyan line) and the highest precipitation value (29 mm, solid red line). I think I found them in the sway frequency plot and they had the lowest and highest range/variability, which was helpful to see that with a first order approximation that the observations aligned with physical interpretations (more precipitation --> more change in tree sway frequency). I couldn’t quite find them in the VOD diagram. Highlighting these two events with bolded lines might help the reader understand the relative sensitivity to single events and strengthen the connections observed between precipitation, evaporation, and VOD or sway frequency.
- Are the authors amenable to including in the acknowledgements or in brief ‘open science’ section in the appendix or supplemental info where the data and code used throughout the manuscript can be found? This would help future studies reproduce these results and follow up studies that are interested in conducting similar experiments elsewhere.
Citation: https://doi.org/10.5194/egusphere-2025-1755-RC3 - AC3: 'Reply on RC3', Sean P. Burns, 05 Jun 2025
- I agree with Referee #2 in that parts of the tree sway methodology can be expanded, even briefly. Suggested expansions:
-
RC4: 'Comment on egusphere-2025-1755', Anonymous Referee #4, 01 Jun 2025
This research investigates promising and novel techniques to predict rainfall interception and, thus indirectly, canopy evaporation. The authors use L-band microwave active microwave attenuation data (vegetation optical depth) from a GNSS doublet and tree sway data measured by a an accelerometer placed on the trunk of a tree canopy. The study is set in an subalpine, high elevation needleleaf forest in in Colorado/USA. The authors demonstrate the ability of both proxies to correlate with onset and drydown of precipiation events, evapotranspiration and modeled interception storage from different land surface model (CLM4.5) parametrizations. The data sets and analysis presented by the authors allow for the conclusion that these techniques are promising tools to measure interception storage and that they hold potential to supplement/validate land surface models that are known to have high uncertainties in interception fluxes the their parametrization of the canopy, and uncertainties in EC water flux measurements during rain events. This study is of great quality. However, the authors should address the comments below before publication.
Major comment
Please elaborate on the robustness of tree sway motion being able to represent interception storage without the need to account for wind speed as a possible confounding factor. In this context, it would be valuable to find sway motion data as a function of wind speed – e.g. in fig. 10 and at least in one of the plot over time – to clarify on this relationship and include this missing piece of information.
Minor comments
- Fig 1
- Please report the the inner circle radius (r=20m) as the authors have done for the outer circle
- Please point the reader of fig. 1’s caption to what the different footprint circles represent to better understand the results, i.e. what is the main take-away from the inner circle radius (apart from GNSS paucity visualization).
- B 527: Please clarify the role of r=20 in this study. Did the authors clip the radius so the “several tall trees” are not included in the footprint?
- B & S3 (GNSS sky view):
- Please show the elevation angle and cutoff to clarify which parts of the canopy will effectively be used for VOD calculation, especially elevation=10°
- Please clarify the rationale behind clipping out another area in NE, close to the northern GNSS gap
- You use a very low cutoff elevation angle of 10°. Looking on fig. S3 – assuming the outer two circles being roughly within θ in (10, 30] – only very low VOD can be found that do not display any pattern expected from forest attenuation and possibly fail to represent true forest VOD. Under this light, please explain why the authors used a cutoff=10°.
- Since the Lambert-Beer angle correction assumes a homogeneous canopy and ignores multipath scattering, any losses observed may be due to scattering caused by multiple layers of vegetation, causing Lambert-Beer to break at low angles. Hence, the referee suggests using a higher cutoff elevation angle (~30°) or would value a discussion why low VOD at lower angles will not affect the overall results. Consider page 12 in Camps et al. (2020) about this question: “Note, however, that only at high elevation angles (elevation angle > 67.5°) is the single scattering albedo correlated with the NDVI, and at lower elevation angles, the presence of multiple scattering makes the tau-omega model [all zeroth order assumption, incl. lambert-beer, ~the referee] more likely to be invalid.”
- 135: Which GNSS frequency is used, please indicate the frequency(ies) in section 2.2.1 since GNSS VOD offers a range of bands to choose from.
- The authors detrend sway motion to alleviate effects of temperature and vegetation water content on short-term changes. However, VOD is also affected by long-term changes in biomass, and vegetation water content. Why did the authos not consider detrending VOD, especially since a trend is visible in fig. 2? This is worth noting in 2.2.1.
- 433/4: The size of the EC footprint has not been explicitly mentioned in the text. Also, which footprint size of VOD as your referring to in this statement? To make a statement about the footprint size (a very relevant discussion) the referee suggests to state that although the footprint sizes between all technique partly or greatly differed, the good correlations could be found etc.
Citation: https://doi.org/10.5194/egusphere-2025-1755-RC4 - AC4: 'Reply on RC4', Sean P. Burns, 06 Jun 2025
- AC5: 'Reply on RC4', Sean P. Burns, 27 Jun 2025
- Fig 1
- AC6: 'Comment on egusphere-2025-1755 (List of proposed manuscript changes)', Sean P. Burns, 27 Jun 2025
Status: closed
-
RC1: 'Comment on egusphere-2025-1755', Anonymous Referee #1, 13 May 2025
Burns et al. measured warm season ET, VOD and tree sway in Colorado subalpine forest to was to track evaporation of intercepted water. They found that a canopy intercepted water can take a tremendous amount of time to evaporate and thus can contribute to ET at the site especially on dry days after a rain event. This work well describes VOD and tree sway as a method of measuring evaporation of canopy interception.Comments:To help distinguish these VOD measurements from more common measurements excluding, I suggest to specify in the acronym that this is VOD plus intercepted water. For example, VOD underscore "int" to distinguish from studies where VOD refers to water only within the tissues.line 66: Please clarify what the "small pockets" are referring to.line 75: Parentheses missing around GNSSline 95: The inclusion of the CLM analysis while exciting comes up abruptly here and the importance of this analysis could be integrated sooner. Suggest to reference Burns et al 2018 or specific the concepts from that paper that will be expanded upon here.line 163: Suggest to replace "use the concept" with "assume"Figure 3, 8: It would be helpful to please clarify in the results why there is not diurnal cycle of VOD where VOD peaks in the morning and decreases into the afternoon even on the dry days. This is common finding in Holtzman et al. 2021 Biogeosciences and Yao et al. 2024 Geophysical Research Letters. Later on VOD does increase due to rain but do the trees not dehydrate when transpiring through the day?336-337: Please describe the B0, G1, A1, and F2 cases here or in the methods sections for reference to the reader.419: This is does not seem like a novel finding given your reference to the common method of removing these data in the introduction. Please clarify the novelty here.477: suggest to replace leaf with needles for clarity.Citation: https://doi.org/
10.5194/egusphere-2025-1755-RC1 - AC1: 'Reply on RC1', Sean P. Burns, 16 May 2025
-
RC2: 'Comment on egusphere-2025-1755', Anonymous Referee #2, 22 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1755/egusphere-2025-1755-RC2-supplement.pdf
- AC2: 'Reply on RC2', Sean P. Burns, 30 May 2025
-
RC3: 'Comment on egusphere-2025-1755', Anonymous Referee #3, 27 May 2025
This paper compiles VOD, tree sway frequency, and flux tower data for a subalpine forest. Additional data include output from a land-surface model, and field/sensor data related to air temperature, bole temperature, bole apparent dielectric permittivity, wetness, and precipitation. Together, these data are used to suggest that VOD and tree sway frequency capture signals related to canopy evaporation. These measurements offer unique insight into canopy storage dynamics from completely independent observations. I don’t have any major suggestions/comments for the authors. The following suggestions are meant to help the reader understand additional context for the tree sway measurements, interpretation of results, and to facilitate reproducibility/follow up studies.
Main Comments
- I agree with Referee #2 in that parts of the tree sway methodology can be expanded, even briefly. Suggested expansions:
- A) Line 145-146. If possible, include a brief description of the accelerometer methods, even though they are included in Raleigh 2022. Can go in the appendix, along with the other measurement expanded details if authors see fit.
- B) Line 147. Why was a 30-minute window chosen? Were other windows tried but this was the best one for the analysis?
- C) Line 148. What interpolation and smoothing function(s) were used?
- D) Line 156-157. How was the value chosen to be 1? Was this done by offsetting the low frequency trend such that the average frequency over the time series was 1? If so, please include this.
- Similarly, Line 138, can there be a brief inclusion of the data processing for VOD? Even if it is a brief summary, it would be helpful. This can be in the appendix.
- Table 1. Is it possible to add the original/raw frequency of measurements for each observation?
- Figure 2. Is the accelerometer associated with Pine 3 or Pine 4? Or are the dielectric permittivity sensor on two different trees? Please clarify.
- Line 347-349. Is it possible to expand on this (how these empirical relationships might facilitate changes in the land-surface model)? This seems like an important consideration, however there is nuance to it. In particular for tree sway frequency, the tree sway frequency is proportional to the inverse square root of tree mass, so while the frequency is observed to be linear, this does not mean that the tree mass (interception) is changing linearly. However, there is no mention of how tree sway frequency relates to mass in a mathematical form throughout the manuscript. I suggest adding this here and/or earlier on in the manuscript so if improvements on the land-surface model can begin there’s the additional context of how physically sway frequency relates to changing tree mass due to interception.
- This is not exceptionally important, and the authors can ignore: I was trying to determine any sort of relative magnitude differences in responses for the precipitation events presented in table 3 and figure 8 (and 9). I tried to find the lowest precipitation event (3.6 mm, solid cyan line) and the highest precipitation value (29 mm, solid red line). I think I found them in the sway frequency plot and they had the lowest and highest range/variability, which was helpful to see that with a first order approximation that the observations aligned with physical interpretations (more precipitation --> more change in tree sway frequency). I couldn’t quite find them in the VOD diagram. Highlighting these two events with bolded lines might help the reader understand the relative sensitivity to single events and strengthen the connections observed between precipitation, evaporation, and VOD or sway frequency.
- Are the authors amenable to including in the acknowledgements or in brief ‘open science’ section in the appendix or supplemental info where the data and code used throughout the manuscript can be found? This would help future studies reproduce these results and follow up studies that are interested in conducting similar experiments elsewhere.
Citation: https://doi.org/10.5194/egusphere-2025-1755-RC3 - AC3: 'Reply on RC3', Sean P. Burns, 05 Jun 2025
- I agree with Referee #2 in that parts of the tree sway methodology can be expanded, even briefly. Suggested expansions:
-
RC4: 'Comment on egusphere-2025-1755', Anonymous Referee #4, 01 Jun 2025
This research investigates promising and novel techniques to predict rainfall interception and, thus indirectly, canopy evaporation. The authors use L-band microwave active microwave attenuation data (vegetation optical depth) from a GNSS doublet and tree sway data measured by a an accelerometer placed on the trunk of a tree canopy. The study is set in an subalpine, high elevation needleleaf forest in in Colorado/USA. The authors demonstrate the ability of both proxies to correlate with onset and drydown of precipiation events, evapotranspiration and modeled interception storage from different land surface model (CLM4.5) parametrizations. The data sets and analysis presented by the authors allow for the conclusion that these techniques are promising tools to measure interception storage and that they hold potential to supplement/validate land surface models that are known to have high uncertainties in interception fluxes the their parametrization of the canopy, and uncertainties in EC water flux measurements during rain events. This study is of great quality. However, the authors should address the comments below before publication.
Major comment
Please elaborate on the robustness of tree sway motion being able to represent interception storage without the need to account for wind speed as a possible confounding factor. In this context, it would be valuable to find sway motion data as a function of wind speed – e.g. in fig. 10 and at least in one of the plot over time – to clarify on this relationship and include this missing piece of information.
Minor comments
- Fig 1
- Please report the the inner circle radius (r=20m) as the authors have done for the outer circle
- Please point the reader of fig. 1’s caption to what the different footprint circles represent to better understand the results, i.e. what is the main take-away from the inner circle radius (apart from GNSS paucity visualization).
- B 527: Please clarify the role of r=20 in this study. Did the authors clip the radius so the “several tall trees” are not included in the footprint?
- B & S3 (GNSS sky view):
- Please show the elevation angle and cutoff to clarify which parts of the canopy will effectively be used for VOD calculation, especially elevation=10°
- Please clarify the rationale behind clipping out another area in NE, close to the northern GNSS gap
- You use a very low cutoff elevation angle of 10°. Looking on fig. S3 – assuming the outer two circles being roughly within θ in (10, 30] – only very low VOD can be found that do not display any pattern expected from forest attenuation and possibly fail to represent true forest VOD. Under this light, please explain why the authors used a cutoff=10°.
- Since the Lambert-Beer angle correction assumes a homogeneous canopy and ignores multipath scattering, any losses observed may be due to scattering caused by multiple layers of vegetation, causing Lambert-Beer to break at low angles. Hence, the referee suggests using a higher cutoff elevation angle (~30°) or would value a discussion why low VOD at lower angles will not affect the overall results. Consider page 12 in Camps et al. (2020) about this question: “Note, however, that only at high elevation angles (elevation angle > 67.5°) is the single scattering albedo correlated with the NDVI, and at lower elevation angles, the presence of multiple scattering makes the tau-omega model [all zeroth order assumption, incl. lambert-beer, ~the referee] more likely to be invalid.”
- 135: Which GNSS frequency is used, please indicate the frequency(ies) in section 2.2.1 since GNSS VOD offers a range of bands to choose from.
- The authors detrend sway motion to alleviate effects of temperature and vegetation water content on short-term changes. However, VOD is also affected by long-term changes in biomass, and vegetation water content. Why did the authos not consider detrending VOD, especially since a trend is visible in fig. 2? This is worth noting in 2.2.1.
- 433/4: The size of the EC footprint has not been explicitly mentioned in the text. Also, which footprint size of VOD as your referring to in this statement? To make a statement about the footprint size (a very relevant discussion) the referee suggests to state that although the footprint sizes between all technique partly or greatly differed, the good correlations could be found etc.
Citation: https://doi.org/10.5194/egusphere-2025-1755-RC4 - AC4: 'Reply on RC4', Sean P. Burns, 06 Jun 2025
- AC5: 'Reply on RC4', Sean P. Burns, 27 Jun 2025
- Fig 1
- AC6: 'Comment on egusphere-2025-1755 (List of proposed manuscript changes)', Sean P. Burns, 27 Jun 2025
Data sets
US-NR1 AmeriFlux Site Data Peter Blanken, Sean Burns, Russ Monson, Dave Bowling, and Andrew Turnipseed https://doi.org/10.17190/AMF/1246088
Tree Sway Frequency Data Mark Raleigh https://zenodo.org/records/5149308
US-NR1 AmeriFlux Site Supplemental Data Sean P. Burns, Peter D. Blanken, and Russell K. Monson http://dx.doi.org/10.15485/1671825
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 589 | 96 | 32 | 717 | 47 | 18 | 28 |
- HTML: 589
- PDF: 96
- XML: 32
- Total: 717
- Supplement: 47
- BibTeX: 18
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1