the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 May 2024
Development of a Peltier-based chilled-mirror hygrometer for tropospheric and lower stratospheric water vapor measurements
Abstract. We have developed a Peltier-based non-cryogenic chilled-mirror hygrometer named “SKYDEW” to measure water vapor from the surface to the stratosphere. Several chamber experiments were conducted to investigate the characteristics and performance of the instrument under various conditions. The stability of the feedback controller that maintains the condensate on the mirror depends on the controller setting, the condensate condition, and the frost point in ambient air. The results of condensate observation by a microscope and proportional-integral-derivative (PID) tuning in a chamber were used to determine the PID parameters of the controller such that slight oscillations of the scattered light signal from the mirror and mirror temperature are retained. This allows for the detection of steep gradients in the humidity profile, which are otherwise not detected because of the slower response. The oscillation of the raw mirror temperature is smoothed with a golden point method that select the equilibrium point of the frost layer. We further describe the details of the data processing and the uncertainty estimation for SKYDEW measurements in terms of the Global Climate Observing System (GCOS) Reference Upper-Air Network (GRUAN) requirements. The calibration uncertainty of the mirror temperature measurement is <0.1 K for the entire temperature range from –95 to 40 °C. The total measurement uncertainty of SKYDEW measurements can exceed 0.5 K in the region where large oscillations of the mirror temperature remain.
Intercomparisons with relative humidity (RH) sensors on radiosondes, the cryogenic frost point hygrometer (CFH), and satellite Aura Microwave Limb Sounder (MLS) were performed at various latitudes in the Northern Hemisphere to evaluate the performance of SKYDEW. These results show that SKYDEW can reliably measure atmospheric water vapor up to 25 km altitude. Data from several SKYDEW and CFH measurements predominantly agree within their respective uncertainties, although a systematic difference of ~0.5 K between SKYDEW and CFH was found in the stratosphere, the reason for which is unknown. SKYDEW shows good agreement with Aura MLS for profiles that are not affected by contamination.
- Preprint
(2527 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-635', Anonymous Referee #2, 29 May 2024
General Comments
The authors of the manuscript titled ‘Development of a Peltier-based chilled-mirror hygrometer for tropospheric and lower stratospheric water vapor measurements’ describe SKYDEW, a new chilled-mirror hygrometer for weather balloons that measures water vapor from the ground up to about 25 km altitude using thermo-electric cooling. The maximum altitude is constrained by the instrument's ability to dissipate heat and by outgassing from the balloon flight train. The latter is a common issue for balloon hygrometers.
The instrument and its principle of operation are documented comprehensively in the text and in the figures, together with a well-funded uncertainty analysis. The data processing is inspired from a 'golden point' approach introduced recently by a Swiss research group. The processing departs from the traditional averaging techniques and appears to be well suited for SKYDEW’s aggressive PID controller, achieving frost point retrievals with a vertical resolution that I estimate from the figures about 50 m. It is unclear from the text how much of the processing is automatic and how much is based on user input. User input is not uncommon for this type of instruments.
SKYDEW is able to measure several kilometers into stratosphere, which is an improvement compared to Snow White, a thermo-electrically cooled balloon hygrometer of previous generation. The approach of using different operating set points for three different regions of the atmosphere is new, as far as I know, and is reasonable.
SKYDEW’s detector is located on top and outside of the instrument housing, similarly to the night version of Snow White. This design may allow for a better ventilation and heat dissipation and overall a faster response. On the other hand, the open design potentially increases the fragility of the sensor and may increase the risk of self-contamination when flown through saturated layers of the atmosphere, as the comparison with MLS retrievals shown in the text suggests. In addition, placing the sensor at the top of the instrument prioritizes ascent data over descent data.
It is unfortunate that only one dual sounding is used for the reproducibility evaluation presented in this paper. The unexplained difference in the stratosphere of about - 0.5 K (or ~ 10%) w.r.t. to the CFH reference is also unfortunate, albeit based only on a very limited number of soundings and only one with CFH and SKYDEW on the same balloon. It appears that a full evaluation of the measurement accuracy of SKYDEW will be only possible further down the road, when more data is available. Nevertheless, the results presented in this initial paper are encouraging and the calculated uncertainties seem reasonable.
The text is scientifically sound and within the scope of AMT, well structured, with some typing, grammar and referencing errors. While the presentation quality of some of the figures may be improved (especially in terms of dpi) they are readable and understandable.
The revision should correct technical errors and expand Section 3 with a few more clarifications.
In view of MLS decline and of the phase-down of hydrofluorocarbons, this presentation about an instrument capable of measuring lower stratospheric water vapor with thermo-electric cooling is relevant to the scientific community.
Specific comments
1: you may consider adding ‘SKYDEW’ into the title of the paper
51: consider rewriting ‘perform poorly in the dry stratosphere’. Operational radiosondes have been used in moist stratospheric conditions (e.g. Vömel et al. 2022).
67: ‘these remote sensing techniques’. Which other ones? You have only mentioned Raman lidar in this paragraph.
133 / Figure 1: Interesting design, with the sensor placed on top and outside the instrument, without inlet/outlet tubes. This departs from CFH/FPH. Could this explain the contaminated profiles shown in Figure 15, or is this due to contamination from the balloon and its flight train? On the other hand, the SKYDEW design may allow better ventilation and heat dissipation, and overall a faster response? Is this assumption correct?
What is the 'cover' in Figure 1 (c) and what is it made of? Is it used to guide the airflow over the detector? And/or to act as a protection against hydrometeors when flying in clouds?
The hot side radiator looks interesting. What is the purpose of the two ‘screws’ on the upper part of the radiator? For heat dissipation?
136: Figure 2 suggests a maximum achievable cooling of about -50 C, not -90 C.
139: What is the wavelength of the light source and what is the modulation frequency? What is the technology (e.g. LED + Photodiode)?
140: What is the overall size (cm x cm x cm) of the SKYDEW instrument?
146: The intensity of the scattered light is used in the processing and in the uncertainty analysis (Sections 3 & 4). Why do you consider it ‘housekeeping data’?
175: Incorrect, liquid water has a vapor pressure higher than ice, not lower.
204: Poor reference for specific humidity and precipitable water. In fact, this sentence may be removed entirely, as it provides no added value to the manuscript.
226: In Figure 2, do the dashed lines correspond to parametrizations using Equations 7 and 8? If so, please mention this and provide parameter values.
288: Do you mean here that during a phase transition, you can distinguish if the mirror temperature corresponds to the dew point or to the frost point solely from the behavior of the scattered light?
293: Is an intentional heating at -12.5°C also performed in NOAA FPH? Hall et al. 2016 do not mention it.
321: From what I understand, Equation (11) fits to the ‘golden points’ in Section 3, i.e. the condensate on the mirror neither grows nor shrinks at the frost point. What is actually the relationship between the (mean) size of the ice crystals and the scattered light intensity in SKYDEW? Consider writing a few sentences why the ‘golden points’ of Poltera et al. 2021 apply to SKYDEW's scattered light signal.
331: ice ‘crystals’, not ‘droplets’.
364: Is the removal of intentional heating data performed automatically by software?
383: This paragraph needs further clarification. Do you extract the ‘golden points’ on the smoothed mirror temperature, or from the original mirror temperature? From Figure 7 (upper left), it seems that the points are extracted from the original profile, but this is not how I understand the text. Moreover, how do you perform the final smoothing of the extracted ‘golden points’? They seem to come at irregular time intervals. Do you linearly interpolate between the ‘golden points’ on a 5 Hz time axis and then smooth with a Gaussian filter of sigma=1.5 seconds?
430: In Table 1, you write ‘ ~ 1.0 K at max’ for this uncertainty component, why not write this here? Moreover, what about the typical uncertainty value when oscillations are ‘normal’?
452: Intentionally heating the mirror first at -12.5 C, such as CFH, seems reasonable for an ascending balloon, as it almost certainly ensures that the mirror condensate is liquid water before the heating stage and ice after it. SKYDEW heats its mirror first at -36 C, i.e. at a temperature where the condensate has very likely already phase-transitioned to ice, which complicates the mirror condensate phase determination. Why not perform the first heating stage at a lower temperature in SKYDEW?
From this paragraph, it is unclear how you determine the phase of the condensate in practice. From the amplitude and frequency of the scattered light fluctuations? Or from the comparison with the partnering radiosonde RH sensor? Or both? Can this be performed automatically by software?
455: Incorrect, cubic ice on the mirror would cause a lower mirror temperature.
504: How large is the air temperature uncertainty uT? Please provide a typical value or an upper bound
511 / Figure 9 (h): the uncertainty due to pressure is surprisingly small compared to e.g., Hall et al. 2016 for FPH? Why is that? How large is the air pressure uncertainty uP? Please provide a typical value or an upper bound
567: You mention at the beginning that several SKYDEW-CFH comparisons have been performed, yet only two soundings are discussed, only one on the same balloon. Is the -0.5 K difference in the stratosphere systematic in all soundings so far, or does it appear only on these two soundings?
588: the MLS data has been averaged over how many +- hours or days?
596: The measurements took place during the summer monsoon, which probably increases the risk of contamination for this type of instruments. Nevertheless, do you know where the outgassing that you mention here comes from? Is it the SKYDEW sensor probe itself that suffered from icing in the saturated troposphere? Or is it from the balloon and its flight train? > 50 m train lines are generally recommended for measuring stratospheric water vapor, have you used such longer lines?
You mention that the soundings on 1 & 3 June passed through saturated layers, but what happened on 27 May? What is the reason for the disagreement with MLS above 20 km on that day? The 27 May sounding has for example not been excluded from the comparison with RS11G on Figure 14.
629: Why is controlled descent a challenge for SKYDEW? Is this because SKYDEW might lose its condensate before balloon turnover? Or is it because of the position and orientation of the sensor?
Technical corrections
Here a list of typing and referencing errors that I was able to spot, sorted by line number or figure number.
21: selects the equilibrium point
25: in regions
46; The Brewer-Dobson circulation
63: integration time
66: Whiteman et al. 2006 is missing from the reference list
89: Hurst et al. 2023
90: I have found the NOAA instrument description in Vömel et al. 1995 (10.1029/95JD01000), not Vömel et al. 1995 (doi:10.1029/95GL02940).
93: water vapor concentrations. Generally, stick to either American or British.
96: cooperation agreement
98: the Fluorescent Advanced Stratospheric Hygrometer
98: the Vaisala RS92 radiosonde
110: analog
123: Put the 3 in CFH3 in subscript
129: Section 5
129: Section 6
136: a temperature difference
150: Vaisala RS41, Intermet 54
162: You have two Vömel et al. 2016 references, please distinguish between the two in the text
172: You have three Vömel et al. 2017 references, please distinguish between the three in the text
174: Fujiwara et al., 2003
181: Vömel et al. 2007 mention -53 C, not -55 C.
182: You have three Vömel et al. 2017 references, please distinguish between the three in the text
193: WMO (2021)
216: between the mirror surface and ambient air, and S
220: thermal conductivity of air,
225: For higher air temperature
230: high cooling power is needed
235: (PT100) which has
251: the control output corresponds to the current
259: A proportional controller alone cannot eliminate
266: which depends
276: at an air temperature of ~0 °C and at ~-13 °C dew point.
285: This implies that the output of the PID controller
308: As in the cloud formation process, the number of IN on the mirror may reflect the temperature dependence of the ice-nucleation process on the mirror.
334: at an as low as possible
343: when the mirror temperature reaches about –36 °C.
406: remove underscore before ‘to reduce this distribution’
462: (Thornberry et al., 2011).
467: Thornberry et al (2011)
472: Pruppacher and Klett, 1997
484: Kizu et al. (2018) is missing from the reference list
492: in long time series
540: A dual sounding with two SKYDEWS
543: worked properly.
548: Smaller oscillations arerequired for better measurements, although the oscillations due to the aggressive setting of the PID controller are needed to detect the golden points.
559: The deviation is large in the troposphere
611: dynamic changes of atmospheric water vapor.
615: The Peltier cooling creates a temperature difference of more than 40 K
726: Missing new line between ‘2022’ and ‘Sakata, R.’
744: Vömel
794: which corresponds to the right axis.
796: Photographs show the condensate on the mirror.
799: Condensates on the mirror at air temperatures of
819: the detected golden points
831: Profile of dew/frost point
837: Frost point profile on 26 November 2021
863: The center panel shows
875: Gray shading indicates the uncertainty of SKYDEW (k = 2).
877: Mirror temperature measurement, u_mirror
878: Resistance measurement u_mrr2: missing ‘divide’ sign between 0.005 and \sqrt{3}
Figures 6 (b) and 9 (b): The sign of the Peltier current is inconsistent w.r.t Figure 2.
Figure 14 (a): The ‘black dashed line’ appears to be missing in Figure 14 (a). Please add the line in the figure, or remove this sentence from the figure caption.
Citation: https://doi.org/10.5194/egusphere-2024-635-RC1 -
AC2: 'Reply on RC1', Takuji Sugidachi, 11 Jul 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-635/egusphere-2024-635-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Takuji Sugidachi, 11 Jul 2024
-
RC2: 'Comment on egusphere-2024-635', Anonymous Referee #1, 03 Jun 2024
Summary:
The manuscript by Sugidachi et al. describes a new Peltier cooled frost-point hygrometer to measure stratospheric and tropospheric water vapor. This development expands to ability to measure stratospheric water vapor, which is desperately needed.
The manuscript describes some instrumental details as well as some comparisons and uncertainty estimates. It is heavily based on previous work on cryogenically cooled frost-point hygrometers and adopts many of the same ideas and principles.
The descriptions are often a little superficial and skip on several details. There appear to be many more profiles than the authors show here, which would allow them to do a statistical comparison between Skydew and CFH or Skydew and MLS. This seems to be missing and would add significantly to support the impact that this instrument may have. As is, the selection of profiles appears a little selective.
Below, I outline several points for improvement of the manuscript. I assume that these will take some time to implement. For these reasons I am hesitant to recommend publication of the manuscript without some major revisions.
Detailed comments:
In the description of the sensor the authors state that a temperature difference of 90°C can be achieved. Here they should clarify that this is the temperature difference between hot and cold side, not between ambient temperature and cold side. Later, they state, that the achievable temperature difference between ambient temperature and cold side can be > 55°C at the surface and 30°C at an ambient temperature of -70°C. In the summary they state a temperature difference of 40 K in the stratosphere. Figure 2 clarifies that these values only apply at a wind speed of 0 m/s. Adding wind of 5 m/s, a realistic temperature difference may probably by around 20°C smaller. Figure 12 seems to indicate a cooling limit of 25 K. In line 365, the authors state that an extreme limit at which Skydew cannot take measurements would be dew-point depressions larger than 60°C at the surface. Considering the statements above, that limit seems to be more than optimistic under real world flight conditions. It would be good to expand that discussion and consolidate the different statements throughout the manuscript.
Related to this, the upper altitude at which the Skydew sensor can measure is not well supported. The authors point out that there is a systematic difference between day and nighttime measurements. Is the 25 km limit (not considering contamination) for nighttime achievable for all geographic locations? Is this limit a maximum under optimal conditions, or an average? Is this an optimistic estimate?
In the abstract, the authors report a difference of 0.5 K between the CFH and Skydew. This does not properly summarize the results in the text and gives a wrong impression of the comparison between the two instruments. This difference was found in only one of two soundings they discuss. There are more comparisons between the two instruments, which are not discussed in the text.
There are over 40 soundings using Skydew with different development stages. The descriptions in this paper, I assume, refer mostly to the current two-stage Peltier cooled version. How many soundings have there been using that version compared to previous ethanol cooled versions? How many comparisons with CFH or other reference quality instruments are there? Can the authors provide statistically significant results? How robust is the instrument performance across many soundings, considering that these instruments are inherently disposable?
Line 91: The data quality of the cryogenically cooled instruments has certainly changed over the last 4 decades. The authors could just delete that statement.
The calibration uncertainty will be the limiting uncertainty for long term trends. The calibration uncertainty is quite reasonable, but can the authors show how they verify this uncertainty in long term production.
Does the instrument measure the temperature of the warm side of the Peltier device to verify that the sensor performs within the expected range? If so, if would be very useful to show these results.
Equation 7 is not obvious. If the heat conductivity across the device is considered small and the heat input into the cold side is small, then the temperature delta across the device becomes proportional to the resistive heating of the device. That does not appear to make sense for a Peltier device. The source for equation 7 is cited as Sugidachi (2014). The source for the same equation in that thesis is cited as Sugidachi (2011). The link in that reference no longer works and this thesis cannot be easily accessed. Thus, it is difficult to trace the reasoning for this equation. (On that note, in other places as well, the authors do not cite the original literature, but rather papers citing earlier publications. It would be better if they directly cited the earlier or original work.)
Section 2.6.1: During the time of the phase transition, the instrument is most likely controlling not around the dew-point or frost-point, but rather around the change in reflectivity between the two phases. During this period, the dew-point or frost-point is not defined by the mirror temperature. However, the control around the dew-point seems to stable until about 1400 s, i.e., super cooled water persists for that long (not 500 s). A region of ice crystals seems to grow in the upper right corner of the mirror. Does the detector even see these very large crystals? Do the authors have any information about the temperature gradients across the mirror? Could the crystal growth in the corner be caused by a significant temperature gradient? (Same questions for the very large ice crystals at the edge of the mirror in Figure 5.)
Lines 307 ff: There is one fundamental difference between cloud physics and the condensation processes on the mirror: The mirror itself can act as an ice-forming nuclei. Therefore, the temperature dependence is likely very different than that for the free atmosphere. The growth rates are possibly similar.
Sections 3.2 and 4.2: The derivation of the golden points and their uncertainty is very unclear. As written, the authors first create 5 Hz data by linear interpolation of the original 1 Hz data, then electrical noise is removed, lastly the data are smoothed using a Gaussian filter with a width of 1 to 3 s depending on altitude. The authors should remove the electrical noise first. How large is the electrical noise and how can it be distinguished from that generated by the PID controller? Secondly, why do they create 5 Hz data, if they are smoothed anyway? The linear interpolation combined with the Gaussian smoothing may even generate a little bit of extra noise, although that may be small. In the uncertainty estimation they use the number of 5 Hz data points in their statistics and assume they are random and uncorrelated. They are clearly not random and uncorrelated since they were created by linear interpolation and smoothing. This needs to be corrected. How is the uncertainty of the mirror temperature in the golden point selection process quantified? Is it the uncertainty of ±1 s in the detection of the golden point and the spread of the mirror temperature over these three data points? Figure 8 does not clarify this question. In line 383 the authors state that the extracted points are smoothed with a Gaussian filter. However, the extracted points are unevenly spaced, and Gaussian smoothing would achieve odd results under these conditions.
Related to the derivation of the golden points and the smoothing, the authors should make a statement how that affects the vertical resolution of the observations and how the vertical resolution is given in the processed data.
The authors also state (line 548 ff) that the aggressive PID tuning and larger oscillations are required for the golden point detection. However, there is no such requirement in the derivation or golden points. The method should work with any amplitude of oscillation since it only requires a short period of constant reflectivity.
The authors find that direct gaussian smoothing of the data leads to slightly lower frost-point temperatures than that derived from the golden point method.
Section 5.2: There are more comparisons between the two instruments, which should be shown and discussed. Is the difference seen in the Lindenberg profile also seen in other comparisons? If so, this should be reported. If not, this should be reported as well. If any of the instruments in the other comparisons did not provide data in any of these additional comparisons, this could also be reported. As is, showing just two comparisons seems a little selective.
Section 5.4: Looking at the data referenced by the authors, the comparison with MLS in the tropopause region seems to look a little better than shown by the authors. The averaging kernel of MLS is broader than 1 km. The Skydew data should be smoothed using the MLS averaging kernel, or at least smoothed to the same vertical resolution. They may need to review how they do the comparison. However, as a result the disagreement on ascent starts at a lower altitude. It would be helpful, if the authors could create a relative difference plot with MLS as reference.
One of the three “contaminated” profiles shows a significant jump in the upper troposphere, which is unlikely contamination. This may indicate additional complications not described by the authors.
Lines 628 ff: The sensor sits on top of Styrofoam box, which is very good for ascent measurements, but very bad for descent measurements. How would descent measurements be better for identifying contamination?
Although the instrument is a frost-point hygrometer, could the authors please add mixing ratio plots for the stratospheric sections of the profiles as they do in Figures 15 and 16? These panels could be added in almost all plots starting with Figure 7. In particular, the comparison plots in Figures 11-13 should show the mixing ratio and the relative mixing ratio difference for the stratospheric part of the profile (or a log plot for mixing ratio, which covers the entire altitude range).
Data availability: Thanks to the authors for making some data available from the YMC-BSM campaign. The GRUAN data archive does not make frost-point observations available and is not a suitable archive for that purpose. It would be better if these data were publicly archived elsewhere.
Figures:
Figure 1: The actual sensor is very hard to see. The right-hand panel could be enlarged to make the sensor arrangement clearer, in particular the coupling of the Peltier element to the large radiator. It appears as if there is a cover over the actual sensor. This cover may accumulate contamination that could potentially be detected by the mirror due to its proximity. This should be discussed.
Figure 2 should show the first two panels with a wind speed of 5 m/s (typical balloon ascent rate), rather than 2 m/s or 0 m/s. These parameters will give a much better representation of the temperature gradients that may be achieved in a sounding.
Figure 4: The images of the condensate on the mirror do not come out well and could benefit from some contrast enhancements. Although the authors specify the air temperature at about 0°C, the mirror temperature is the relevant temperature (about -12°C). There is a red area underneath the enlargements, which also changes between the images. What does it represent?
Figure 5: The caption lists the air temperature but should better give the mirror temperature at which these images were taken (they are listed in the legend of each panel). The text referring to this Figure should also list the mirror temperature, not the air temperature.
Technical comments:
Line 43: Replace “reduces” with “decreases”.
Line 49: change to “… employ capacitive RH sensors.”
Line 85: replace “cooler” with “colder”.
Line 136: Replace “at” with “a”.
Line 235: Delete “that”.
Line 545: Insert: … almost “always” …
Line 561: appear
Line 588: Figure 15 (not Figure 16)
Line 590: Replace “trail” by “train”.
Line 615: Replace “over” with “of more”.
Line 621: Remove the comma.
Line 624: “A dual sounding …”
Temperature differences are sometime expressed in °C and sometimes in K. Please use K for temperature differences throughout.
When referring to Figures, please use “Figure x” consistently and avoid using “Fig. x”.
Citation: https://doi.org/10.5194/egusphere-2024-635-RC2 -
AC1: 'Reply on RC2', Takuji Sugidachi, 11 Jul 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-635/egusphere-2024-635-AC1-supplement.pdf
-
AC1: 'Reply on RC2', Takuji Sugidachi, 11 Jul 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 349 | 118 | 34 | 501 | 17 | 16 |
- HTML: 349
- PDF: 118
- XML: 34
- Total: 501
- BibTeX: 17
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1