|The manuscript "The use of O2 1.27 micron absorption band revisited for GHG monitoring from space and application to MicroCarb" is a detailed and thorough inspection of the oxygen airglow phenomenon as observed by SCIAMACHY on ENVISAT, however, it does not adequately project these results into the purported conclusions regarding the architectural choices for the MicroCarb instrument.|
The title includes "...application to MicroCarb", the abstract and introduction discuss uncertainty in retrievals between A-band and delta band with respect to confounding aerosol signatures as well as atmospheric emission. The question posed (somewhat indirectly) by the authors is "Should GHG instrumentation use the 1-Delta band for airmass determination?". The answer to the question is presented as a competition between errors induced by atmospheric emission vs. errors induced by aerosol extinction (and extrapolation of that effect). This is a reasonable premise for a manuscript, however it ignores other factors such as (1) the relative signal to noise ratio (SNR) of practical instrumentation (2) the errors due to instrument trades in spectral resolution (3) biases induced by spectroscopic errors. Some of (2) is mentioned in this manuscript, but it is not quantitative. The cursory mentions of SNR (1) are made for existing instruments and not put in context with MicroCarb SNR, which is not described. As for (3), the authors are utilizing HITRAN, which is a world-class standard, but is NOT what is used for OCO and GOSAT missions, (ABSCO), nor for TCCON (GFIT), which find that retrievals approaching 1 ppm accuracy require (among other things) a dedicated spectroscopic formulation that goes beyond HITRAN.
Before providing some detailed comments, I will make a primary recommendation - re-write this manuscript as two independent manuscripts. The first manuscript should focus on the airglow retrievals and model intercomparisons of the SCIAMACHY data. The second manuscript should build on the airglow model validation of the first paper and then present a fully quantitative intercomparison of the methods for airmass retrieval from the differing O2 near-infrared bands. This first manuscript is essentially complete and represents a subset of the information currently reported. The second manuscript, if meant to answer the question about the use of 1-Delta band for MicroCarb, will need to address much of the detailed comments below. If the authors choose to retain a single manuscript, my 'presentation quality' rating of 'poor' will be difficult to overcome because the effort as presented here is imbalanced and does not provide a cohesive conclusion.
A forward model of aerosol effects is formulated in Eqs 5-7, but not used to make any quantitative statements. For application to GeoCarb, I expected the relative contributions of aerosol loading in the 0.76 and 1.27 micron regions to be presented. I also am concerned that the retrieval of aerosols will depend critically on the bandwidth of the instrumentation, as alluded to in the dicussion about reducing GeoCarb's resolution to cover the whole 1.27 micron band. These factors surrounding aerosol effects must be quantified and then weighed vs. other sources of error such as the airglow uncertainty.
The airglow model is presented to be independent of meteorological and chemical perturbations, which is a solid argument for its use in the simple subtractive approach. The authors dismiss issues at high solar zenith angle (SZA) in two different ways (1) the simplistic time fields do not capture the rapidly changing conditions near the terminator (2) the night-time kinetics are not important during daytime measurements. I find two issues with this approach. First, there is still CO2 to be measured at high SZA (i.e. latitude) - so what is the practical cut-off? Second, the rapidly changing area of the SZA dependence is your best chance to observe/quantify what will otherwise be a bias. Another way to frame
these points is that a detailed model of the high SZA dependence would allow for airmass retrievals to extend to high SZA/latitude where airglow bias would shrink considerably, and perhaps reveal other contributing biases. Since the presented airglow model assumes steady state and large time zones, this region will be dominated by model biases - at the very least, the MicroCarb mission should understand the extent of where these biases will effect retrievals.
The proposed configuration of MicroCarb is presented as retrieving two bands of oxygen 'for safety'. Does this presume that the A-band retrieval is sufficient? Wouldn't a two-band retrieval (which is implied to be a good way to retrieve aerosol) be best? It is unclear what 'for safety' means, some kind of reduced risk presumably, but the manuscript provides no quantitative comparisons of A-band vs. delta-band vs. combined dual band retrievals.
The entire Appendix A is review material and is un-necessary for supporting the publication.
in the absract third paragraph (why are there 3 paragraphs!) the text implies CO2 can be retrieved with 'only' the 1.27 micron band measurement
in the first sentence of the introduction the main driver of climate change is attributed to CO2. I recommend qualifying this as 'human-induced climate change' since geologic climate shifts are driven by solar and orbital effects as well as GHG loading from geological sources.
page 3, line 15, "...modelling may be more accurate." this was stated as fact in the abstract. As a central theme of this paper, it seems to be inconsistently referred to in many ways.
page 8 line 5, "...synthetic absorption emission spectrum." this is an odd thing to state
FIgure 6. If I understand the context, the emission=absorption case is what is described in the text and used in the model, but this figure implies a massive error, I think this is just my confusion, but it shouldn't so be hard to follow. Either way, this is a fundamental effect that seems trivial to spend a section explaining.
Page 16 line 9-12, "This is why it is not practical..." seems like a death-sentence for MicroCarb, I understand the spectral resolution is better, but the listed issue "back-scattered solar radiation" is not quantified for MicroCarb (vs. SCIAMACHY) in this document
Figure 10, which is reproduced from another source, is totally un-necessary for readers to understand the physics
Page 18 lines 19-20 "At a given solar zenith angle..." this point is central to the thesis of the effort, it should be mentioned in the abstract and the conclusion
page 22 lines 4-5, the radiometric properties of SCIAMACHY surely deserve a literature reference
Figure 13. The descrepancies between measurement and model are indicated to be minor throughout the discussion, however, this plot suggests significant (maybe 30%) of the airglow (at higher altitudes) is unmodeled. Perhaps this error is not propogated into MicroCarb results, but no proof has been given.
FIgure 15. In the text the 1-Delta band is stated to be in the spectral and brightness range of the CO2 bands, but these plots seem to indicate that only the A-band has dynamic range that encompasses both CO2 bands, with the 1-Delta band falling in-between, is this a quirk of the selected simulation? My belief is that a 1-Delta only retrieval, especially with the limited bandwidth of MicroCarb, would not effectively span the radiative dynamic range (contrast) that provides high precision.
Page 29 line 25-26 "...uncertainty is assumed to be proportional to square root of signal." Such a weigthing scheme reduces the impact of a large dynamic range, it also seems unjustified unless the noise in the instrument is dominated by its source (the sun), usually remote sensing instruments are dominated by detection noise which is spectrally flat.
Page 29 line 38 following on the last comment, the uncertainty may not be calculated from the SNR, but it IS used to weight the fit, which does something, also the "dispersion of the results" could be either instrument noise, model error, or retrieval failure, some description of the relative impacts of these factors is needed to understand the value given
Page 35 line 14-17, "It is widely recognized..." here the main question is answered as being a known, however, this manuscript did not convince me and no reference is given.
Page 36 lines 9-12, Here are some quantitative statements, presumably the nadir GHG observation that is imagined here is that of MicroCarb? This list of quantitative results is in the conclusion, and the necessary retrieval method implied by them is described in the following section of the conclusion. This is a very odd structure for a manuscript.
Page 37 line 9-10 Here we are in the conclusions still, and another result is provided fro the first time, albeit very qualitatively "...accuracy almost compliant...". The number to comply with was never stated, the model result not given and the superiority over A-band results is simply stated as fact.
Page 37 lines 20-25 This imples MicroCarb could/should do better with a resolution/bandwidth trade, which contradicts much of the premise of the rest of the mansucript that seems to confirm MicroCarbs configuration
Page 37 lines 25-29 Here, at the end, another potential instrument is described, something called CO2-M, and a recommendation is given about including the 1.27 micron band. Presumably this is a follow on possibility? If anything, the effort here frames a capability to model and predict good instrument design choices, but what is given here is hardly more that a statement about necessary resolving power, and the study claims to provide justification for the entire band. When it comes to instrument design trades, resolving power is only one of many factors.
Appendix B page 59 line 9, reference to equation 6 is wrong, equation 6 is about aerosols
Appendix B page 60, the discussion about partition sums references HITRAN quite a bit, but the partition sums in HITRAN are literature traceable to Gamache and the TIPS code/tables, the references are given in HITRAN. The 'care' needed to extract energy levels from HITRAN should be alleviated with HITRANonline, which now book-keeps energy levels directly, the energy levels for oxygen are traceable to Yu et al. If one needs to calculate upper state energies, the transition energy is added to the lower state energy.
Figure B2. THe labeling must not be correct, 'statistical weight' would be a straight line vs. J', what is plotted is probably population of the level. This figure is completely basic and un-necessary.
Figure B5. The axes label and caption seem to imply a ratio is plotted, however the values and units are not consistent with a ratio.
page 67 line 24, what is E_mn(lambda)?
Page 86 lines 20-23 I find it surprising that the observations "...suggest a fast increase of the emission with decreasing altitude around 50 km..." the chart in Fig 13 shows a decrease in emission below 50 km.