Review of the AMT manuscript by S. Brunamonti et al. with title:

SI-traceable validation of a balloon-borne spectrometer for water vapor measurements in the upper atmosphere

General comments

This paper is of interest for the AMT audience for several reasons: 1) It demonstrates that laser absorption spectrometers are a promising alternative to cryogenic frost-point hygrometers under the particularly difficult measurement conditions encountered in the upper atmosphere (variable pressure, low and variable water concentration). 2) It gives a blueprint of a laboratory validation for such measurement devices. 3) It shows the importance of choosing a more advanced (compared to the Voigt profile) line shape model (here the qSDVP) and demonstrates how to obtain the necessary parameters, which are not contained in the HITRAN data base.

The experiment is well-designed and the underlying measurements and data analysis are presented in detail, allowing fellow scientists to reproduce the measurements if they wish.

The paper is well written: It is easy to follow the authors through the chapters, thanks to the clear structure, the precise language and the supporting figures.

Specific comments

1 Introduction, p2, l25; State that only pressure broadening parameters are assessed and the line strength is not (since only pressure was varied, while temperature was held constant). This was not immediately clear to me when first reading the paper.

2.3 Gas handling system, p4, l13; Flow rates not consistent with what is stated in Figure 1 (0.05 to 3 slpm vs 0.05 to 4.5 slpm)

Figure 1; Should there not be a vent somewhere between the mixing solenoid valve and MFC 3? If not, where does the excessive flow go? Please clarify this.

p4, l19; how was the temperature controlled?

P4, l25; it would be interesting for fellow scientists trying to reproduce the experiments to know how the gas cylinder used for the secondary reference gas mixture was “conditioned” and what kind of synthetic air (upper boundary of water content?) was used to prefill and pressurize the bottle.

P4, l28; it would also be good to specify the type of the SilcoNert®2000-coated stainless-steel cylinder into which the reference gas mixture was expanded. Despite the coating, surface effects might be different for different bottle sizes.

3.2 Pre-processing, P9, l2; I imagine that the spectral range was about 0.845 cm-1, so you obtain a spectral-point resolution of 1.69 10^(-4)  for your stated 5000 datapoints? Please mention the spectral range to make clear how you got to the spectral-point resolution.

3.4. Quadratic speed-dependent Voigt profile (qSDVP), p13, l14; I do not agree with the interpretation of the residuals obtained with the qSDVP as being free of any structure exceeding the normal noise level, although admittedly the features visible in Figure 6c and d are very small. Have you tried to see if an additional fit parameter (D2 != 0) or a higher order line shape model would be able to suppress this feature? If so, it would be nice to mention this. In any case, in my opinion it would be preferable to state that there is still some small structure observable, but that it is largely reduced compared to the VP. Stressing the argument that the QF obtained with the improved fit equals the SNR, one could nonetheless justify the choice of the qSDVP as line profile?

4.3. Linearity (extended-range validation), p18, l14; What about the repeatability of this measurement? Have you determined the ~180ppm concentration several times? From Figure 9b it looks like all measurements (except for the 180ppm one) are slightly too high, as if the undiluted gas actually had a slightly higher water concentration…

1. Conclusion, p21, l10; “without systematic residuals”, I do not agree, as discussed above.

Technical corrections

P9, l15, replace “while secondary reference gas mixtures (panel d) at 3 pressure levels (60−200 mbar)” by “while secondary reference gas mixtures (panel d) were measured at 3 pressure levels (60−200 mbar)”

P 14, l4, delete double “the” in “while the the qSDVP fitting uses”

The color-code used in the figures for different pressures (Figure 3, 4, 6, 7) should be unified. A different color scheme should be used, since the different blue lines are hardly distinguishable.

P 25, l15-26; references are not in correct order (K after L)

Very nice work and paper, clearly exposed and overall rigorous in contents and discussions.

First of all, I fully agree with all comments, questions and suggestions of the other two referees.

In particular, as already underlined, this works illustrates and confirms the interest of using laser absorption spectroscopy in particular in the mid infrared, to obtain high accuracy and selective measurements of atmospheric species. The presented instrument additionally possesses the uncommon feature of an open absorption cell which allows fast renewal of the atmospheric sample with minimal surface effects, especially valuable with sticky molecules such as water. This advantage is clear when looking at the long transient times (hours) observed when closing the cavity as is done for this study. About this point, it would be nice that the authors add a short discussion about the fact that the performance of the instrument can be considered the same when the cavity is open to ambient air. In particular, are there any turbulence effects which add noise to the measurements, or thermo-mechanical deformation of the optical setup which may increase measurement drift relative to what obtained here?

This work also illustrates that the Voigt line profile is largely inadequate for an accurate description of collisionally broadened molecular absorption lines. This is actually a well-known fact since the time that high-resolution spectra are being obtained by using narrow laser spectral sources - compared to spectra obtained at lower resolution with traditional Fourier or grating spectrometers. Nonetheless, it is very instructive to see the impact of the choice of spectral lineshape on the retrieval of molecular mixing ratios, and to see that linear and accurate results may be obtained by using an advanced line shape model - with parameters determined by a multiline fit at several pressures, as shown in this work.

I have only one criticism concerning the long-term stability. The Allan-Werle stability analysis was performed for timescales only up to 500s (8 minutes). However, the importance of AW stability plots goes beyond the determination of the time for which a minimum AW deviation is attained, which defines the time for optimal averaging used in this work. The behavior of AW deviation at longer times provides essential information on instrumental drift. Do the measurements keep falling close to the optimal value or else do they drift away, and by how much and over how long? The AW deviation at long time scales provides this information on long-term stability of measurements and allows to assess the need for a re-calibration in case measurements are required to a level below the long-term drift. As it may be complicate to run an AW plot over more than one day, one can replace that by taking individual measurements averaged over the optimal time periodically and during a period which may be representative of the duration of a measurement campaign (one or 2 weeks). Besides, in the present case, long term drift may explain some observed effects. For instance, if the measurement at 180 ppm used for determination of the amount fraction of the secondary reference was performed a sufficiently long time before the measurements for the other amount fractions shown in figure 9, this might explain their systematic positive offset from the expected values.

Specific comments/questions:

1. What is he finesse of the Ge etalon used for spectral calibration of the laser scans? Is it just an un-coated flat Ge slab, or does it have reflective coatings on its faces?

2. Concerning the spectral resolution, the authors discuss briefly the “point resolution” and how that is fixed by the number of data points over a laser current scan, and their group averaging, however they should address the more fundamental question of the spectral resolution afforded by the line-width of the laser. Is it narrower than the point resolution? Do they observe any excess noise in correspondence with the sides of an absorption line, before averaging several scans? At low pressure where the line sides are steepest, and at high H2O content for better S/N, excess noise relative to the spectrum baseline could reveal the effect and magnitude of the laser line-width.

3. Any physical or practical reason for fixing Delta₂=0 in the fit ?

4. In table 1, the caption states that the relative uncertainty levels vary between 1.4-1.47% for all conditions but that does not seem to hold for the first point in the table (0.04/2.51=1.59%)

Review

Brunamonti  et al  AMT  2023  83            

SI-traceable validation of a balloon-borne spectrometer for water vapor measurements in the upper atmosphere

AMT  2023  83

The authors present a strong, detailed, extensive study on a performance evaluation and optimization of their balloon-borne MIR-TDLAS-hygrometer ALBATROSS, which is definitively  worth publishing.

The paper deals with a topic in atmospheric science, which has remained to be important for decades:  The accurate determination of the water vapor amount fraction in the upper troposphere / lower stratosphere (UT/LS ).  Water vapor is causing the largest part of the greenhouse effect, is a e.g. key player in the formation of clouds (which strongly influence earth’s radiation balance as well as atmospheric chemistry, particular at high altitudes), and is one of the key reactants in atmospheric chemistry by being the main source for the OH radical – the key atmospheric “cleaning agent”. 

The authors have selected tunable diode laser absorption spectroscopy (TDLAS) as analytical method and developed a particularly compact and light-weight, open-path (hence gas-sampling-free) TDLAS-system (named ALBATROSS, described elsewhere) based on mid-IR (6,01µm = 1663,9 cm-1) Quantum cascade lasers (QCL). The spectrometer is specially designed for balloon-borne high-altitude measurements. Instead of “Herriott”- or “White”-type Multipath absorption cells (H-MPC / W-MPC) - known from many earlier TDLAS hygrometers - ALBATROSS uses a segmented Circular-Multipass cell (sC-MPC) with a “flat/2D” star-shaped beam pattern, a quite low optical volume (140 cm³) and a cell path length of 611 cm.  

The key points of the paper claimed by the authors are :

1. SI-Traceable validation of ALBATROSS (using a metrologically validated permeation- and dynamic-mixing-based water vapor source)
2. Evaluation of ALBATROSS’ accuracy/precision, incl. pressure + amount fraction dependence
3. Comparison of 2 spectroscopic line shape models in their influence to ALBATROSS’ performance

I find the paper certainly deserves publication also as it is one of the very few studies where a metrological evaluation of a TDLAS hygrometer is targeted. 

However, as is will discuss below, I think a careful revision of the paper and the claims of the authors is needed:

Introduction: Due to the significant importance of  the  topic and the  long lasting efforts of the airborne hygrometer community dating back way into the 1980s and 1990s I think the introduction should be revised  to  include essential representative and  important work in airborne hygrometers, e.g., as FPH are mentioned as golden standard the review paper by Hall https://doi.org/10.5194/amt-9-4295-2016 should be relevant and mentioned. Further, the FISH hygrometer by C Schiller, reviewed by M Krämer in https://doi.org/10.5194/acp-15-8521-2015, is one of the most extensively used airborne hygrometers and one of the key reference instruments in AQUAVIT, and should be taken into account and mentioned.  Also there are plenty of airborne TDLAS hygrometers instruments e.g. by Sargent https://doi.org/10.1063/1.4815828, dating  back in the  90s  by Durry https://opg.optica.org/ao/abstract.cfm?uri=ao-38-36-7342        or Scott and Herman https://opg.optica.org/ao/abstract.cfm?uri=ao-38-21-4609. Particular relevant should be high flying ballon-borne open-path direct TDLAS hygrometers previously used for UT/LS sounding: CHILD by  Gurlit et al https://opg.optica.org/ao/abstract.cfm?uri=ao-44-1-91 should be cited here. There is also work on new open-path hygrometers based on cMPC e.g. by Witt et al https://www.mdpi.com/2076-3417/11/11/5189  which  I think should also be mentioned. 

Particular stratospheric H2O vapor accessed via balloon platforms is strongly influenced by photochemical conversion of CH4 to H2O, asking for the need to simultaneously monitor traces of H2O and CH4, which is covered by some balloon sensors e.g.  https://opg.optica.org/ao/abstract.cfm?uri=ao-44-1-91 . A fact which should be considered and mentioned too.

The dominant topic of the paper is hygrometer validation. The community has realized in the past a few fundamentally different type of validations: A) Field comparisons (Problem: lack of repeatability and lack of boundary parameter control) ; B) “Lab-like” parallel comparisons, e.g. AQUAVIT, (Problems: maintaining  identical or at least comparable measurement and sampling conditions for all instruments and implementation of metrological references), and  C) Rigid single instrument validations, preferentially to a SI-traceable water vapor source (Problem: large total effort, lack of H2O sources  suitable for atmospherically  relevant conditions, i.e. accurate definition of trace H2O levels - variable low gas pressures - low air temperatures). These differences should also be part of the introduction in order to avoid comparing “apples with oranges“. In addition to AQUAVIT other validations of the above categories should cited / taken into account/ analyzed, e.g:       

Buchholz + Smit > field comparison https://doi.org/10.1007/s00340-012-5143-1;             

Filges + Gerbig > field comparison  https://doi.org/10.5194/amt-11-5279-2018;               

Buchholz + Ebert > metrological standard https://doi.org/10.5194/amt-11-459-2018 ;  

Buchholz > metrological primary standard https://doi.org/10.1007/s00340-014-5775-4

For AQUAVIT (which was the largest parallel hygrometer comparison under variable p-T-H2O conditions), the authors should not highlight the (insufficient) performance of very young - not matured – instruments ( “exceeding 100%”) and give their underperformance the same weight like the very mature CORE hygrometers, which have been used and improved over decades. The performance of the non-calibrated, absolute, open-path TDLAS “APicT” in AQUAVIT certainly also relates well to the paper here, and could be mentioned in the paper. Some of the main findings of AQUAVIT were indeed the still quite large total discrepancies (-+ 10% relative) between the very mature “core” hygrometers. Also it took a complicated decision making process to define a “comparison reference” i.e. a suitable metrological H2O source or metrologically validated reference instrument which is compatible with the special (low temperature) boundary conditions of the AIDA chamber and their  huge size.

Experimental:

The authors target a SI-tracebale validation of ALBATROSS, where ALBATROSS is claimed to be a gas sampling-free, and calibration-free open-path Mid IR spectrometer.

The open-path approach promises to avoid H2O adsorption problems. However, open path also causes a very complicated tradeoff in system design, due to the complete lack of “sample control” during field-use, so that gas pressure, gas temperature, residence time, sample homogeneity must be measured, evaluated or assumed.  Additionally p and T are often not measured within the optical sample volume but outside of it, leading to further heterogeneity errors in measured p and T wrt to the gas sample.

In my understanding a validation of the open-path ALBATROSS was not tackled or described in the paper. Instead a closed-path version of ALBATROSS was used, which I think, is a big change with respect to the initial claim. Of course, the validation of the closed-path version is highly important and demanding, but closed-path studies are certainly not fully sufficient to validate the open-path version and certainly not under UT/LS field conditions. The title of the paper is therefore misleading and should be considered to be changed ( e.g. > “validation of  a  closed-path Albatross”).

In order to deduce the performance of the open-path version careful consideration and evaluation of  p, T sensor location and calibration is needed, which is however not given in the paper. Witt et al  in https://www.mdpi.com/1424-8220/23/9/4345 recently evaluated a comparable open-path C-MPC under dynamic situations and  found considerable systematic deviations caused by spatial gas temperature inhomogeneity and by the un-even statistical spatial weighting caused by the special C-MPC beam pattern. This findings are probably of high relevance for open-path in-field-use as well as for high-accuracy validations in closed-path cMPCs  as  presented in this manuscript by the authors.

The authors aim on calibration-free first-principles evaluation of the hygrometer signals, which is indeed a very powerful capability for field use (see e.g the  airborne HAI Hygrometer). In a cal-free mode, however, the TDLAS-instrument integrates any H2O spectral absorption over the full light path i.e. anywhere between the laser chip and the detector chip. Any “parasitic” = unwanted water along the absorption path “outside of the absorption cell” will lead to systematic, potentially drifting offsets and needs to be carefully evaluated and removed. Particularly complicated are situations where the gas pressure also is heterogenous along the path (e.g in sealed laser or detector housing). This problem is carefully described in Buchholz 2014 https://iopscience.iop.org/article/ 10.1088/0957-0233/25/7/075501. How this is solved / or avoided in the present study must also be described, in particular if ALBATROSS is claimed to be cal-free. It is unlikely that this problem is completely absent in the  ALBATROSS design. Parasitic water vapour offsets can of course be removed to first order by calibration, but not in a cal-free TDLAS hygrometer. 

With respect to this topic it should also be analyzed where the zero air blank values (e.g. 1.46 ppm in Fig 2 compared to 0.59 ppm in fig 9 ) comes from, how stable they are and e.g  how much of this is caused by  parasitic water in the spectrometer itself.

In the spectral evaluation a “spectrum normalization” via a division through an “empty cell spectrum” is used. As the “empty cell” still had non-stable “zero” water levels of 1,5 ppm, the spectrum normalization actually also introduces an effective offset (and to a certain extend a parasitic water vapor ) correction. This approach and the offset correction cannot be used in the open-path configuration. The alternative approach “polynomial baseline reconstruction” does not provide offset correction so that the parasitic contributions  should be effective. The authors should add data on this if possible or discuss this  effects and their quantitative influence on the absolute accuracy of  both ALBATROSS  versions.

Used preparative water vapor references: 

Primary Permeation source:  It should be better clarified which components in the entire setup (fig 1) define the “SI-traceable permeation source”. Is this the permeator only, or the permeator and  MFC1 and 2, or even more components? This needs to be clarified as only this subsystem provides the property of being SI-traceable. As I see it now, the “permeation source subsystem” is embedded into a larger gas mixing system containing further MFCs plus an additional pressure controller(s?), pressure sensors, gas and cell body temperature measurement. These all should be shown on fig 1. and better explained in the text. For the entire validation to be “SI traceable” all relevant measurement data need to be SI traceable. Traceable calibration data and accuracy and expanded uncertainties should be provided for all measurement parameters (p, T, flow etc) required for the TDLAS evaluation procedure, which is not the case. Figure 1 lacks also an excess flow outlet before MFC3. 

The absolute accuracy and stability of the reference H2O concentration and the gas handling system will influence the TDLAS validation and e.g. depends on accuracy and stability of the H2O blind value, which needs to be determined and should be given in the text.

Due to the lack of traceability information for the used validation setup I can’t see that the “entire validation setup” is SI traceable. Due to this deficit, the paper claims and the title should better changed to, e.g  “Validation of a closed-path balloon-borne spectrometer with a permeation-based SI traceable H2O-source”.

Secondary water standard: The bottled H2O mixture generated is analyzed (if I got it right) only by the closed-path ALBATROSS. Hence the assigned bottle value of 181 +- 0.06 ppm “collects”  all uncertainties (and all systematic errors) from the closed-path Albatross validation using the primary permeation source. The +-0.06 ppm (=3,3 E-4 relative!) can thus only be “precision”. Here the accuracy and the uncertainty of the bottle assignment should be added and discussed, which then needs to be taken into account for the “expected H2O amount fraction” in fig 9, and for the evaluation of the uncertainty of the linearity relation. Looking at the fitting function in fig 9 the differential linearity and the 1,008 slope seem certainly excellent. However, the large offset of 590 ppb (which is very close to 2% ! at 30 ppm and would extrapolate to 15% at tropopause concentrations of 4 ppm) definitively needs further explanations by the authors. For me this indicates an accuracy problem of closed-path ALBATROSS or/and this secondary standard setup. Also for both values (m and b) uncertainties should be provided.

Other points to be considered are the likely dependance of the H2O amount fraction form the bottle pressure, as well as sampling influences by the sampling line including the pressure reducer, more information on the sampling system and the adsorption minimization would be helpful.

Taking all this into account, bottle-based secondary standards might be useable as a high concentration H₂O source, but to be useful to a broader community they certainly need more evaluation work.

Spectroscopic retrieval:

The spectroscopic retrieval section is quite extensive and specialized for publication in AMT.

In my view the full fitting model is not sufficiently described: It is not clear how many and which water lines (or other interfering species) are fitted, or e.g how large the pressure dependent influence from neighboring lines is and how and if it is compensated. Which H2O isotopic composition is assumed?  An H2O stick spectrum showing the fitted as well as the ignored lines would be helpful here. Also the description of the physical model behind the spectral evaluation and in particular a complete set of input parameters and their total uncertainties is not given. A total uncertainty evaluation of a cal-free system seems therefore not possible.

If the cal-free evaluation is the goal, then all spectral parameters plus all auxiliary measurements needed  (= p, T, L  …. ) must be stated with their (expanded) accuracies/uncertainties. Here I would expect an uncertainty table for all input parameters, as well as more information on p-, T-sensors their location and traceable calibration, which is not given.

Also the uncertainty influence of the fitting process itself as well as e.g. the uncertainties of the linearization of the spectral axis/laser tuning  should be discussed.

Particularly in gas spectrometers the real gas temperature in a weakly thermally conducting low pressure gas can cause  problems. Concerning the HMP110 used here: this T sensor is specified by the manufacturer with an accuracy of 0.4K (not 0,2K) for the extended T range (needed for UTLS use). Also comments by the authors are recommended if/how they deal with the systematic T-offsets /uncertainties (and the effect in the TDLAS evaluation) caused by invasive air temperature measurements, i.e. evaluation of PT100’s self-heating and thermal gas to sensor transfer problems (particularly at low pressure). EURAMET project 1459 “Air Temperature Metrology – ATM” could be considered here.

The magnitude of the “temperature problem” also strongly depends on the spectral line selection: H2O line identification and lower state energy of the fitted lines therefore should be given in the paper.

The influence of individual spectral data uncertainties can be quite large and often strongly limits the achievable total uncertainty of cal-free spectrometer realization. As I understand the authors paper, they are taking fixed line strength S(296K) and broadening  G₀ for the spectral evaluation from HITRAN, further they need T dependance of broadening and S (which also comes from HITRAN with their uncertainties) and the line pressure shift Do  (again HITRAN + uncertainty) and then finally the “new” qSDVP braodening parameter G₁ (which also needs to get an uncertainty  from the  parametrization). With typical HITRAN uncertainties for S 1-10 % (depending on the line selection) Voigt broadening (another U= 2-5 % ) , broadening T coefficient (5 -20% and more ), plus  p, L, T, fit process, tuning and spectral axis uncertainties  it certainly  takes  further explanations how the closed-path ALBATROSS reaches 1.5% total uncertainty in cal-free mode. The best short-term accuracy can certainly be achieved by a hygrometer calibration to a very good reference and not via a  spectroscopic cal-free approach, due to the large amount of spectral input parameters with fairly  large uncertainties. 

The transfer of the closed-path validation presented in the paper to the open-path balloon-version, depends particularly strong on the accuracy of the spectral data i.e. H2O line selection  or the temperature coefficients of the broadening. What measurements this requires and  how  this  should be described is shown e.g in Pogany et al for traceable H2O strength https://doi.org/10.1016 /j.jqsrt.2015.06.023 and in Nwaboh https://doi.org/10.3390/app11125341 for traceable determination of H2O broadening incl T dependence for TDLAS. 

Line shape study:           

The authors compare the applicability of two line shape models: Voigt (VP) and quadratic speed dependent  Voigt (qSDVP) and then optimize the qSDVP approach.            

Their VP evaluation is not very extensive and based on a single fixed set of parameters taken from HITRAN: The to be expected pressure dependent line shape deficits are not taken care of. It should be noted that Buchholz AMT 2018 had proposed to correct this parametrizable, perfectly long-term stable,  systematic deviation caused by the Voigt profile deficits e.g via a look-up table approach. This correction approach allows faster fitting and avoids too many fit parameters, which have caused in his spectrometers noise-like fitting instabilities.         

For the qSDVP evaluation the authors use a restricted qSDVP parameter set, allowing only one additional “broadening parameter”, and hence should be better called “simplified qSDVP” to be precise.  As I understand the paper, the authors use the Albatross-permeation standard-comparison via an iterative parametrization to “determine” the “optimal” qSDVP broadening parameters for their setup (while checking S). The parametrization of the width parameter of the simplified qSVDP follows two goals A) to match the spectrometer response function (Albatross H2O concentration) and the reference concentration (permeation source plus mixing system) and B) to minimize the fit residual i.e. to remove systematic deviations in the line shape fitting (optimization of QF).  However, there is no uncertainty provided for the outcome of this process, leaving it open, how accurate this qSDVP parameter really is. Literature comparisons of this spectral parameter are also not given, making it somehow an instrumental parameter.  

As the goal of the parametrization was to improve the “correlation” between the permeator and the TDLAS system, it is “no surprise” that the parametrized qSDVP evaluation yields pretty much the “input data set”, while the QF optimization improves the apparent system precision by minimizing the fit residual. However, the problem I see is, that the reference permeator information was used twice: First for the determination of the spectral information and then the “trained” qSDVP-TDLAS was compared to its previous reference in the learning situation. And the result of the second step is not really surprising, it’s a  pretty good match.  

For me this approach seems essentially like a more elegant way to calibrate the spectrometer response function by using the reference H2O concentration. The uncertainties of this process are not sufficiently discussed. Also the high correlation caused by that approach is not studied or taken care of. An elaboration of this problem would require a further comparison with a third independent preparative or analytical H2O system, which has not been shown in the  paper. 

Therefore I think that the authors cannot claim a demonstration of a calibration-free hygrometer.

Not in closed-path configuration and even less so with an open-path cell.

Essentially they have developed a novel (?) calibration procedure instead.

In contrast to a classical calibration they are not aiming on a direct correction of the instrument function but realized a “physics-informed approach” to remove line shape deficits.       

This is also valuable(!) but it remains a calibration process.

Also I think further work is needed to investigate the accuracy and (longer term) stability of this parametrization / parametric calibration, and how often it needs to be repeated.

In essence the paper contributes a wealth of highly interesting and valuable data!

But I think that the claims derived from the data and the results should be carefully and conservatively revised. What I see is a “Use of a SI-traceable permeation source for the characterization/calibration of a closed-path Mid-IR QCL TDLAS hygrometer suitable for balloon-borne, extractive UTLS-hygrometry” 

overview over the above proposed additional references

• Hall, E. G., Jordan, A. F., Hurst, D. F., Oltmans, S. J., Vömel, H., Kühnreich, B., and Ebert, V.: Advancements, measurement uncertainties, and recent comparisons of the NOAA frost point hygrometer, Atmos. Meas. Tech., 9, 4295–4310, https://doi.org/10.5194/amt-9-4295-2016 , 2016.
• Meyer, J., Rolf, C., Schiller, C., Rohs, S., Spelten, N., Afchine, A., Zöger, M., Sitnikov, N., Thornberry, T. D., Rollins, A. W., Bozóki, Z., Tátrai, D., Ebert, V., Kühnreich, B., Mackrodt, P., Möhler, O., Saathoff, H., Rosenlof, K. H., and Krämer, M.:
  
  Two decades of water vapor measurements with the FISH fluorescence hygrometer: a review, Atmos. Chem. , 15, 8521–8538, 2015., https://doi.org/10.5194/acp-15-8521-2015
• R. Sargent, D. S. Sayres, J. B. Smith, M. Witinski, N. T. Allen, J. N. Demusz, M. Rivero, C. Tuozzolo, J. G. Anderson; A new direct absorption tunable diode laser spectrometer for high precision measurement of water vapor in the upper troposphere and lower stratosphere. Rev Sci Instrum 1 July 2013; 84 (7): 074102. https://doi.org/10.1063/1.4815828
• Georges Durry and Gerard Megie, "Atmospheric CH₄ and H₂O monitoring with near-infrared InGaAs laser diodes by the SDLA, a balloonborne spectrometer for tropospheric and stratospheric in situ measurements," Appl. Opt. 38, 7342-7354 (1999) https://opg.optica.org/ao/abstract.cfm?uri=ao-38-36-7342
• David C. Scott, Robert L. Herman, Christopher R. Webster, Randy D. May, Gregory J. Flesch, and Elisabeth J. Moyer, "Airborne Laser Infrared Absorption Spectrometer (ALIAS-II) for in situ atmospheric measurements of N₂O, CH₄, CO, HCL, and NO₂ from balloon or remotely piloted aircraft platforms," Appl. Opt. 38, 4609-4622 (1999) https://opg.optica.org/ao/abstract.cfm?uri=ao-38-21-4609
• W Gurlit, R Zimmermann, C Giesemann, T Fernholz, V Ebert, J Wolfrum, U Platt, and J P. Burrows, "Lightweight diode laser spectrometer CHILD (Compact High-altitude In-situ Laser Diode) for balloonborne measurements of water vapor and methane," Appl. Opt. 44, 91-102 (2005) https://opg.optica.org/ao/abstract.cfm?uri=ao-44-1-91
• Witt, F.; Nwaboh, J.; Bohlius, H.; Lampert, A.; Ebert, V. Towards a Fast, Open-Path Laser Hygrometer for Airborne Eddy Covariance Measurements. Appl. Sci. 2021, 11, 5189. https://doi.org/10.3390/app11115189 https://www.mdpi.com/2076-3417/11/11/5189
• Buchholz, B., Böse, N. & Ebert, V. Absolute validation of a diode laser hygrometer via intercomparison with the German national primary water vapor standard. Phys. B 116, 883–899 (2014). https://doi.org/10.1007/s00340-014-5775-4 https://link.springer.com/article/10.1007/s00340-014-5775-4
• Buchholz, B. and Ebert, V.: Absolute, pressure-dependent validation of a calibration-free, airborne laser hygrometer transfer standard (SEALDH-II) from 5 to 1200 ppmv using a metrological humidity generator, Atmos. Tech., 11, 459–471, https://doi.org/10.5194/amt-11-459-2018 , 2018.
• Filges, A., Gerbig, C., Rella, C. W., Hoffnagle, J., Smit, H., Krämer, M., Spelten, N., Rolf, C., Bozóki, Z., Buchholz, B., and Ebert, V.: Evaluation of the IAGOS-Core GHG package H2O measurements during the DENCHAR airborne inter-comparison campaign in 2011, Atmos. Meas. Tech., 11, 5279–5297, https://doi.org/10.5194/amt-11-5279-2018 , 2018.
• Buchholz, B., Kühnreich, B., Smit, H.G.J. et al. Validation of an extractive, airborne, compact TDL spectrometer for atmospheric humidity sensing by blind intercomparison. Appl. Phys. B 110, 249–262 (2013).https://doi.org/10.1007/s00340-012-5143-1 https://link.springer.com/article/10.1007/s00340-012-5143-1
• Witt, F.; Bohlius, H.; Ebert, V. Evaluation of Spatial Gas Temperature and Water Vapor Inhomogeneities in TDLAS in Circular Multipass Absorption Cells Used for the Analysis of Dynamic Tube Flows. Sensors 2023, 23, 4345. https://doi.org/10.3390/s23094345 https://www.mdpi.com/1424-8220/23/9/4345
• B Buchholz V Ebert Offsets in fiber-coupled diode laser hygrometers caused by parasitic absorption effects and their prevention May 2014 , Measurement Science and Technology 25(7):075501 https://iopscience.iop.org/article/10.1088/0957-0233/25/7/075501
• A Pogány, A Klein, V Ebert, Measurement of water vapor line strengths in the 1.4–2.7µm range by tunable diode laser absorption spectroscopy, Journal of Quantitative Spectroscopy and Radiative Transfer, 165, 2015, 108-122, https://doi.org/10.1016/j.jqsrt.2015.06.023. https://www.sciencedirect.com/science/article/pii/S0022407315002290
• Nwaboh, J.A.; Werhahn, O.; Ebert, V. H₂O Collisional Broadening Coefficients at 1.37 µm and Their Temperature Dependence: A Metrology Approach. Appl. Sci. 2021, 11, 5341. https://doi.org/10.3390/app11125341   https://www.mdpi.com/2076-3417/11/12/5341