the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Feb 2026
CO2 and Heat exchange across the Nocturnal Canopy–Atmosphere interface in the Amazon rainforest
Abstract. We investigated the characteristics of the nocturnal boundary layer (NBL) above and within the Amazon rainforest canopy during the 2022 dry season. The study aims to determine how NBL dynamics influence nocturnal CO2 and heat exchange across the canopy-atmosphere interface. Utilising observations from the CloudRoots-Amazon22 field campaign conducted at the Amazon Tall Tower Observatory, we distinguished between the strongly and weakly stable regimes to study the effect of radiative cooling and wind shear on CO2 and heat exchange. Our results reveal a distinct, stable layer above the canopy with an average height of 150 to 188 m, which develops due to strong radiative cooling of the canopy top. Below the canopy, the cooling marks the build-up of a well-mixed layer within the canopy. In the weakly stable regime, increased turbulence at the canopy top was observed, leading to a significant observed CO2 flux of 3.82 𝜇mol m−2 s−1 above the canopy. In contrast, in the strongly stable regime, turbulence was almost absent, and the observed flux was only 0.16 𝜇mol m−2 s−1, suggesting a decoupling of the canopy and the roughness sublayer. The decoupling was confirmed by the 2–3 times decreased vertical heat transport in the strongly stable regime. Even though our method includes typical observational uncertainties, our results show significant differences between CO2 and heat exchange between the two regimes, stressing the importance of correctly representing the nocturnal dynamics in tall canopies like the Amazon Rainforest.
- Preprint
(4939 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-684', Anonymous Referee #1, 21 May 2026
- AC1: 'Reply on RC1', Anna Huitema, 18 Jun 2026
-
RC2: 'Comment on egusphere-2026-684', Anonymous Referee #2, 28 Jun 2026
In this manuscript, the authors study the nocturnal CO2 exchange processes by using the multi-level tower facility over the Amazon rainforest. The observations are impressive, but I have significant concern if those have been fully utilized. After reading the manuscript, it appears to me that the authors are invested more towards discussing how to evaluate the nocturnal boundary layer height, and then just summarizing the results of the budget without specifically unravelling the scales of the turbulent processes that govern the exchanges of CO2 between the canopy and the atmosphere above. As a result, this manuscript never reads like a scientific article but more like a report, thereby limiting its novelty.
The authors state that in the weakly stable regime the shear production term in the TKE budget plays a significant role in shaping the CO2 transport between the canopy and the atmosphere aloft. This is not surprising, since in such conditions the shear is the only mechanism by which the turbulence is generated and hence the CO2 transport correlate well with that term. It is more important to understand the length scales involved. The authors have high frequency observations and thus it should not be an issue to investigate the cospectra of CO2 and heat fluxes. Given they have multi-level observations (which is rare to have, especially for scalars), such an analysis would significantly strengthen the study, and the finding would be more focused towards understanding the processes rather than summarizing the observations. With this, I recommend a major revision. My other comments are given below.
Comments:
‘Accurately representing the CO2 exchanges ----‘, I think the authors need to elaborate on the challenges here.
Paragraph in Line 50: This paragraph sounds to me like summarizing the results in the Introduction section. The authors should focus more on motivating the present study with respect to previous ones rather than stating the results in the Introduction.
Line 85: ‘tall 90 tower’, tall 90 tower or 325 m tower?
Line 85: ‘Moisture and CO2 fluxes ---‘, Were the moisture and CO2 fluxes WPL-corrected?
Line 100: ‘To not limit ourselves ---‘, I do not understand what the authors mean.
Line 105: What sort of data limitations?
‘U_g being the geostrophic wind’, Is U_g really the geostrophic wind? Why do you need to specify that as a geostrophic wind, would it not suffice to just mention as the wind speed at the highest height?
Line 145: ‘----- are done’, This sentence reads odd.
‘In addition, we aligned ---‘, Was this alignment done at each height? What was the turbulence intensity of the flow? If the turbulence intensities were large, this alignment can be highly inaccurate.
Line 160: Why you need to parametrize the dissipation rate? Can't you directly compute it from the u spectra?
‘We expected the latent heat ---‘, I do not understand this, since you have the moisture flux measurements why you need to invoke this assumption?
Line 260: ‘canopy crown’, What does the +- sign refer here?
Line 285: How does it decoupling parameter compare the against the previous studies, such as the one by Peltola et al. (2025)?
Line 325: ‘Dissipation below ----‘, This term is parameterized and therefore I have significant concerns. The authors should reevaluate this term from the velocity spectra. I am unsure why the authors state that this was beyond the scope of this study. If you have the data, this should not be a too cumbersome exercise.
Line 370: Why you are attributing this residual to horizontal advection? You have not included the dissipation rate of temperature , which also would play a significant role in deciding the heat budget. This could also be done with the high frequency temperature observations. Yes, within the canopy the temperature fluctuations could be sufficiently damped and might not show an inertial subrange but above the canopy it should be OK to see a well-behaved temperature spectra and heat flux cospectra.
16: Line 405: Again you are ignoring the dissipation terms in the CO2 budget.
Reference:
Peltola, O., Aslan, T., Aurela, M., Lohila, A., Mammarella, I., Papale, D., ... & Laurila, T. (2025). Towards an enhanced metric for detecting vertical flow decoupling in eddy covariance flux observations. Agricultural and Forest Meteorology, 362, 110326.
Citation: https://doi.org/10.5194/egusphere-2026-684-RC2 -
RC3: 'Comment on egusphere-2026-684', Anonymous Referee #3, 30 Jun 2026
The authors investigate nocturnal CO₂ exchange processes using a multi-level tower system over the Amazon rainforest. The study addresses an important topic, and overall the manuscript is well written and generally meets the scope of the journal. However, I have several comments and suggestions that should be addressed to improve the clarity, readability, and scientific presentation of the manuscript.
1. The manuscript is considerably long, which makes it difficult to follow in several sections. I recommend carefully revising the text to remove repetitive or less relevant material and focus on the key findings. A more concise manuscript will improve readability and maintain the reader's interest.
2. The introduction includes a summary of the study results, which would be more appropriate for the Results or Discussion section. Instead, the introduction should focus on the scientific background, motivation, relevant previous studies, and the research gap. The objectives and novelty of the study should be clearly stated at the end of the introduction, highlighting how this work differs from previous research and addresses existing knowledge gaps.
3. Please clarify whether any outlier detection, quality control, or data-cleaning procedures were performed before the final analysis. If so, describe the methods used.
4. Line 25: The phrase "in its immediate surrounding" is unclear. Are the authors referring to the eddy covariance footprint? If a footprint analysis was performed, please describe the methodology and discuss its implications for the results.
5. Please clearly define the variables U, U* (friction velocity), and U, when they are first introduced in the manuscript.
6. Figure 2: The figure would benefit from including units for all variables. Please ensure that the units are shown either on the axes or clearly stated in the figure caption.
7. Section 3.2 (Temporal Evaluation of Stability Within and Above the Canopy): The first paragraph of this section is difficult to follow and contains several unclear statements. I recommend rewriting this paragraph to improve clarity and better communicate the main findings.
8. The study is based on approximately two weeks of observations. Please discuss whether this observation period is sufficient to draw broader conclusions regarding nocturnal canopy-atmosphere interactions in the Amazon rainforest. It would be helpful to acknowledge this limitation and discuss its implications.
9. The Results and Discussion sections would be strengthened by a more comprehensive comparison with previous studies. Please discuss how your findings agree with or differ from the existing literature and explain the possible reasons for any differences.
10. The Results section is overly detailed, and several paragraphs are difficult to follow. Consider condensing the presentation by focusing on the most important findings and improving the clarity of several sentences. A more concise presentation will make the results easier to interpret and more engaging for readers.
Overall, this is a valuable study with the potential to make an important contribution to understanding nocturnal CO₂ exchange in tropical forests. Addressing the comments above will improve the manuscript's clarity, scientific rigor, and overall impact.
Citation: https://doi.org/10.5194/egusphere-2026-684-RC3 -
CC1: 'Comment on egusphere-2026-684', Bruce Hicks, 07 Jul 2026
Overall, this is good stuff. There are plenty of changes that a careful re-read after a few weeks of time off would detect, and I recommend that the authors take the opportunity of the present lengthy review process to rethink the text and its language in detail. This is not to say that there is anything greatly wrong, but why not turn the review delay into a positive?
An immediate reaction is that the authors have set themselves a big challenge with a limited dataset. In particular, while I applaud their attention to nocturnal intermittency, I suspect that examination of this topic is severely constrained by the sampling protocol. To be blunt, 30-minute averaging has been shown to hide most of the signal indicative of this phenomenon. In my own experience, while 30-min might satisfy old-fashioned micrometeorologists, a separate 5-min (or shorter) dataset is necessary to look into the basic processes that control the nighttime air-surface exchange of all quantities, including momentum.
To my mind, the historic focus on log-linear profiles and their consequences is old stuff, maybe best forgotten. I am convinced that this is one of many areas where artificial intelligence will provide opportunities to delve into datasets of many kinds, including the impressive archive assembled by ATTO. My present belief is that the gradients detected at night are the result of how many intermittent occurrences occur during a 30-min period. Consequently, I look at Table 1 with a suspicion that strong stratification might be best quantified using the standard deviation of CO2 concentrations within 30-min periods rather than using the gradients. This would, of course, redirect much of the analysis that follows, and hence while this represents my own view of the dataset and the questions addressed in the present text, I cannot suggest anything more than the inclusion of a paragraph to address the intermittency matter a bit more thoroughly.
My own examination of intermittency on and above forest canopies indicates that temperature is affected much as is CO2. The signal is harder to extract, but it is still there.
The discussion around line 115 caused me to pause a bit. Normalizing the CO2 gradient must involve the CO2 eddy flux determination. The latter is a 30-min average made up of a bunch of intermittent events (I suspect – prove me wrong!) as is the gradient. My mind is struggling with what happens when one considers the ratio of these two, which I think the normalization does.
Equations 1 and 2 are the same, but applied over different height regimes.
Section 2.5.1 needs some clarification. I could not find a definition of ν (I believe the symbol in the definition of the dissipation is Greek nu) but my own opinion is that the various relationships introduced in this paragraph are rather old and based on analyses about which I could argue because they manipulate hard-to-measure quantities in ways that lead to statistically uncertain results. In any case, why not just use the measured TKE without embellishment? Has Stull’s “15” been verified?
In Section 2.5.2 we find that results published by others, elsewhere, are used to modify actual observations prior to analysis. This was the case earlier in the manuscript and gives the impression (I hope incorrect) that results could depend on assumptions made regarding the selection of models to be used prior to data analysis. Nevertheless, the 1 – 12 W m-2 difference between model and observations is outwardly impressive, until it is realized that this is the absolute difference range. I would like to see the average difference and its standard error.
Figure 3 is of great interest. In particular, I would interpret the profiles (CO2 as well as temperature) as indicating linearity with height within the canopy and a different linearity with height above it. This would align with experimental results elsewhere (many not yet published, so I cannot be too outspoken about this).
Section 3.2.1 poses some problems that would be solved if there was a listing of nomenclature. I finally found that Ric is taken to be 0.25. I consider this to be an average based on countless studies of gradients in the past, when instrumentation was not yet adequate to examine the matter in a manner that would account for chaotic variability and intermittency.
Section 3.3.1. Here, I cannot comment much, but once again I would like to see how you might expect nocturnal intermittency to affect the analysis and conclusions.
Line 358. I cannot relate “enhanced shear” to the “weakly stable” situation. Increased stratification means grater shear, surely. And I interpret the causative factor as being increased transport of warmer air into the canopy from aloft due to intermittency.
Section 3.3.3 I believe that at night a major contributor to changes in CO2 concentration is the variation of soil efflux. Changes in soil temperature are central issues, with changes in atmospheric pressure being suspected contributors. As soil cools during the night, CO2 efflux might well decrease (in summer). In daytime, the efflux rate might decrease with increasing temperature. The temperature influence on sub-surface microbial activity is well documented in general, but doubtlessly there must be some sort of species dependence.
For future activities, I would recommend (1) a separate archive of 5-min average (or shorter) observations, (2) soil temperature data, (3) inclusion of discussion about soil efflux changes and (4) supporting weather data such as the strength and height of the low level jet (when it exists).
Citation: https://doi.org/10.5194/egusphere-2026-684-CC1 -
RC4: 'Comment on egusphere-2026-684', Peter A. Taylor, 08 Jul 2026
Looking over the previous reviewers comments there are many valuable points. I would however raise three more issues concerning the horizontal homogeneity of the site and the boundary layer depth calculations.
In the cited paper by Andreae et al (2015) there are several comments about terrain inhomogeneity. These include, in Section 2.1, “From the perspective of micro-meteorological flux measurements, this is not an ideal type of terrain because it induces significant up-slope and down-slope circulations. The effects of local topography on the local flux measurements from the small towers are the subject of ongoing investigations.” Figure 1b of the Andreae paper shows the topography in the region around the ATTO site. There appears to be considerable topographic variability within a 5 km circle around the 325m ATTO tower. Andreae state “The tower site is located approximately 12 km NE of the Uatumã River (Fig. 1b). As is typical for this region in the central Amazon Basin, there is little large-scale relief, but at smaller scales a dense drainage network has produced a pattern of plateaus and valleys with a maximum relief height of about 100m ….”
My main concern would be the assumptions made in Section 2.5 concerning horizontal homogeneity. “To evaluate the governing equation with our observations, we made the following two assumptions: 1) The atmosphere is horizontally homogeneous (∂ψ ∂x = ∂ψ ∂y =0) and 2) subsidence is negligible (w = 0) are done.” These lead to Equations 3,4,5. Remainder terms are included and will include advection associated with horizontal inhomogeneity. “The Rθ encompasses the molecular diffusion term, the latent heat term and the advection of heat (Stull, 1988).”
The problem is that these remainder terms are generally as large as the other terms, at least in the weakly stable case (Figures 6, 7, 8).
We are told (line 375) that “To the east (main wind direction) of the ATTO, a swampy area is located …”, but there is little mention of wind direction data elsewhere, although we are told “we found that NLLJ formation (associated with low turbulence) occurred under Northeast winds, in line with the lower roughness area to the Northeast.” Key questions are whether wind direction was measured, and did it have any impact on boundary-layer depths and other boundary layer parameters. Maybe wind direction could be added to profile plots like Figure 5.
In discussing boundary-layer depths and Fig 1. Line 102 states “To not limit ourselves to hNBL at measuring heights, we fitted a 2nd-order logarithmic function to vertical profiles of θ, CO2 and U (Stull, 1988).” Checking though Stull I could not find this but envisage a form, for θ, like
(θ – θ0) / Δθ = A (ln (z-zc) – ln z0θ) + B (ln (z-zc))2 where zc is canopy height and z0θ is a temperature roughness length. Or the 2nd order could be ln(ln (z-zc) ?
We should be given details of the “2nd-order logarithmic function” used and confirmation that a displacement height was used. Values determined of A, B and z0θ might also be interesting. These should include U and CO2 profile fits.
As I see it the best hNBL estimate should be the simple TKE < 0.2 TKE0 criterion. In Fig 2 this and the estimate based on U criterion seem to have the most variability. This variability may be real and averages over 12 different days at selected hours seems a questionable way to present these data. Each day may have totally different external parameters, like wind speed and direction?
As with the other reviewers I expect that, with careful investigation, these data from the CloudRoots-Amazon22 campaign can lead to an improved knowledge of nocturnal boundary layers over forests and the associated CO2 and heat fluxes.
Citation: https://doi.org/10.5194/egusphere-2026-684-RC4
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 841 | 447 | 81 | 1,369 | 114 | 113 |
- HTML: 841
- PDF: 447
- XML: 81
- Total: 1,369
- BibTeX: 114
- EndNote: 113
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors aim to understand how the nocturnal boundary layer (NBL) dynamics influences CO2 and heat exchange between the air layers within and above the Amazon rainforest canopy. The field observations within and above the canopy layer are unique, even though the authors focus on a relatively short period of about two weeks during the Amazon dry season.
I am disappointed about the analysis methods and confused by many detailed descriptions of analyses. I encourage the authors to think more carefully about the key factors in driving CO2 and heat exchange and to demonstrate more clearly what the field observations reveal. I believe readers could learn more from the field analysis without the introduction of various unnecessary assumptions.
Major comments:
1. There are two radiative coolings at night: longwave radiative cooling from the surface and the canopy layer, and the cooling associated with vertical radiative divergence within the air layer, either above or within the canopy. For the development of the NBL, radiative cooling below the NBL is generally much more important than vertical radiative divergence, based on previous field observations as far as I understand. It is therefore important to distinguish the two especially when discussing radiative cooling from the soil surface and the canopy surface. I was often confused about which process was discussed in the manuscript.
With cooling occurring below the air layer, whether from the soil, the canopy surface, or both, the development of turbulent mixing under stable conditions is primarily determined by wind shear as demonstrated clearly in Fig. 5.
2. With the above considerations in mind, my major confusion is why the authors used six criteria to classify the observation based on previous investigations. Although the fundamental physics governing the atmospheric boundary layer is the same, the observation heights and canopy structure in this study differ from all the studies listed in Table 1. Why can the authors not analyze their own observations directly and then highlight the unique characteristics of different stability regimes based on what they observe? After reading the abstract, I expected a comparison between the air layers above and within the canopy and kept looking for distinct stability regimes in each layer. Only after reading the manuscript several times, I realized that all six criteria are based on observations above the canopy layer. Thus the distinction between strongly and weakly stable regimes applies only to the air layer above the canopy.
3. Since the study deals with turbulent transfer across the two layers, a brief description of turbulence measurements such as sensor types and data processing methods would help readers understand the data used in the study even if detailed information is available elsewhere.
Because the study concerns the CO2 and heat exchange between the two layers, it is also important know the canopy density, tree types, and whether low vegetation is present near the soil surface.
In addition, because the analysis is based on measurements from two towers, the manuscript should specify how far apart the two towers are and whether they are within the same forest stand.
4. I am particularly puzzled by evaluation of the boundary layer height. From what I understand, the measurement extends only to 298 m, assuming above the ground. With an average tree height of about 20 m, the maximum depth of the air layer above the canopy layer is therefore about 278 m. Considering the development of the stable air layer above a cooling surface, the NBL impacted by the surface can be quite shallow. Therefore, the NBL heigh, h_NBL, might be directly inferred from vertical profiles of turbulent variables measured from the two towers. Furthermore, any coupling between the two layers should be readily observable from the vertical variation of turbulent variables across the two layers under different conditions.
5. The description of the data used in this study is confusing. See my detailed comments below.
Detailed comments
Figure 1: Is this figure intended as a schematic illustrating the five criteria? Also are NLLJs frequently observed under the strongly stable conditions?
L.112: What is h_theta?
L 119: How do the authors justify the assumption of U=U_g? It seems to me that this is an unnecessary assumption, especially if the goal is simply to classify the strongly and weakly stable regimes.
L 123: If “sufficient turbulence above the canopy is required for the application of the criteria”, why is it necessary to separate the strongly and the weakly stable stratification?
L 126: At which measurement level is TKE assumed to be 0.2 TKE_0?
Eq. (2). Ri_can is not actually an in-canopy layer Richardson number as stated in L. 133. Instead, it represents a Ri across the canopy top.
L. 157: Do the authors mean horizontal advection of TKE?
L. 158: is theta_v derived from sonic anemometer temperature measurements?
L. 168: Assuming latent heat release to be zero at night is questionable. Even under very stable conditions with weak turbulent mixing, evaporation generally continues throughout the night.
L. 180: What do the authors mean by “sufficient to get a realistic estimate of radiative cooling”? See Hoch et al. (2007 JAMC, 46, 1469).
L. 187: There is no molecular viscosity in the CO2 balance.
L. 196: With such a valuable two-weeks observational dataset, only eight hours of the data are used for the study? That corresponds to less than 6% of the nighttime observations. What characterizes the excluded hours ? Where they associated with Intermittent turbulent mixing? Since clouds are mentioned here, where the selected hours limited to cloud-free conditions? By the end of the manuscript, it appears that Scu clouds were frequently present even during the dry season. Is there a way to identify impacts of cloud cover on surface radiative cooling and the subsequent development of the NBL? How representative are the selected data for this study overall? How many nights were ultimately included in the analyses?
L. 200: Should this refer to Fig. 2 instead of Fig.5?
L. 238: Only 8 hrs data were used for the study. How many independent data points can be obtained from a two-hour average?
Fig. 3: Are these profiles derived from the 8-hours observations mentioned earlier?
L.243: “…, a shallow stable daytime layer within the canopy was present”. Based on Fig.3, both the entire NBL and the within-canopy layer appear stable. Is Fig. 3 constructed from the three nights listed in Table 2? If ambient wind and temperature vary from night to night, then Fig. 3 does not represent the temporal variation of theta and CO2 during any individual night. What exactly does the figure represent then?
L. 250: S2,…S4 and W3 are not introduced here without prior explanation. I eventually fount them in Table 2, and they are only discussed later in L 294. Does this mean that all analyses are based on the three nights listed in Table 2? On 13 August, the NBL apparently transitioned from weakly stable to strongly stable conditions. It would be interesting to see how this transition occurred.
L. 253: A case study illustrating the evening transition would be particularly informative.
L. 255. “Black error bars”? All the horizontal bars in the figure appear red. How were the error bars calculated? Do they represent variations between nights?
L. 257: The two proposed mechanisms sound good. Which one is more important for the present study?
L. 273: “from 11 nights”? I am totally lost here. The “eight hours” mentioned at the beginning of section 3 gave the impression that the entire section was based on that limited dataset. Does the observational dataset vary between subsections of section 3? As Table 2 does not cover the full period from 18:00 to 06:00, please specify the data used for each figure and analysis.
Figure 5. I find this figure is the most informative although I still don’t know which nights are represented.
L. 313: “the four selected periods”, where were they described?
Figures 6, 7, 8: What do the shaded areas represent?
L. 448: Clouds. Yes, it would be interesting to see their impact on the NBL development.
L. 486: The study focuses on nighttime conditions during which stable layers are present throughout. Is the discussion here referring to longwave radiation emitted from the canopy top or to radiative divergence cooling within the atmosphere?
Fig. B1: Interesting! Whenever winds are strong over the canopy layer (W1-W4), turbulent mixing in the air layer above the canopy can penetrate into the canopy layer resulting in enhanced heat transfer. When winds above the canopy layer are weak (S1-S4), the positive heat flux below the canopy crown (presumably) in the nearly neutral canopy layer (Fig. 5a) may develop because shear-generated turbulent mixing transports warm air downward from above the canopy while turbulent development near the soil surface remains constrained.