Review of “Understanding the drivers of near-surface winds in Adélie land, East Antarctica” by Darvinche et al.

This manuscript presents a method for diagnosing the terms in the horizontal momentum budget, including separation of the pressure gradient force into terms due to the large-scale pressure gradient and pressure gradient terms driven by near surface temperature distribution (katabatic and thermal wind). This decomposition is used to assess the relative importance of the different forcing terms on spatial wind speed variability and wind speed variability on seasonal and 3 hourly time scales. This work builds on early similar research (Parish and Cassano, 2003; van den Brooke and van Leipzig 2003; Bintanja et al. 2014). The results are worthy of publication and the manuscript is generally well written. I do have some concerns, described below, which warrant major revisions before the manuscript can be accepted for final publication,

Major comments

The method of determining the minimum height above which the potential temperature profile is assumed linear is not explained clearly but seems inaccurate. This is a fundamental aspect of the work presented and will impact all additional results so it is critical that this method is clearly described, accurate and justified. As described starting on line 129 an initial linear vertical potential temperature gradient is estimated between 500 and 350 hPa. I assume that this is done at each 3 h time step and for each model horizontal grid point but this should be stated explicitly in the text. The vertical potential temperature gradient is then calculated and compared to the gradient between 500 and 350 hPa. I assume that this new gradient is calculated at each model vertical level moving up from the surface but this is not stated explicitly and needs to be. The minimum height to be used for the assumed linear potential temperature profile is then based on the height at which the vertical potential temperature gradient exceeds the average gradient between 500 and 350 hPa by a factor of 5. The logic in this seems flawed since what is desired is separating the portion of the profile, near the surface, where the gradient varies with height from further aloft where the gradient is nearly constant with height. Using a constant factor to compare the gradients only determines the height at which the gradient is larger than that in the 500 to 350 hPa layer, which is not the metric that is relevant. In this case what is relevant is assessing how the gradient changes with height - a second derivative of potential temperature with height. When this second derivative becomes small enough the profile can be assumed to be linear. The authors should consider using this more direct way of assessing the height at which the the potential temperature profile switches from being curved to being linear. If the original method will be retained the authors need to better justify this approach by showing a comparison with the more direct method described here. And, the text needs to more explicitly describe the process used to determine this minimum height above which the potential temperature profile is assumed to be linear.

Another concern comes from using 350 hPa as the upper height for the linear approximation of the potential temperature profile. How often is this height above the tropopause. It seems like it would be better to calculate the linear profile over a fixed depth above Hmin - maybe just 100 or 200 hPa - to minimize the possibility of estimating a linear gradient over different layers of the free atmosphere with possibly different air masses and potential temperature gradients.

The use of the term thermal wind in your decomposition is confusing. The thermal wind, as defined in atmospheric dynamics text books (e.g. Holton and Hakim) refers to a change in geostrophic wind over some depth of the atmosphere. This is not what this term represents in your decomposition? Parish and Cassano (2003) have the same term in their decomposition and refer to it as the integrated deficit term while Cassano and Parish (2000) referred to this as an adverse pressure gradient force term since it often opposes the downslope flow due to a deepening of the boundary layer with downslope distance and thus a larger integrated potential temperature deficit. This term needs to be renamed to more accurately describe what it represents physically.

Figure 4: There are obvious discontinuities in the pressure gradient force components (e.g. KAT and THWD between D17 and D85) seen in this figure. The source of these clearly non-physical results need to be discussed. Do these artifacts reflect a shortcoming in the decomposition that makes the results less trustworthy?

Figure 7: I found that showing the direction of the momentum budget terms as the equivalent geostrophic wind to be confusing. It would be clearer to simply show vectors in the direction of each momentum budget term scaled by their magnitude. In this way it will be clear in which direction each force is acting rather than the reader needing to rotate the vectors mentally by 90 deg. If the authors wish to keep the vectors scaled relative to a geostrophic wind speed the magnitude of each term can simply be divided by the Coriolis parameter, which will retain the same magnitude as currently shown in Figure 7 but without the direction being rotated 90 deg from the true direction each force is acting.

Minor comments

Line 14: remove latitudes after sub polar - it is redundant and not needed

Figure 1 caption: Last sentence of caption describing color of dots does not match what is shown in the figure.

Line 69: model should be model’s

Line 73: What is meant by “the data are slightly better correlated to our model”? This sounds like you are selecting observational data that matches the model which is not appropriate - you cannot preferentially choose observations that match your model and ignore and de-emphasize those that don’t.

Table 1: List lon, lat and elevation for DC-tower. I assume that this is the same as for the DC-AWS but this should be confirmed by listing these values in the table.

Line 86: What is meant by model bases? Please clarify.

Lines 93-94: As written it seems like the 30 snow/ice layers are each 20 m thick. I think what you mean is that the total depth of snow/ice is 20 m and there are 30 layers distributed over this depth. Please rephrase.

Table 2: It would be more informative if the table listed the start and end distance from the coast for each section and gave the range of terrain slope in addition to the average slope.

Line 111: winds variability should be wind variability and near-surface should be capitalized since it is at the start of new sentence.

Line 132: larger that should larger than

Line 192: Delete boundary between surface and layer. I think you are referring simply to the surface layer here.

Table 3: It would be helpful if the relative magnitude of each term was given. For example, the terms could be normalized relative to the LSC term or total PGF to indicate how much larger or smaller each term is relative to the LSC or overall PGF forcing. This could be given as a percentage in parenthesis after the seasonal value is listed. The total PGF should also be listed in this table.

Figure 7: It would be helpful to add a panel showing the total pressure gradient force, which can then be compared to the other terms in the momentum equation.

Figure 7: Similar to the comment regarding Table 3, showing figures of the ratio of KAT, THWD and TURB to LSC would be very helpful, especially if a color bar with different colors above (forcing greater than LSC) and below (forcing less than LSC) was used. This would clearly show where each forcing term exceeds the LSC forcing.

Line 330: Replace outputs with output