|L8: The references are also calibrated for mole fractions, which is required for the IM. Please specify that here.|
L28: I am not convinced that the differences in quoted repeatabilities between the two methods IM and RM are really statistically significant. It would be better to avoid value judgements about which method is “better” and simply state the repeatabilities as observed under the measurement conditions. Are they really the “optimal” conditions (L11)?
L58: FTIR has also been developed recently for this purpose, especially at BIPM as is referenced in this paper. Perhaps reword this as “Optical (infrared) spectroscopy now offers this possibility following strong developments in recent years in FTIR and especially for the laser light
Figure 2. From the plots it is clear that the linewidths in the calculated (fitted) spectrum do not match those measured, leading to the typical “second derivative” shaped residuals (and a position error in the case of the 636 line). Is this due to the accuracy of the Hitran widths, or the linewidth model employed in TDLWintel? Could the Aerodyne authors comment, perhaps in the Figure caption. It would be desirable to reassure the reader that the resultant errors in retrieved amounts are systematic and will calibrate out under identical measurement conditions, but they may lead to dependence on such conditions (eg. temperature, pressure) if they vary.
L125: strictly speaking, the r18 and r17 isotope ratios differ from the 628/626 and 627/626 ratios by a factor of 2.
Figure 4: I cannot understand the relationship of the third column in the table (labelled std dev (‰)) to the Y axis on each figure, ppm2 or ‰2. I would expect this to be the square root of Allan deviation at the minimum, in units of ppm (upper plots) or ‰ (lower plots). These are sqrt(AV) at a particular averaging time, not strictly the same as standard deviations. But the numbers do not line up:
For example for 626 in the top left plot, the minimum AV is ~20 ppm2 at 16 s, so sqrt(AVmin) = 4 ppm. The given value in the plot table is 0.01 ‰. 4 ppm seems rather high (1% or 10‰ of 400 ppm) for the laser measurement.
Please explain and/or correct the numbers in the table
Eq 1. It is confusing to use the same symbol (M) for the measured amounts (RHS) and the ratio to the WG measurement (LHS) – the former has the units of an amount of gas, the latter is dimensionless. Could you be clear in the text by using a different symbol for the LHS, say M’, or R for ratio, or I for index and referring to the appropriate quantity in the text strictly according to which is being used (for example L 235 and following).
L159 et seq: As stated in the text, the std dev of the corrected isotope ratio should be lower than that for the uncorrected ratio if the drift correction is effective, by an amount that depends on the drift, but only after taking into account that the random error in the ratio (M’) will be sqrt(2) times larger than in either of the individual M values (so if there is no drift, the std dev of M’ would be 1.4 times the std dev of M). But the statement line 160-164 that the std dev should not get larger with n is not correct – if the correction procedure were perfect the std dev should decrease by sqrt(2) or a factor of 1.4 from n=5 to n=10 measurements. In fact it increases in every case, so the drift correction is not perfect over that timescale. Why?
Further, in Table 1 caption, are these really RELATIVE standard deviations (in ‰ of ‰), or are they actual standard deviations of the isotope ratios in ‰? This was changed from the original version. Please be very wary of using ‰ for a relative value of 2 quantities in isotope work because it is so easily confused with a delta value itself, also quoted in ‰.
L174, L1276: should this be adsorption, not absorption?
L188: please refer to comment on Figure 2 above where such a lack of fit is evident in the linewidths.
L197: should this read “when eq 3 is brought into eq. 2…”?
L198: … for either or both of those …
L236: This should read either “The CO2 mole fraction is calculated by multiplying M’626(t)dc by the known M626(t)WG of the working gas” or more strictly “The CO2-626 mole fraction is calculated by multiplying M’626(t)dc by the known M626(t)WG of the working gas” (These will be the same for the same isotopic composition).
L258: the meaning is not clear – what quantity is “The WG(t)”
L263: “As measurements were not conducted on CO2 of similar isotope composition” Not clear, of similar composition to what?
L330: The Picarro analysis is based on 626, not whole CO2 – there is an inherent assumption that the 626/totalCO2 ratio is constant. This is a potential source of error if the isotopic composition of different reference gases varies significantly, but from Table 5 the variation in isotopic composition is not significant for this error (Griffith 2018). However the narrow range of delta13C across the reference gases does not provide a wide span for calibration using the RM – for IM it does not matter. See L390.
L370, 371: Tans, Crotwell and Thoning (2017) also successfully implemented the IM and should perhaps be referenced here – it is used routinely at NOAA in the generation of reference gases.
L584: For the benefit of those reading only the abstract, intro and conclusions, please spell out Ratio Method and Isotopologue method here.