the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Lower-cost eddy covariance for CO2 and H2O fluxes over grassland and agroforestry
Abstract. Eddy covariance (EC) measurements can provide direct and non-invasive ecosystem measurements of the exchange of energy, water (H2O) and carbon dioxide (CO2). However, conventional eddy covariance (CON-EC) setups (ultrasonic anemometer and infrared gas analyser) can be expensive, which recently led to the development of lower-cost eddy covariance (LC-EC) setups. In the current study we test the performance of a LC-EC setup for CO2 and H2O flux measurements at an agroforestry and adjacent grassland site in a temperate ecosystem in northern Germany. The closed-path LC-EC setup was compared with a CON-EC setup using an enclosed-path gas analyser (LI-7200, LI-COR Inc., Lincoln, NE, USA). The LC-EC CO2 fluxes were lower compared to CON-EC by 7–13 % (R2 = 0.91–0.95) and the latent heat fluxes were higher by 2–3 % in 2020 and 23 % in 2021 (R2 = 0.84–0.90). The large difference between latent heat fluxes in 2021, seems to be a consequence of the lower LE fluxes measured by the CON-EC. Due to the slower response sensors of the LC-EC setup, the (co)spectra of the LC-EC were more attenuated in the high-frequency range compared to the CON-EC. This stronger attenuation of the LC-EC requires a larger spectral correction and as a consequence larger differences between spectral correction factors of different spectral correction methods. At the agroforestry site where the flux tower was taller compared to the grassland, the attenuation was lower, because the cospectrum peak and energy-containing eddies shift to lower frequencies which the LC-EC can measure. With the LC-EC and CON-EC systems was shown that the agroforestry site had a 2.3 times higher carbon uptake compared to the grassland site and both had an equal evapotranspiration when simultaneously measured for one month. Our results show that LC-EC has the potential to measure EC fluxes at various land-use systems for approximately 25 % of the costs of a CON-EC system.
- Preprint
(12851 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on amt-2024-30', Anonymous Referee #1, 07 Apr 2024
Review of Lower-cost eddy covariance for CO2 and H2O fluxes over grassland and agroforestry, submitted to AMTD by van Ramshorst et al.
Summary
This manuscript presents comparisons between lower cost and conventional eddy covariance instrumentation at two landscapes – a grassland and an agroforestry site in Germany. The work shows compelling evidence that the lower cost system can be successful at measuring CO2 and H2O fluxes in these landscapes. The work is of interest to the readership of AMT and is of generally good quality. However, I think it can be improved significantly through more hypothesis driven objectives, clearer writing, and more concise presentation. Some logging errors (a sonic at 2Hz instead of 20Hz) should be more adequately addressed. I have major, minor, and technical comments that should be considered. I rate each of its scientific significance, scientific quality, and presentation quality as “Good”.
Major comments
- We don’t learn until L279 that in 2020 the CON-EC system is sampled at 2 Hz but logged at 20 Hz, so the values are repeated ten times, which causes harmonic oscillations. This needs to be addressed in the methods and more explicitly considered. I’d tend to just report the fluxes and cospectra from 2Hz and not 20, but maybe there are other ways to do this (e.g. focusing on spectra from the gas analyzers). Even consider really emphasizing the 2021 dataset through the paper and then using 2020 for a more supportive role.
- There are many graphs – are they all needed?
- Fig 11 is not gap-filled and loses a lot of value that way. Can a simple gap-filling approach (e.g. with REDDYPROC) be used to assess the impact of these site and instrument differences in a real use-case?
Minor comments
- The introduction is 6 paragraphs and could probably be a bit more concise. Eddy covariance isn’t mentioned till the end of paragraph 2 and defined in paragraph 3. I would tend to combine everything before L32 into paragraph 1 (and write more compactly); perhaps L42-59 can also be more integrated and concise (e.g. the repetition of L51-52 – spectral…spectral, corrections…compensate; or L59 “which is an additional source of uncertainty in itself”).
- L60 the introduction could finish with clearer hypotheses about the expected findings, rather than objectives only.
- L90 or so could add specifications on precision of the instruments, not just their response times
- L99 and 114 consider commenting on why the intake is 20cm below the sonic anemometer and risks of spectral attenuation since they would be sampling eddies at different wind velocities.
- L151 could add some brad description of the difference in theoretical background or implementation of the Horst and Ibrom and HL09 approaches
- L198 have the residuals of the regression always been tested for normality?
- L206 and elsewhere – there are a lot of multiple “respectively” chains and perhaps this and other sentences like it can be simplified
- L219, 220, 224 all compare the systems or the sites, and the words “lower relative” or “higher” can be quantified with percent or absolute differences.
- L295 it’s not clear to me – the most affected…by aliasing?
- Fig 4 could include bias – both average and in different ranges (e.g. in the positive and negative FCO2 ranges separately)
- Fig 6 and likely elsewhere (e.g. Fig 7 and Table 3), “EC” is used rather than “CON-EC”. I’d prefer the latter to be explicit and consistent throughout the text.
- Fig 5a legend for LC-EC should be a green line; caption should have “light blue” as two words rather than one.
- L345 consider more strongly highlighting that there is a 10-15% difference in results based on the correction method
- L462-470 could almost be in the intro to lead to hypothesis-driven objectives
- L494 explain why the bowen ratio decreased in these conditions
- L537 be clearer on where and when this difference occurred
- L549 I think “lower in magnitude” and not in absolute number (since the LC system is higher at noon than the CON system in Fig 3 for CO2 flux)
Technical comments
Generally there are many small issues in word order, punctuation and orthography, and vague writing that can be clarified and improved. I’ve noted some here.
- L3 could add who this lower cost system is made by? (equivalent to L5 reference to Licor)
- L8 no comma needed before seems
- L11-12 is both vague and wordy (it’s rather obvious that stronger attenuation requires larger spectral correction, and that thus the factors are amplified; perhaps there are better ways of saying this – more quantitative, more hypothesis oriented)
- L15 ET is never exactly ‘equal’ – perhaps “not statistically different” or similar
- L19 no comma needed after climate change, “the” not needed before increased
- L32 consider “Direct observations with” before “eddy covariance”
- L37 remove of
- L77 the quotes should go after agroforestry and not project
- L78 “if and” can be removed
- L81 the comma after tower is not needed
- L94 add s to time
- L104 “This” is ambiguous; in general it should be avoided without a noun following it
- L140 move “a” to after “narrow”
- Table 2 could add tau-nom
- L198 chagne were to was
- L230 “in the current paper” isn’t needed.
- L286, change look differently to looks different.
- L320, 321, 325, consider “varied” instead of was/were varying
- L360 remove also
- L369 clarify “This” and consider changing “led to consistently.”
- L374 “which” would grammatically refer to the current LC-EC setup, but makes more sense in the sentence to refer to the predecessor. Reword sentence.
- L379 I think “led” not “lead”
- L397 remove exact
- Fig 11 caption – add “AF” after agroforestry (consider also defining ET)
- L408 remove very
- L414 care not carefulness
- L420 remove comma after site
- L489 could remind us how those results are poor
- L490 remove “of all”
- L491 consider predicts instead of shows and removing probably
- L492 use either rather than both
- L509-510 – be more concise
- L536 add such before as
Citation: https://doi.org/10.5194/amt-2024-30-RC1 -
RC2: 'Comment on amt-2024-30', Anonymous Referee #2, 17 Jun 2024
Van Ramshorst et al. made a comparison of two eddy covariance setups using low-cost and conventional instrument(s) for a grassland and an agroforestry site in Germany. The authors showed that the low-cost EC system can measure fluxes at both landscapes. In general, the manuscript is of good scientific quality, it is well written, and fits to the scope of AMT. However, some issues need to be addressed before publication. I have major, minor, and technical comments which are listed below.
Major comments:
R1: In the abstract and conclusion it is written that low-cost EC system costs approximately 25% of the conventional EC system. Is this estimation made on material costs only or also working hours, building up the system, testing, setting up the logging software, etc.? How much time did you invest for preparation? The LICOR systems are standardized, have a robust measurement performance, and setting them up in the field may be less difficult than a custom-built setup. Then a customer may still buy the conventional instrument, which could also be the open-path CO2/H2O gas analyzer of LICOR.
The authors can add a recommendation section with regard to preparation, costs, testing of software, and instruments of the LC EC setup. Estimates of the acquisition costs for the instruments could be added to Table 1. Summarizing the main aspects regarding flux pre- and post-processing may be also useful to add.
R2: The authors wrote that sonic measurements of CON EC were sampled at 2Hz instead of 20Hz during the 2020 campaigns (L279-292). This issue needs to be introduced earlier, preferably in the methods section. This introduces a significant uncertainty to the flux damping calculation. Instead of doing the flux analysis using 20 Hz data, it may be better to use the real 2Hz data of the CON EC setup for comparison.
R3: I think you shouldn’t write that the LC system ‘‘potential to measure EC fluxes at various land use systems’’ (L16 and L560). Fluxes are reported only for two ecosystems. Also, we learn that the performance of CO2 flux measurements is better than for latent heat fluxes, and a competitive performance was achieved above tall canopies. I think this should be the final remark. It can be added to the conclusion that existing research infrastructures equipped with EC setups like ICOS can be used for further validation of the instrument setup and testing of flux damping methods. This can also be an aspect for a recommendation section.
Minor comments:
R4: In general, the manuscript is a technical study. The ecological aspect is kept short in the discussion, which is fine. However, the reader gets an introduction about the potential benefits of agroforestry regarding climate change (L19-33). I think this section can be shorter.
R5: The intakes of both setups were 20 cm below the center of their sonic anemometer (L99 and L113). The influence of the vertical separation on flux damping should be discussed. (Kristensen et al., 1997).
R6: To determine minimum and maximum lag, the nominal lag was multiplied with 0.75 and 1.5, respectively (L137-L138). How were these numbers determined?
R7: Were values different from 40 s tried as upper limit for H2O flux calculation (L141)? Did you apply a time lag filter? What is the time lag range of quality-assured fluxes?
R8: Please add references for each of the preprocessing steps (L147).
R9: A theoretical background should be added to the methods of Horst et al. (1997) and Ibrom et al. (2007) (L165-L166). Please check line 164. Low-frequency corrections are described in Moncrieff et al. (2004).
R10: Absolute limits for flux filtering were applied. Are the flux limits based on manual screening? If so, please add it.
R11: Figure 2: Please adjust the x-axis. It’s easier to compare the campaigns if both x-axes have the same starting date and change day of year to dates which are easier to read like Aug 15. The distance between the plot labels, (a) and (b), y-axis can be increased a little bit.
R12: Figure 3: Move the titles in the middle. The distance between the plot labels, (a) and (d), y-axis can be increased a little bit.
R13: Figure 4: Put the same titles on top of each row. There’s a lot of information in the plot. I propose to summarize the information in a table.
R14: Figure 5: Same as for figure 4. Also, the sections 3.2.2 and 3.2.3 are difficult to read because a lot of numbers are mentioned in text. A table can be helpful here. Align the format of the x-axes of (b), (d), and (f). Start and end date should be the same.
R15: Sec. 3.2.5: Maybe put this section to the method part? The influence of the concentration correction on fluxes is not in the discussion.
R16: Figures 6 and 7: Add the ensemble averaged (co)spectra to (e) and (a) and add titles to each row. How many (co)spectra were used for averaging? Explain the inertial subrange in the method section. Choose a different color for the temperature (co)spectra. They are difficult to distinguish.
R17: Figure 8: is the figure needed? Since it is not referred in the discussion, put it in Appendix or supplement.
R18: Sec 3.3: Put the three observations in a list, which makes the text easier to read.
R19: Figure 9: Put titles on top of each row and align the format of the x-axes and if possible also of the y-axes. Maybe the latter only for CO2. Make a space between g and C in y-label.
R20: Figure 10: Put titles on top of each row. The labels (a) to (f) look kind of large compared titles and labels. In the legend, the identifier for box plots, median and whisker, are hardly visible. Put numbers on top of each box showing how many values were used. Make the box borders, medians, and whiskers more visible. You may want to leave out the outliers. Then you see the range of the boxes better.
R21: Figure 11: Make a space between g and C in y-label.
R22: Sec. 4.1: The beginning of section 4.1 is a summary and not a discussion. (L360-370)
R23: L397: The unpublished work by Callejas-Rodelas et al. (2023) …. I can’t find Callejas-Rodelas et al. (2023) in the references. If a manuscript is not published, it’s better to write in the references a remark in submission or in discussion instead of always repeat unpublished work. Anyway, I think it’s published now: https://www.sciencedirect.com/science/article/pii/S0168192324002016
R24: L462-470: Maybe move this to the introduction or method section?
Technical comments:
L4: comma after study
L19: no comma
L20: leave out the “-” between carbon and dioxide.
L32 and L31: both sentences start with Eddy covariance.
L36-L37: add Heiskanen et al. (2022) as reference.
L37: no comma after impacts
L60: comma after study
L72 and L74: Both sentences start with based on gap-filled meteorological data. Consider changing.
L81: remove the tall after 3m. 10m sounds tall but not 3m.
L94: add s to time
L106: comma after 2021
L170 and L172: make CO2 straight.
L186: put brackets around 1.
L191: make H italic.
L194: comma after however
L198: add s to regression.
L203 and L204: comma after 2020 and 2021
L213: put bar on top of u*. The bar refers to an average.
L229: H is straight here. Change to italic. Also at other occurrences.
L230: figure not shown is sufficient
L234: Use a large ‘’-’’ which makes clearer that a range is meant here. Same for all occurrences.
L246: r or R2?
L257: change biggest to largest (same in L262) and add ‘‘the’’ after between
L266: comma after in general.
L314 and L315: comma after campaign.
L353, 354, and 355: repetition of similar. Consider rephrasing.
L356: change has to had and sites to site.
L357: is there a which between ratio and was missing?
L407: replace big by large.
L421: remove ‘’friction velocity’’ and commas
L422: comma after air
L427: remove tall after 3 m.
L477: put brackets around 1.
L486: confirm instead of confirms.
L499: had instead of has
L530: closed-path
L549: comma after 2021
References:
Heiskanen, J., Brümmer, C., Buchmann, N., Calfapietra, C., Chen, H., Gielen, B., Gkritzalis, T., Hammer, S., Hartman, S., Herbst, M., Janssens, I., Jordan, A., Juurola, E., Karstens, U., Kasurinen, V., Kruijt, B., Lankreijer, H., Levin, I., Linderson, M.-L., Loustau, D., Merbold, L., Lund Myhre, C., Papale, D., Pavelka, M., Pilegaard, K., Ramonet, M., Rebmann, C., Rinne, J., Rivier, L., Saltikoff, E., Sanders, R., Steinbacher, M., Steinhoff, T., Watson, A., Vermeulen, A., Vesala, T., Vítková, G., and Kutsch, W.: The Integrated Carbon Observation System in Europe, B. Am. Meteorol. Soc., 103, E855-E872, https://doi.org/10.1175/BAMS-D-19-0364.1, 2022.
Kristensen, L., Mann, J., Oncley, S. P., and Wyngaard, J. C.: How Close is Close Enough When Measuring Scalar Fluxes with Displaced Sensors?, J. Atmos. Ocean. Tech., 14, 814–821, https://doi.org/10.1175/15200426(1997)014<0814:HCICEW>2.0.CO;2, 1997.
Moncrieff, J. B., Massheder, J. M., deBruin, H., Elbers, J., Friborg, T., Heusinkveld, B., Kabat, P., Scott, S., Soegaard, H., and Verhoef, A.: A system to measure surface fluxes of momentum, sensible heat, water vapour and carbon dioxide, J. Hydrol., 188, 589–611, https://doi.org/10.1016/S0022-1694(96)03194-0, 1997.
Moncrieff, J. B., Clement, R., Finnigan, J., and Meyers, T.: Averaging, Detrending, and Filtering of Eddy Covariance Time Series, Kluwer Academic, Dordrecht, 7–31, https://doi.org/10.1007/14020-2265-4_2, 2004.
Citation: https://doi.org/10.5194/amt-2024-30-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
396 | 100 | 26 | 522 | 19 | 22 |
- HTML: 396
- PDF: 100
- XML: 26
- Total: 522
- BibTeX: 19
- EndNote: 22
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1