Dear Authors,

Thank you for your detailed and authoritative responses to the referees' comments. I have carefully reread your well-written paper that certainly merits publication in AMT. At this stage, there are some very minor and technical issues remaining, which should be addressed before publication. Please find these listed below.


Minor corrections:
======================
(page and line numbers refer to the manuscript version that keeps track of revision changes)

[0] Some company names appear in capital letters, others do not. It seems preferrable to keep a common format. I suggest capitalizing only the first letter in company names, which is an easier read, especially for long names.

In the same veins, there is excessive use of the brand name to denote the CRDS analysers, but other instruments such as the modified QCLS/ICLs instrument are not presented by insisting on the original brand each time. This poses the risk of inadvertent advertising of one particular instrument as compared to others and hinders easy understanding for people outside the field. In general, it is preferable to use brand/model names only when instruments are presented for the first time and when required for the understanding. Otherwise instruments should be referred to by their operational principle or instrument function (eg "hygrometer" instead of "Sensirion", etc.). Please modify Figures 10, 11, 12, 13 and the text accordingly.

[1] The typesetting guidelines of Copernicus journals require a space between quantity and unit. This also applies to dimensionless units such as the % sign (ie 78 % instead of 78%, ...).

[2] p 7, l 17-18 : " ... CH4 (Cert.-Nr. CB11361, WMO X2004A for CH4 (Dlugokencky et al., 2005)). C2H6, CO and N2O are compared to NOAA ..." -> CO2 is missing in the list even though it is mentioned later on in chapter 5, p 17. Please complete the list.

[3] p 16, Fig 8 and discussion of frequency lock : There is a regular spike pattern in Fig 8 with a period of about 30/6 ~ 5 min. This supports the hypothesis that shifts are not only caused by mere laser drifts, but are also caused by some well-timed mechanism. It is not clear whether this is meant by "high-frequency shifts are evident, including discontinuities". The discussion of the spikes could be added to the discussion on p 16.

[4] p 22, l 20-21 and p 23 Fig 13 : According to the figures, the QCLS-CRDS and the QCLS-FLASK comparison shows a good agreement in relative deviations, except for CH4, where QCLS-CRDS gives a 1-sigma of 1.4 ppb, but the flask comparison seems to give a different number (1-sigma = 3.9 ppb (n=40)). This needs to be commented.

[5] p 21, l 12+, discussion 13CO2 measurement : "We estimate the required isotopic composition that could explain a large bias of up to 17 ppm (see Supplement Section 3) in CO2 to be 98.447% primary isotopologue and 1.079% secondary isotopologue or δ13C = −19.6 ‰ which seems reasonable according to B. Coplen et al. (2002)." This statement seems to be in direct conflict with the Supplement, where it is derived that d13C must be -37 ‰ to have an impact of 17.2 ppm. Please correct and be more detailed about the -19.6 ‰ and your calibration procedure that should be crucial in infering a particular isotope composition.

The discussions in the main text and in the supplement also merit clarification. In particular, the definition of x_retrieved remains dubious, as remains the meaning of "possible influence estimate". It might help to differentiate between two effects of opposite sign : HITRAN conventions (a ~ -5 ppm effect when measuring background air CO2) and the systematic bias that could play a role here (up to about +12 ppm). HITRAN assumes 13C isotope abundances of VPDB (d13C = 0) to derive line intensities, and, for the purpose of demonstration, one assumes that these are correct. When we then measure CO2 as 13CO2 we can use the rule of thumb that a relative change of 13C will translate into a relative change of CO2 even if the concentration of CO2 remains unchanged (less 13C leads to less CO2 measured and higher 13C leads to higher values of CO2). This means that the QCL measurement of air with d13C = -9.7 ‰ leads to CO2 that is roughly 1 % (or 4 ppm at 400 ppm CO2) too low (effect of -4 ppm). A positive offset of +10 ppm can only be obtained if the instrument is calibrated with a gas of d13C = -40 ‰. This seems to be reasonable provided the tank CO2 derives from fossil fuel. Air, which has a d13C that is higher than fossil fuel CO2 by ~ 30 ‰ will yield a 0.03 * 400 ppm = +12 ppm higher CO2 abundance. Note also that line 15 on p 3 of the Supplement wrongly refers to methane.

To which degree has the 13C isotopic composition of CO2 to be known to reach WMO compatibility ?

Technical corrections:
======================
(page and line numbers refer to the manuscript version that keeps track of revision changes)

p 1, title : use subscripts in chemical formulae
p 1, abstract : "... and central US." -> "... and central US." (note you use USA without full stops later on ....)
p 3, l 5 : "since the pre-industrial era, where ..." -> consider replacing "where" by "and"
p 3, l 11 : "Aircraft provide ..." -> "Aircrafts provide" (eventually "aircraft provides")
p 4, l 25 : consider replacing "... required to operate the instrument on research aircraft." by "... required to operate the instrument on a research aircraft."
p 4, l 27 : "The spectrometer is split into an electronics compartment and an optics compartment." -> "The spectrometer is split into an electronics and an optics compartment.
p 4, l 28 : "... , etc.." -> "... , etc."
p 5, l 3 : "... , etc.." -> "... , etc."
p 6, l 2 : "every half of a second." -> "twice per second."
p 6, l 17 : "... avoiding injecting large vibrations into the ..." -> "... reducing vibrations of the ..."
p 6, l 18 : "This translates to a net flow rate of 25 SLPM". Different entities use different standard conditions (IUPAC, NIST, EPA ...) SLPM is therefore not a well defined unit. It would help to recall "your" standard conditions here (just once in the text). Even better, you could (just once) give the molar flow rate in mole/s here. Note that since 1982 IUPAC defines T = 273.15 K and p = 100 kPa as standard conditions.
p 6, l 19 : "when operating with a cell pressure" -> "when operating at a cell pressure"
p 7, l 3 : "Polytetraflouroethylene" -> "polytetraflouroethylene"
p 7, l 16 : "cross-calibrated using a Picarro CRDS against NOAAA standards" -> "cross-calibrated against NOAAA standards using a CRDS "
p 7, l 24 : "Typical calibration distribution diagrams ..." sounds like a technical term. It could be easier to understand using "Histograms of typical calibration measurements are provided ..."
p 8, legend Fig 2 : A letter "H" appears in the drawing, which is not explained in the Fig legend.
p 10, legend Fig 4 : change all chemical names to small letters, add axis title to x-axis. What is the difference between quantities and units used on the y-axis in this and in the previous (Fig~3) figure ? Shouldn't they have same names and spellings ? Since "adu" does not refer to a proper name, it should be spelled using small letters.
p 11, legends and axis labels Fig 5 : font size is too small
p 11, Fig 5, y-axis label : The plot does not show the optical depth but the transmittance
p 11, Fig 5, graph labels : replace "microwindow" by "fit window"
p 11, l 6 : "... shift variables are held constant at the mean over its last 10 values." -> "... shift variables are held constant at their means over the last ten values."
p 13, l 6 : "Eq. (1)" instead of "Eq. 1"
p 13, Table 1 : Broadening coefficient (gamma_air)
p 15, l 2 : "Power drawn from the aircraft remained under 50 A at all times and settled at approximately 40 A." Replace "power" by "current" or "electrical current".
p 15, l 3 : "The volumetric flow rate at ... SLPM". You specify the flow of an absolute amount of gas and not a volume flow. I suggest dropping the term "volumetric".
p 15, l 9 : "will decrease the signals standard deviation" -> "will decrease the standard deviation of most of the signals"
p 15, Fig 7 left panel : Legend is difficult to read, please increase font size. Plot could be made more readable by excluding data for tau > 500 s (does not seem significant anyway) and restrict y-scale from 1e-4 to 1e1.
p 15, Fig 7 right panel : It appears that the confidence interval given for linear fit parameters ignores the uncertainties of the x-values. If this is so, it is better not to insist on the confidence intervals of the parameters. Give R-squared with number of digits required to see the deviation from 1.
p 16, Fig 8 : Please increase the thickness of lines in legends. Traces are difficult to identify otherwise.
p 20, l 2 : "(Gordon et al., 2017))" -> "(Gordon et al., 2017)"
p 20, l 15 : put "H2O" in upright letters
p 20, l 18 : Please add + or - sign to bias constants in order to avoid ambiguities, assuming that "+" means that the QCLS is higher than the reference.
p 21, l 22 : "(see 2.3)" -> "(see Section 2.3)"
p 21, l 23 : "Eq. (1)" instead of "Eq. 1"
p 21, l 27-28 : "... precision (uncertainty) estimates based on ambient measurements at stable conditions of 0.3 (2) ppb, 0.02 (0.1) ppm and 2.0 (5) ppb ...". This notation is non-standard and quantities in parentheses might therefore be confused with uncertainties. I suggest writing "... precision/uncertainty estimates based on ambient measurements at stable conditions of 0.3/2.0 ppb, 0.02/0.1 ppm and 2.0/5.0 ppb for ...".
p 23, Fig 13 : y-axis label "Norm occurrence" should come with units in 1/ppb or 1/ppm
p 23, Fig 13, central panel : symmetrize y-axis (from -2.x to +2.x)
p 23, Fig 13 caption : Please add legend for colour code of flask samples.
p 23, Fig 13 middle panel on top : Is there a reason for the skew in the distribution ?
p 24, l 2 : "(5 to 10mins interval)" -> "(5 to 10 min interval)"
p 24, l 10 : "above the eastern U.S.." -> "above the eastern US."

Dear authors,

Thank you for addressing all points. I recommend publishing "as is".

Best wishes,