General Comments:

This manuscript describes the novel application of a scanning open-path laser absorption system to the real-world problem of fugitive methane emissions from Alberta oil sands tailings ponds and mine faces. The 2D methane concentration fields obtained with the laser system are used in conjunction with an inverse dispersion model to derive emission estimates. The advantages of this approach regarding the establishment of continuous spatially representative time series of emission rates is demonstrated, particularly through the quantitative characterization of the spring pulse in the methane flux from a pond as the ice breaks, something that has long been speculated to be of potential importance.

A comprehensive comparison of the open-path laser based results with previously published fluxes is presented and convincingly illustrates some of the issues with using snap-shot flux chamber measurements to try to quantify emissions representative of a large, heterogeneous surface source. Of course the presented method is not without issues; as explained below, there are reasons to believe that the derived emission rates are more likely to be under- rather than over-estimates. Nevertheless, methods such as the one described in this manuscript are approaching maturity and will hopefully help to modernize industrial emissions reporting protocols in the coming years.  

The manuscript is well written and can be published after minor modifications. Details on suggested improvements are given below.  

Specific Comments:

Page 5 Line 9: specify the period of time

P10L19: Using the lowest concentration inside your relatively small domain in close proximity to strong sources may not give you a proper background concentration. Overestimating the background concentrations will result in an underestimate of the emissions. Can you add some evidence that your background concentrations are low enough (by comparison with time series from other measurements, for example)?

P14L10: As I see it, it is an indication that the diurnal cycles of the concentrations are due to a common (probably meteorological) driver, but I’m not sure you can deduce that there is no bias due to local sources.

P22L12: these errors are so tiny compared to most of the other uncertainties involved that they’re not really worth discussing in this much detail – unless you can separately quantify and discuss other errors (e.g. background concentrations, winds, non-stationarity of meteorological conditions) in similar detail.                                                                                                                                                                                                                                                           

P23L1: I think the DEM / height above ground issue may be significant and worth a bit of extra effort. If the depth of the mine (below the measurement level) is underestimated, then the wind speeds may also be underestimated, and by extension the same will be true for the dispersion / dilution of the emitted methane, resulting in an underestimate of the emissions. Perhaps the fastest way to get a rough quantitative estimate of the magnitude of the difference would be a Monte-Carlo approach of varying the elevations of the beam paths above ground, to simulate the mine being deeper than given by the DEM. Also worth exploring might be the dependence on surface roughness (z₀).

Figures and Tables: there are more figures than necessary, with some redundancy that can easily be fixed. A few figures, and all the tables, should be moved to a Supplement to improve the flow.

Specific suggestions:

Figure 3 is mostly a repeat of Figure 1. Combine these two, and mark the center release points.

Figure 4 currently appears in the text before Figure 3; relabel.

Utilize the space in Figure 6 more efficiently by eliminating the large white space in the middle; use a broken x-axis.

Insert the data in Figure 8 into Figure 9; this will make it easier for the reader to line up the melt, and it reduces the figure count. Mark the April 28 peak on the graph for easier identification.

Eliminate Figure 7; redundant.

Figure 10: mark the daylight / nighttime difference by shading.

Figure 12: the correlation is surprisingly good! You can add error bars (interquartile range or so) to indicate the high uncertainty associated with the first monthly median.

All tables should go into the supplement. The information in Tables 1 and 2 is already present in Figs. 11 and 14.

Technical Corrections:

Throughout, references still need to be standardized (currently a mixture of footnotes and citations) and completed (e.g. “et al” if multiple authors).