the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Nov 2024
Correcting for water vapor diffusion in air bag samples for isotope composition analysis: cases studies with drone-collected samples
Abstract. Traditional methodologies, such as mass spectrometry and laser spectroscopy, have been widely employed for precise water vapor isotope measurements. Nevertheless, these techniques are limited by logistical challenges in fieldwork, consequently constraining the temporal and spatial resolution of measurements. Specifically, water vapor isotope measurements are primarily limited to near-surface levels, while measurements associated with processes aloft connecting tropospheric water vapor to surface precipitation are notably scarce. Portable sampling devices, such as air bags and glass bottles, have therefore become necessary alternatives for collecting, storing, and transporting gaseous samples in diverse environments prior to analysis with less portable instruments. In drone-based high-altitude vapor sampling, air bags are preferred for their lighter weight and greater flexibility compared to glass bottles. Nevertheless, they present specific challenges, such as potential sample contamination and isotopic fractionation during storage, primarily due to the inherent permeability of air bags. Here, we developed a theoretical model for water vapor diffusion through the sampling bag surface, with parameters calibrated through laboratory experiments. This model enables the reconstruction of the initial isotopic composition of sampled vapor based on measurements obtained within the bag and from the surrounding environment. We applied this correction method to air samples collected at various pressures up to the upper troposphere using an air bag-mounted drone that we developed, thereby estimating the initial isotopic composition and uncertainty based on our observations. The corrected observations closely match the IASI satellite data. Our correction method significantly enhances the reliability and applicability of water vapor isotope observations conducted using drones equipped with air bags. This approach leverages the strengths of drone-based air bag sampling while mitigating its limitations, thus facilitating the convenient collection of isotopic data throughout the troposphere.
- Preprint
(2010 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 03 Jan 2025)
-
RC1: 'Comment on amt-2024-151', Anonymous Referee #1, 16 Dec 2024
reply
This manuscript describes methods by which to interpret isotope and humidity data from drone-mounted permeable gas sampling bags. The conceptual model appears to be generally useful for inferring atmospheric conditions based on gas composition inside sampling bags. However, the conceptual derivation of the model should be more robustly described, improving precision of communication but especially considering all relevant effects such as temperature and pressure. There is also a lot of improvement needed in precision of notation and terminology, such as the definition of diffusion, silent substitution of delta notation for isotopic ratio in a key equation, and others that I have noted below in the detailed comments. Finally, the actual sampling techniques are not described, so the example drone flight profile is difficult to interpret.
L90 differential diffusion does cause fractionation, but they are not synonymous.
L90 there are two relevant gradients causing fractionation: one in concentration of water (causing mass flux), and one in isotopic composition (resulting in no net mass flux). The latter can cause fractionation even if humidity is the same inside and outside of the bag. Neglecting the second gradient may be justified if it is small, but it should not be ignored completely by the theoretical derivation.
L125 I have several comments about Eq 1:
(1) it neglects effects of pressure and temperature differences across the bag membrane. (2) wouldn't it be more general to formulate this equation in terms of partial pressure of water vapor instead of mass concentrations? That would partially resolve #1
(2) L124 It's not flux toward the bag, but into the bag, correct?
(3) defining k in g/kg adds a potentally confusing dimensionality to k, so it is not dimensionless as might be assumed and so that F takes on the units of g/m2/s instead of the si base units kg/m2/s that readers might assume without reading carefully. Maybe there is a good reason for the choice; please tell us.
L133 and L141 the definitions of alpha and lambda are both crucial equations. Assigning them equation numbers would make them easier to find.
L139 the mass balance assumptions are not clear. How can M be constant if there is water flux into the bag? Does this mean we must assume that all vapor transport into the bag is balanced by an equal mass of non-water vapor transport out of the bag? Or maybe this theory only works if dM/dq is very small? What are the limits of this assumption?
L141 this definition of lambda can be loosely defined as a diffusion coefficient, but it is probably better termed (non-dimensional) conductance. It defines the rate of net mass flux in response to a gradient in concentration, but neglects gradients in pressure, temperature, and isotopic composition. A standard definition of a diffusion coefficient would have dimensions cm2/s. Further confusion arises in sec 3.2, where lambda is called "permeability" L218. Please choose consistent terms.
L167 it would be helpful to be extra clear here that the alphas being obtained are those due to fractionation due to mass flow through bags.
L181-183 there is a lack of clarity here in notation. Delta 18O and delta 2H do not appear in eq 11. Substituting delta notation for ratio notation has no consequences for the alpha in Eq 11 because the standard ratio cancels out, but it would be kinder to readers to justify the use of delta notation either by deriving Eq 11 in terms of deltas or to explain here that ratios of R and ratios of deltas are equivalent.
L202 flow rate is not measured in psi.
L209 "can” or "did"? And what is a parallel sample?
L210 I'm not following—which bias is this? I am guessing this is in the laser spec, but it would be nice to be clear.
L210 this paragraph is difficult to follow because it appears to use jargon specific to the piece of equipment used (but not fully specified—was it a Picarro A0101?).
Table 2.1 is not needed
L228 alpha_delta is an unfortunate choice in nomenclature. The standard variable is alpha, which can be made more specific by listing the isotopes involved (eg alpha H2/H1), but it adds only confusion to add "_delta" because the delta notation has nothing to do with the isotope fractionation factor. Alpha is defined in terms of isotopic ratios (i.e., not in terms of delta).
L232 injected how? Liquid? This is ~10^-2 ml, correct? L255 And 10^-3 ml in experiment 3? I'm surprised this was easier than using lab air and later adjusting the humidity or isotopic composition of a testing chamber.
L259 "can” or "did"?
L266-268 the logic of how the sampling system works is important. Full details might not be appropriate here, but a citation to them would be nice. At minimum I would expect an outline of how it works, given that understanding the results depends on understanding the methods.
L269 of course the bags do not deflate because of mass loss, but because of increased pressure outside the bag, and the pressing danger would therefore seem to be preventing ingress of new air, not egress of sample. Do the one-way valves protect against this?
L280 the vapor is not measured in situ. Samples are removed from their locations and measured elsewhere.
L287 this sentence illustrates what I mean in my comment L266: readers cannot appreciate the sampling environment in any useful detail, so there is no way to fully understand why, for example, "it is difficult to experimentally estimate λ for different altitudes".
L291 longer sampling time where? To collect the air or to analyze the samples?
In general, section 3.4.1 seems to all collapse to "mass was estimated proportional to pressure at the sampling altitude and pumping time". The overly detailed presentation makes the logic seem more complicated than it is.
L306 comment 1: what does "diffusion model correction process" mean? Parameterization? Correction of model structure?
L306 comment 2: I don't think lambda_surface and lambda_alt are proper variables. Lambda is a property of a bag that should not depend on altitude. Its apparent dependence on altitude in this work is due to errors estimating masses. Therefore, unless I am missing something, the better variable to report here as a source of error is the estimate of air mass.
L306 comment 3: mismatches between model and data are not sources of error, they are themselves the error.
Sec 3.4.2 the uncertainty section is difficult to follow and needs a revision for conciseness and clarity. The section includes too much information (e.g., how mean parameter estimates were obtained—i.e., nothing to do with uncertainty), is not well organized, and is also not always specific when it needs to be. An example of this last point is the ¼ estimate for pumping time. Is this really uncertain to that degree? It seems more likely (lacking actual experimental details), that pumping time is well known and the real variable is mass captured.
L376 this temperature (and pressure) dependence should be recognized in the theoretical development.
Fig 3 it would improve the accessibility if the caption told us which experiment these data come from
Fig 4a misspelling of Environment. Also, Environment should be defined in the main legend, not in each panel
Fig 4 I don't understand why the isotopic equilibration models cross over each other at long time. What equation exactly generates the solid lines model fits?
L398 should specify HD16O—or omit it, since the sentence is about 18O
Fig 5 delete "(a, b) (a-b)" at the beginning of the caption
Fig 5 is difficult to follow. What is "real value"--it looks like initial isotopic composition inside the bag, but why does it not change with time and why is it measured at different times compared to the colored dots? What equation exactly generates the solid lines model fits?
For figures 4 and 5, plotting the humidity inside the bags over time would help a lot in illustrating the processes and ensuring the models are describing the processes correctly.
For figures 5 and 6, I appreciate the idea to diagram the processes, but I found the diagrams unhelpful because they illustrate only the magnitudes of fluxes and ignore the crucial differences in humidity inside the bag.
L405 referring to Fig 4 as the first scenario and Fig 5 as the second and third scenarios is confusing, because they are not so labeled in the figures and difficult to keep straight in the text.
L431 again it would be helpful to be explicit about which equation is "the model"
L441 ok but (1) there are other ways to flag for unrealistic results, so focusing on this one seems odd; and (2) leaving in those six data points would presumably not have much effect on the results, so this detail seems distracting.
Minor global comment: the word "value” is redundant almost everywhere.
L456 "model corrections” can never affect d18O in the bag. I think this is saying that applying corrections for vapor pressure differential and fractionation by the bag changes the estimate of the atmospheric d18O. L454-459 why say all this twice? The correction process is the same for 2H and 18O, and the d response follows. This description makes it sound more complicated than that.
Fig 10 it is not clear what "Picarro” means here. Is this the measurement of ambient vapor at the surface at the time of sampling aloft? Or is this bag samples vs. satellite-inferred estimates?
L551 the methods in this manuscript have nothing to do with laser spectrometry; the samples could as well be measured by other methods.
Citation: https://doi.org/10.5194/amt-2024-151-RC1 -
RC2: 'Comment on amt-2024-151', Anonymous Referee #2, 18 Dec 2024
reply
Wang and others have developed a theoretical model to describe water vapour diffusion through the surface of a sampling bag. They calibrated the model's parameters using laboratory experiments. This model allows for the reconstruction of the initial isotopic composition of the sampled vapour by using measurements taken from both inside the bag and the surrounding environment. I believe it is an important work but not well explained and supported in several sections, and needs a through revision.
L67 Distinguish between aerial mobile measurements for water vapour isotopes and mobile water isotope measurements, i.e. picarro mounted on a van. The latter is quite common and does not require sample storage anymore.
L88-90 check usage; this should be isotopologues
L104 What diverse conditions?
L110-111 Picarro direct observations - as in measurement on the Picarro? And satellite data of what?
L125 State the boundary conditions and assumptions clearly under which equation 1 is valid
L130 Equation 2 should be based on isotope notations, and the isotopologue for which they apply can be in subscript. This is not clear yet.
L202 Convert the flow rate to volume per unit time
L228 Provide more details on the experimental setup, including the type of airbags used and the specific model of the Picarro analyzer
L252 Provide more details on the experimental setup, such as the specific amounts of water injected and the isotopic values used, to give a clearer picture of the conditions tested.
L277 Drone flight path and sampling strategy need to be better explained. Also, samples aren't measured in situ.
L305 I would suggest a net uncertainty or error propagation of some kind to be calculated and reported for
λ_surface, α_δ, λ_alt. Currently, this section does not explain the uncertainties (only how they are calculated) or how they affect the results.
Fig 4 How are the equilibriation lines intersecting? And mention which experiment generates this.
L441 explain why <1 permil is unrealistic
L460 I would explain this in the methods and bring it back in the discussion as a model sensitivity to its parameters
L485 How different are these storage times to really affect the measurements? Can this be incorporated as part of the correction in the model?
L490 This section has been introduced several times in the paper but is not discussed enough here. I would expect some prior information about why they may be different based on the remote sensing method but necessary to fit wider regions or global models. I would also expect the authors to mention other such repositories like TES and SCIAMACHY.
Fig 10 The left panels are of d2H, but the figure caption and subsequent discussion on d18O. I expect the satellite data to be that of d2H. What am I missing here?
Fig 10e Explain why, for higher elevation samples, the satellite dD differs more with measured/corrected data than other altitudes.
Citation: https://doi.org/10.5194/amt-2024-151-RC2
Model code and software
Correcting-for-water-vapor-diffusion Di Wang and Camille Risi https://github.com/DishiWANG0608/Correcting-for-water-vapor-diffusion.git
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 116 | 28 | 7 | 151 | 6 | 4 |
- HTML: 116
- PDF: 28
- XML: 7
- Total: 151
- BibTeX: 6
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1