The authors report on a DIAL system operating in the 1.645 um region to detect methane. The results are compared to a ICOS site and successfully measured increased concentrations of methane created by a fire. The work is well described and the results are impressive. I recommend publication of this manuscript with some minor changes.

Line 83.

“The wavelength measurement alternates between the two DFBs using a fiber switch (sSWT) toggling every 10 s, thereby tracking the slow drift of each seeder locked to the laser cavity.”

Line 180

“The Licel acquisition module integrates 2000 laser shots for both ON and OFF wavelengths in real time, resulting in ON and OFF profiles recorded with a temporal resolution of 4 s.”

Please clarify the above two statements. Is that 4s for both, or 2 sec each? Where did 10 sec go?

Figure 3. Switching the axis of the range is confusing

Line 259

“The uncertainty bars of the ICOS measurements represent the dispersion of the measurements within each hourly interval.”

Why not match the resolution of the lidar system? Is there a limitation, or is this a choice?

Line 267.

“The differences in methane concentration between the ICOS measurements (at 100 m) and the lidar measurements averaged over one hour up to a range of 3.5 km remain below 50 ppb. A fire occurred at a waste sorting facility located in the city of Paris on the evening of April 7..”

Roughly how far from the point sensor (at 100 m) is the center of mass of the lidar results?

Figure 6. Can you highlight the region in the left image used in the right image. The error bars and overlap is very difficult to see at the plum tail, maybe include zoomed in insert.

Review for 1.654 micron differential absorption lidar measurements for atmospheric methane using an Er:YAG laser by Dimitri Edouart, Fabien Gilbert, and Claire Cenac

This paper presents initial methane mixing ratio profiles retrieved from a DIAL instrument.  This paper is well written, complete, and can be published as is. 

Between line 30 and 31, A new paragraph might be added that discusses the need spatial, temporal, and resolution of the methane measurements needed by the science community to help put this work in perspective.

Figure 2:  Because you do not have temperature measurements, you would like to choose a temperature insensitive absorption line for the methane.  Can you comment on the temperature sensitivity of the absorption lines chosen?

Figure 6:  Using a swept window in both time and range may provide a better representation of the time series rather than the :pixelate” plot shown in figure 6 (left).

Line 259:  Why are you comparing the CH4 mixing ratio measure at 2.1 km when the tower observation is 5 km away?

Line 261-262.  Why is a line fit used to determine the differential absorption coefficient.   You should be able to directly use the return signal and the DIAL equation to retrieve the mixing ratio.  Perhaps a few sentences should be added to describe how you are completing the retrieval would be in order.

Line 364:  “Figure 10 (right) …”  Figure 10 is missing from my copy of the paper. 

This paper reports on an Er:YAG based methane DIAL for the retrieval of methane profiles when pointed horizontally. The paper is well structured, however, the structure causes some confusion on the logic behind selecting different resolutions and spatial averages.  I recommend publication after addressing some minor comments below as well as some introductory remarks on driving requirements for the spatio-temporal resolutions and what science can be achieved with those.

• Define ICOS acronym in abstract
• Line 27: define TCCON
• Line 74: When introducing the fSWT, please provide attributes like switching speed, and most importantly, what the optical cross-talk is as this will drive the spectral purity of the entire transmitter. Please also discuss the switching strategy. Is the wavelength switched on every shot? 
• Line 76: Please discuss the logic for the wavelength selections (i.e. MERLIN chooses offline wavelengths on the higher side of the online whereas HALO chose offline wavelengths lower than the online.  Please also discuss the laser stabilization approach. Does the laser cavity have enough bandwidth to acquire lock on each shot, or do you operate on integers of the free spectral range to alleviate the need for high-speed cavity locking?  I see this topic is touched on around lie 165, but more detail would still be good like what is the FSR, what is the metric used to ensure you are operating on cavity modes?
• Line 129-131: You indicate the signals are temporally and spatially averaged to meet SNR requirements, but do not discuss what the SNR requirement is, and over what scales averaging is required to hit the SNR requirement.  It’s also unclear what the electrical bandwidth of the receiver is?  Do you electronically band limit the receiver to match the vertical resolution after averaging/smoothing?  If no, please discuss the potential sources of noise aliasing by not electronically band limiting. 
• Line 161: It may be unclear to readers why you would opt to select H2O differentical cross section that is equal or close to zero, rather than minimizing it. Please elaborate on this topic and also include Refaat et al. 2012 where this original concept was presented. It would be good to include a vertical differential cross-section for CH4 and H2O to show the readers the vertical sensitivity to methane and water vapor.
• Section 3.1: Please elaborate on why you are only using a 25 point moving average (resulting in 47 m resolution on the DIAL retrieval). The differential cross-section over this delta R is almost negligible and therefore makes it difficult to achieve high precision.   Furthermore, increasing the range over which the signals are average increases SNR by deltaR^3/2 power.  Please discuss why larger range averages are not utilized and what science is driving to higher spatial resolutions.
• Lines 220-230: as with the previous comment, please discuss why you opt to significantly increase the temporal averaging at the expense of retaining higher vertical resolution?  Why not utilize a range dependent delta R for the DIAL retrieval to better leverage the 3/2 power scaling for the vertical averaging. A discussion on the error propagation as to how you arrived at these average schemes would be good for the reader. 
• Section 3.2 – it is now noted that the resolution is reduced to 470 meters, but it is framed with respect to the cross sections. Please discuss why the differentical cross-sections are computed over a 470 meter window and also discuss why the lidar signals are not smoothed to this same spatial resolution?
• Line 340: The laser spectral purity may be 99.6, however, the system spectral purity may be different owing to the performance of the fast optical switch.  Please report what the optical cross-talk is of the switch and how that affects the spectral purity of the pulsed laser.