This study investigates the C3H8 retrieval from ground-based FTIR spectra at Xianghe, and discuss the C3H8 column variation in North China, based on these new FTIR measurements. The technical details and uncertainty discussion are generally well provided in current form, but the result part, such as data comparison and trend explanation, are somewhat less satisfactory. Overall, I suggest the publication on AMT after presenting more information for data interpretation. Specific suggestions are listed below.

1. Method 2.3: Line 20-25: It is still not clear why perform a profile retrieval for H2O column concentration. Because each species could have large variability in vertical scale. Moreover, suggest providing more technical details about the how to perform a profile retrieval.
2. Section 3.2: Since the large difference exists for seasonal variation of C3H8 column concentration between model and FTIR measurements, it would be better not present this comparison in the main text, unless the authors could provide more evidence or information to explain these differences. For example, the authors could collect some surface observation of C3H8 concentration in Xianghe or surrounding regions that used for comparison to FTIR retrieval near the surface.
3. Section 3.3 Line 20-25: What is the significance by providing the ratio of ∆C2H6 to ∆C3H8? What does the trend of this ratio mean?

Section 3.4: The authors compare FTIR measurement to MkIV data here, but the basic information about MkIV measurement were not well described. Readers might be very interested about the principle of technique used for C3H8 measurement in MkIV and the accuracy of these data. Based on these information, we can rule out the systematic difference deviation between FTIR and MkIV.

Review of “Atmospheric propane (C₃H₈) column retrievals from ground-based FTIR observations at Xianghe, China”

General comments:

This study used ground-based FTRI Mid-infrared observations at Xianghe to retrieve C3H8 column through optimal estimation approach. Compared with CH4 and H2O, the absorption of C3H8 of is weak at 2964.5-2970.0 cm^-1, thus retrieving C3H8 is challenging. Although the authors conducted uncertainty analysis, I still have some concerns about the accuracy and importance of the C3H8 retrievals.

Specific comments:

1. Due to the weak absorption of C3H8, I suspect that the a posteriori C3H8 strongly depends on a priori C3H8. Figure 6 shows the seasonal variation of FTIR C3H8 is different from the two model simulations. How is the monthly variation of the a priori C3H8? Is it similar to the FTIR retrieval? It would be interesting to see to how the a posteriori C3H8 vary if different a priori profile (for example, using profiles from CAMS and WACCM as a priori) is used.
2. As CH4 and H2O have stronger absorption than C3H8, and CH4 and H2O absorption lines are not perfectly fitted, how it will affect the accuracy of C3H8 retrieval should be discussed.
3. In Sect. 2.4, how uncertainty is calculated? Apparently, we cannot get x_t and b_t. More details are needed.
4. Toon, G. C., Blavier, J.-F. L., Sung, K., and Yu, K.: Spectrometric measurements of atmospheric propane (C3H8), Atmos. Chem. Phys., 21, 10727–10743, https://doi.org/10.5194/acp-21-10727-2021, 2021. should be cited. It used observation around 2967 cm⁻¹ to retrieve C3H8.

Review of “Atmospheric propane (C3H8) column retrievals from ground-based FTIR observations at Xianghe, China” by Zhou et al., amt-2024-67

This manuscript presents a 4 year data record of C3H8 (and C2H6) column amount measurements at the Xianghe TCCON site. Retrievals are carried out from high resolution spectra in the 2964.5-2970.0 cm-1 range (C3H8 Q-branch). Several spectroscopic line lists are compared in terms of spectral fit quality. The uncertainty of retrieved C3H8 columns is estimated and the seasonal variation, as well as the correlation of C3H8 with other hydrocarbons, is discussed. Results are compared to two different models and to FTIR measurements at other sites.

The subject of this paper has relevance within the scope of AMT. New data are presented and I find the overall quality of presentation clear and concise. 

Yet, the paper currently leaves the reader with a number of open questions. Since Toon et al. (2021) have already presented first measurements of C3H8 with a similar methodology, I find that more work is needed with respect to the interpretation of the results before publication, so that this paper adds value to the existing literature. The big question to me is: Do we trust these column retrievals and if so, why are C3H8 columns so low in Xianghe?

General comments:

G1: I am concerned that the retrieved C3H8 values from this study do not line up with similar measurements and models. I do find it surprising that C3H8 columns 50 km downwind of Beijing should be one order of magnitude lower than C3H8 columns in Pasadena (10^15 vs. 10^16 molec/cm2). You mention on page 2: “we expect that the C2H6 and C3H8 concentrations are relatively high in this region” - I agree and would like to ask you to please expand your study in a way that resolves this apparent conflict. Similarly, it appears that the present measurements are in broad agreement with CAMS/WACCM during wildfire season in northern summer, but not during the winter when gas is being consumed. What are the atmospheric situations when you do observe high wintertime C3H8? Can we understand these results in the context of the existing measurements and models?

G2: Please discuss why the residual structure in Fig. 2 has structures that are larger than the ones found by Toon et al. (2021). Please present average spectral fit residuals in Fig. 2 (with std/min/max) to build confidence that your fit works. Are the average fit residuals understood?

Minor comments:

M1: You note the range of spectral resolution you have across the full MIR range. What is the spectral resolution in your window at ~2970 cm-1?

M2: Page 4, lines 21-22: add C2H6 to list of interfering gases

M3: the matrix S (not S_R) on page 6, line 16 and in eq. 8 needs to be introduced in the text

M4: You calculate the retrieval error due to uncertainties in spectroscopic parameters, but did not include all relevant parameters like line position or line shift. Please explain why these can be neglected in the uncertainty budget.

M5: How do you fit the spectral baseline (aka “background curvature”)? Please explain in the text.

M6: Page 11, lines 14-19 and page 14, lines 14-16: Please provide a more convincing explanation why the C3H8-CH4 correlation is weaker than the correlation with other NMHCs. There are oil and gas producing/consuming regions in Northern China (e.g. Changqing oil field/city of Beijing) and with a life time on the scale of weeks-months natural gas related C3H8 emissions from such basins/cities could easily reach Xianghe (compare page 2, lines 20-23).

M7: How long is the integration time for one measurement?

M8: I do not understand Fig. 9: why are there negative values for C3H8?

M9: Is it possible that the C3H8 regularization is too loose?

M10: If differences between retrievals and a priori data are big: How trustworthy are the a priori profiles of C3H8 from WACCM and are they very different from the ones in CAMS? Would it not make sense to scale only the lower layers of the C3H8 profile, instead of fitting/scaling the full profile, especially since the retrieved profiles do not appear to differ substantially from the prior above ~10 km? How many layers are there below 10 km?

M11: Which interfering species has the most impact on C3H8 retrieval accuracy/precision?

M12: Have you checked the correlation between C3H8 and H2O/HDO? What did you find?

M13: Maybe mention somewhere that the DLR and HITRAN2020 line lists for H2O are very similar; I believe for the main isotopologue: HITRAN2020 = DLR, except for line positions, but better to double-check.

Technical comments:

T1: page 2, line 4: “of the order of 10 years” -> “on the order of 10 years”

T2: caption of Table 1: “specie” -> “species”

T3: page 6, line 16 and caption of Table 3: 2040 -> 2004

T4: page 6, eq. 8: S_ij -> S_{ij}

T5: page 7, line 13: pressure-dependent parameter, temperature-dependent parameter -> pressure-dependence parameter, temperature-dependence parameter

T6: page 10, line 17: “might due” -> “might be due”

T7: page 11, line 14: “separately(“ -> separately (“

T8: page 11, line 16: “have” -> “has”

T9: page 11, line 16: “it is probably due to that the” -> “it is probably due to the fact that the”