The following review refers to the revised version of the manuscript “Characterisation of J(O1D) at Cape Grim 2000–2005” by S. R. Wilson. First, I like to thank Dr. Wilson to consider my suggestions and those of Referee #1 pertaining to the original version of the manuscript. I have reviewed the “Response to Referees” document prepared by Dr. Wilson and feel that all concerns raised with respect to the original reviews have been addressed appropriately. The revised version of the manuscript is in my opinion now appropriate for publication in ACP with mostly minor revisions (see below). My only major concern is that I don’t think that the strong increase of the cloud modification factor at solar zenith angles (SZAs) larger than 70° (e.g., Figure 7) is due to natural causes. Instead, I think this increase is an artifact of measurement errors at large SZAs. Reasons for my suspicion of measurements artifacts are provided in my line-by-line comments below. Based on this, I suggest that the attribution of the observed effect to large cloud enhancements at large SZAs is de-emphasized or removed completely.
Prompted by my suggestion in my original review, Dr. Wilson has now prepared an “Uncertain Estimate of J(O1D),” which he has submitted to be published as a supplement. I greatly welcome this idea because I think that an uncertainty budget it required to support the conclusions of the manuscript. Unfortunately, this supplemental document is poorly written, has numerous errors (which I discuss below), is in large parts incomprehensible to me, and incomplete. At the very least, I don’t have confidence that the “combined error” provided at the end of Table 2 of the document is reasonable.
This leaves a conundrum: on one hand, the manuscript is appropriate to be published with mostly minor revisions. On the other hand, the supplement (which I consider essential because some of the conclusions of the paper hinge on good knowledge of the measurement’s uncertainties) requires major revisions.
I also want to point out that the “ratio-Langley technique,” which forms the basis for the calibration of the measurements presented in the paper, is unique and to the best of my knowledge not used by anyone but the author. This technique has been described previously by Wilson and Forgan (1995), but unfortunately no uncertainty budget of the method is included in this Applied Optics publication. I feel that it is in the best interest of Dr. Wilson to invest the time to prepare a solid uncertainty budget of the method, which would not only benefit the manuscript at hand but also future applications of this technique.
*****Specific comments pertaining to the revised manuscript (Major concerns are prefixed by “(MAJOR)” *****
P2, L5: What is the confidence interval of the quoted “total uncertainty” of 25%? Is it 1 sigma or 2 sigma or something else?
Section 1.1.1: $F(\lambda)$ is introduced as a spectral quantity. All formulas of section 1.1.1 do not include the argument “(\lambda)”. It should therefore be pointed out that the argument “(\lambda)” was omitted in Eqs. (3) - (6) for simplicity.
P4, L17: Global irradiance (E) is not the energy striking a horizontal plane. Rather, it is the radiative power (not energy!) per unit area incident on a horizontal surface.
P4, L21: Technically, F is the “spectral actinic flux” not the “actinic flux”
Eq. (5): There should be a $\mu$ in front of the $E_0$ in the parenthesis of the formula
P5, L22: Uncertainty of what quantity?
P8, L19: What is the lower value for alpha? 1.7 or 1.73?
P8, L26: Insert “because the product sigma * Phi * F can be different from zero outside this wavelength range.” after “photolysis rate”
P9, L7: Replace “terms” with “terms contributing to the calculation of J(O1D)”
P9, L8: Again, what is the confidence interval of the quoted “total uncertainty” of 25%? In the preceding sentence, a 1-sigma uncertainty of 9-12% is attributed to the irradiance measurements. This may imply that the combined uncertainty of 25% quoted here is also one sigma. This would mean a 50% 2-sigma uncertainty. Is the uncertainty really this large? If so, many conclusions presented in the following sections have a rather shaky foundation. Also, replace “This does” with” This uncertainty does”
Section 3.2: What is the source of the ozone column used for modeling?
P9, L18: What is the confidence interval of “12%” cited here? I also assume that the model has some uncertainty, for example the ozone column used in the model has some uncertainty. So measurement and model could disagree by more than 12% (the value pertaining to the measurements alone), and could still agree to within their combined uncertainty.
P10, L27: I don’t understand “driven by the rate of change of the solar zenith angle at midday”. The SZA is almost identical from year to year. The small difference arising from the fact that a solar year does not fit exactly into either 365 or 366 days should have a sub-1% effect. On the other hand, variability in ozone is largest in spring and this might be the dominant factor explaining the larger variability in late winter / early spring (September and October).
P11, L17: I note that clouds are also the norm at Ushuaia.
(MAJOR) P13, L13: The good agreement of the 99 percentile with the clear-sky model calculation is serendipitous considering the uncertainty of both measurement and model. Because of these uncertainties, it is not justified to conclude that 1% of all data are affected by cloud enhancement.
(MAJOR) P13, L15: At SZA=75°, about 40% of all measurements seem to be enhanced above the clear sky flux (the exact percentage is hard to determine visually from Figure 5). Enhancement of the actinic flux can only occur of if the disk of the Sun is not occluded and additional radiation is scattered towards the instrument by clouds near to the Sun. (See for example Crawford et al., 2003). I think that a percentage of 40% is highly unlikely, and based on my experience, I feel that even 25% enhancement at 65° is unrealistic. Rather, I think that the discrepancy between the 99 percentile and the clear sky model at large SZAs is a consequence of systematic errors in either the measurement and model. Errors of this magnitude are actually expected based on the uncertainty estimate given earlier. The conclusions in this paragraph should be toned down considerably.
P13, L19: The behavior is NOT consistent with known cloud impacts published for example by Crawford et al., 2003 and Mateos et al., 2014.
(MAJOR) P13, L26: The upper part of Figure 7 (median) suggests that clouds have no effect at SZA=75° and actually increase J(O1D) at 80°. I think this is unlikely and not supported by the paper by Mateos et al., 2014. For example, Mateos et al. (2014) conclude that the CMF is always smaller than 1 for overcast conditions. The large increase of CMF as a function in SZA shown in Figure 7 could therefore only be explained by a sharp increase of broken-cloud cases with increasing SZAs. I think this is highly unlikely. Again, I feel that measurement uncertainties are responsible for this effect. I recommend to change the focus of the interpretation: more emphasis on uncertainties / measurement artifacts and less on a real effect.
P14, L1-4: “As scattered … properties” should be deleted based on my previous comment.
P16, L4: Change “8%” to “-8%” to be consistent with the quoted “-20%” mentioned later
P16, L15: Change “20%” to “-20%” to be consistent with the quoted “-20%” mentioned earlier
Caption Figure 1: Indicate what satellite was used.
*****(MAYOR) Specific comments pertaining to uncertainty estimate (supplement)*****
As mentioned in my general remarks, I believe that the uncertainty estimate that is provided as a supplement needs to be substantially improved. The author should consult the following publications for guidance: BIPM, 1995; Taylor and Kuyatt, 1994. Since the actinic flux F is calculated from several variables as described in Section 2.1. of the supplement, the calculation of the combined uncertainty entails to ascribe an uncertainty to each of these variables, and apply the procedure outlined in the guidelines mentioned above. It would also be helpful to reiterate the advantages of the ratio Langley technique compared to the standard Langley technique. For example, why is the uncertainty reduced by using a sunphotometer as a reference? Is that instrument calibrated against the reference spectrum from Chance and Kurucz (in this case, the uncertainty of the calibration must be at least as large as that of this reference spectrum) or was it calibrated against a standard issued by a Standards Laboratory such as NIST?
The following specific comments refer to the line numbers of the supplement.
L31: The argument of E_dir should be $\lambda_r$, not $\lambda$
L35: $f(\theta) is a correction term for the departure from the ideal (Lambertian) angular response. Hence, it should be specified that $c(\lambda_r)$ is the calibration factor for $\theta$ = 0.
Eqs. (5) and (6): $\lambda_r$ should be changed to $\lambda$ in the last term of both equations (otherwise the last factor would be 1). Also the quantity $c(\lambda) is not defined.
Eqs. (7) and (8): Change $S_{\downerror}/$E_{\downerror}(\lambda_r)$ to $S_{\downerror}(\lambda_r)/$E_{\downerror}(\lambda_r)$. (In standard math notation, the argument has to be repeated after every variable and cannot be applied across an operator (i.e., “/”) .
L58: Regarding “needs to be less” is awkward. The actual rather than the desired uncertainty should be quoted.
Sections 2.2.1 through 2.3 are by and large incomprehensible to me. These sections should be replaced with a well-structured uncertainty description as outlined in the guidelines mentioned earlier. For a complete uncertainty budget, all components that contribute to $F(\lambda)$ (Eq. (10) need to be considered. For example, $c(\lambda, \theta)$ is not discussed at all in Section 2.2.1 while the quantity $c^{rel}(\lambda)$ addressed in Section 2.2.2 has not been defined.
L72: Which “two measurements”?
L95: Why are Dobson wavelengths discussed here? I thought the sunphotometer wavelength is 342 nm?
L112: It is stated that the measurement noise is the major driver of the overall uncertainty and that the uncertainty is only weakly dependent on wavelength. How can that be? Measurement noise is largest at the shortest wavelengths where photos are scarce. So the uncertainty should greatly increase towards shorter wavelength. This is also supported by Figure 2.
Table 2: What is the confidence interval of these uncertainties?
*****Typos / language pertaining to manuscript*****
P2, L3: include “J(O1D)” in commas or parentheses to emphasize that this quantity refers to “rate of production… ozone”
P5, L24: Replace “measurement detector have typically been used” with “systems have been used in the past to determine J(1OD).”
P5, L25: Replace “spectrometer or a diode array” with “spectrometer, and a diode array”
P7, L9: Replace “by the the global” with “by the global”
P20, L20: Replace “of the ratio of the SRAD direct beam irradiance to the sunphotometer.” with “of the ratio of the SRAD-derived direct beam irradiance to that measured by the sunphotometer.”
P7, L24 and L25: Remove “e.g.” and parenthesis pertaining to “global” and diffuse”
P8, L7: Delete “upon”; replace “this will” with “this effect will”
P8, L13: Replace “diffuse” with “diffuse component”; replace parentheses around alpha
P8, L15: insert space after “(McKenzie et al., 2002)”
P8, L18: Remove space after “(Kylling et al., 2003)”
P14, L25: Insert parenthesis after “1000 m” (Also space between “1000” and “m” is missing!)
P15, L2: Move comma after “(Gerasopoulus et al., 2012)”
P15, L5: closing parenthesis is missing.
P15, L17: insert comma after “coast”
P15, L22: insert space in “90m” and “100m”
P16, L12: change “clouds” to “cloud”
Caption Figure 4: Change “(red dashed line)” to “(red solid line)”
Caption Figure 5: Change “fits to two” to “fits using two”
*****References*****
BIPM, Bureau International des Poids et Mesures. Guide to the Expression of Uncertainty in Measurement. ISO, Internat. Organization for Standardization, 1995.
Chance, K. and Kurucz, R. L.: An improved high-resolution solar reference spectrum for earth’s atmosphere measurements in the ultraviolet, visible, and near infrared, J. Quant. Spectrosc. Ra., 111, 1289–1295, doi:10.1016/j.jqsrt.2010.01.036, 2010. 18395, 18396, 18397]
Crawford, J., R. E. Shetter, B. Lefer, C. Cantrell, W. Junkermann, S. Madronich, and J. Calvert, Cloud impacts on UV spectral actinic flux observed during the International Photolysis Frequency Measurement and Model Intercomparison (IPMMI), J. Geophys. Res., 108(D16), 8545, doi:10.1029/2002JD002731, 2003.
Mateos, D., di Sarra, A., Bilbao, J., Meloni, D., Pace, G., de Miguel, A., and Casasanta, G.: Spectral attenuation of global and diffuse UV irradiance and actinic flux by clouds, Q. J. Roy. Meteor. Soc., doi:10.1002/qj.2341, 2014.
Taylor B. and C. E. Kuyatt, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, NIST Technical Note 1297, National Institute of Standards and Technology, Gaithersburg, Maryland, available at http://www.nist.gov/pml/pubs/tn1297/
Wilson, S. R. and Forgan, B. W.: In Situ Calibration Technique for UV Spectral Radiometers, Applied Optics, 34, 5475–5484, 1995. |
- Printer-friendly version
- Supplement
