Insights into the effect of spatial and temporal flow variations on turbulent heat exchange at a mountain glacier

Multi-scale interactions between the glacier surface, the overlying atmosphere and the surrounding alpine terrain are highly complex. The high heterogeneity of boundary layer processes that couple these systems drives temporally and 10 spatially variable energy fluxes and melt rates. A comprehensive measurement campaign, the HEFEX (Hintereisferner Experiment), was conducted during the summer of 2018. The aim of this experiment was to investigate spatial and temporal dynamics of the near-surface boundary layer and associated heat exchange processes close to the glacier surface during the melting season. The experimental setup of five meteorological stations was designed to capture the spatial and temporal characteristics of the local wind system on the glacier and to quantify the contribution of horizontal heat advection from 15 surrounding ice-free areas to the local energy flux variability at the glacier. Turbulence data suggest that the temporal change in the local wind system strongly affect the micrometeorology at the glacier. Low-level katabatic flows were persistently measured during both night time and daytime and were responsible for consistently low near-surface air temperatures with small spatial variations at the glacier. On the contrary, local turbulence profiles of momentum and heat revealed strong changes of the local thermodynamic characteristics at the glacier when larger-scale westerly flows disturbed the prevailing katabatic 20 flow forming low-level across-glacier flows. Warm air advection from the surrounding ice-free areas significantly increased near-surface air-temperatures at the glacier, with strong horizontal temperature gradients from the peripheral zones towards the centerline of the glacier. Despite generally lower near-surface wind speeds during the across-glacier flow, peak horizontal heat advection from the peripheral zones towards the centerline and strong transport of turbulence from higher atmospheric layers downward resulted in enhanced turbulent heat exchange towards the glacier surface at the glacier centerline. Thus, at 25 the centerline of the glacier the exposure to strong larger-scale westerly winds promoted heat exchange processes at the glacier surface potentially contributing to ice melt. On the contrary, at the peripheral zones of the glacier turbulence data indicate that stronger sheltering from the larger-scale flows allowed the preservation of a katabatic jet, which suppressed the efficiency of https://doi.org/10.5194/tc-2020-78 Preprint. Discussion started: 31 March 2020 c © Author(s) 2020. CC BY 4.0 License.

The manuscript would benefit from the addition of some context for the general meteorological conditions during campaign, especially timeseries of temperature and wind speed/ direction during the 5 selected days. This would provide the reader with a more intuitive introduction to the meteorology between relationships are discussed in later figures. These figures should also include an indication of time periods defined as 'katabatic' and 'disturbed' as this is unclear. In the discussion section, the authors should reflect further on the (potential) implications for measurements and modelling of turbulent heat fluxes, wind speed and air temperature distributions on other glaciers. Along with this the authors could provide more recommendations for future research.
Specific comments to improve the paper are provided below, but in general the paper is very well written, and figures well presented. My only concern with the analysis presented is the use of ratios to normalise temperature and wind speed in Figure 6, 7 and Table 1, and I would suggest the authors instead use anomalies (in K and ms -1 , respectively). This is especially important for temperature, where the fractional difference for the same change in temperature (in C) become smaller as daily mean temperature (in C) increases. If the authors wish to retain the current method, the theoretical basis for using ratios needs more explanation. The discussion of temperature differences between sites and situations is also very hard to compare with the current figures (see specific comments), but a change to anomalies and addition of timeseries of from each site should address this.
While the use of scatter plots makes it a little hard to interpret the density of data in certain figures, the ability to use colour warrants this approach. For some figures (Fig 9 and 10), histograms added along the x and y axes would enable the reader to see differences in the distribution that are discussed in the text (e.g. https://matplotlib.org/3.1.0/gallery/lines_bars_and_markers/scatter_hist.html).
In short, with some changes to clarify ambiguities of method and the presentation of additional results to support some statements, this manuscript will make a good addition to the literature.

Specific comments:
41 -the sensitivity of melt rate to air temperature is not only controlled by net longwave and turbulent heat flux, but also controlled by snowfall-albedo feedbacks -consider changing "controlled" to 'strongly affected' or similar.
48 -'several studies' -worth adding additional references to this sentence or rewording.
49 -"near-surface warming" -it is unclear what is meant here -the katabatic models discussed in the previous sentence predict enhanced turbulent heat fluxes due to increased wind speed, not temperature. Please revise.
122 -please list the model numbers of the other instrumentation, including the young anemometers, the 2d sonic anemometer and the air temperature, rh and pressure sensors. Please also note if the t/rh sensors were passively or actively ventilated and if any corrections were made to raw data aside from the eddy-covariance data.
127 -it would be useful to expand further on the choice of 1-minute averaging period, as this departs significantly from often-used averaging periods of ~30 minutes. Perhaps present some of the analysis mentioned or comment on the effect of the short averaging period on, e.g. average heat fluxes. 223 -'Flux footprints tend to be smaller during disturbed situations." Figure 3 shows a larger overall footprint area -perhaps worth clarifying that footprints for individual periods are smaller but the more varied orientation during disturbed conditions results in a larger overall footprint, if this is the case.
227 -Do you think the different instrumentation contributes significantly to the differences observed between level 3 and the lower two levels?
227 -Do you mean a secondary larger-scale wind system above level 2? If so, please clarify.
234 -"This extreme increase of wind speed with height is confirmed by preliminary numerical simulations (not shown)". As the reader cannot assess this without presenting the data, please remove or modify this sentence.
268 -'on 2018-08-20' -I presume you mean on all case-study days? Please revise 277 -'the temporal variability of flux profiles increased significantly for disturbed situations' -it is very hard to assess this statement from Figure 5 -please add further statistics to describe the mean and variability of the fluxes or reword. Figure 6 -consider moving TT3 to the x axis on these plots as it is functioning here as a common variable (hence is more like the 'independent' variable). Figure 6 -it is hard to assess the density of points in the scatter plot -consider using a transparency for the points so that more dense data shows as darker shades. Figure 6 -the colour scale for disturbed conditions would be better to avoid white tones as the are hard to read. Scale used in Figure 9 would be better.
308-332 -there are many statements in this section at are not clearly supported by the data presented in Figure 6. The addition of timeseries of WD/WS and temperature from multiple sites would be of great benefit here.
310 -"significant increase in the near-surface air temperature of several degrees (Fig. 6d-f)" -this cannot be ascertained from the current figure 6 as the units are normalised. Please use anomalies as suggested in general comments section or provide additional results to support this statement.
314 -"Local air temperatures at the higher altitude station TT4 showed the lowest sensitivity to changes in wind direction at TT3." It is unclear how the data support this statement -please clarify and revise.
315 -"The katabatic flow seemed to persist at the higher altitude station TT4 when at the same time all transect stations already evidenced a westerly flow (Fig. 6b)." It is unclear how the data support this statement -please clarify and revise.
317 -"Air temperatures at the glacier tongue (WT1) appeared to be strongly affected by up-valley flows (Fig. 6f)." It is unclear how the data support this statement -please clarify and revise.
326 -"explain a larger spatial variability of the air temperature" -It is unclear how the data support this statement -please clarify and revise.
329 -Are the cooler temperatures during katabatic flows affected by diurnal changes in temperature? Ie. are katabatic conditions more common during cooler periods at night time? Table 1 -what is UT ?
342 -'all four turbulence stations' do you mean 'all three turbulence stations' or 'all 6 turbulence sensors'. Also please list what height data is from 361 -'showed small spatial differences' -this is very hard to interpret from Figure 7 -a histogram of differences between fluxes at different stations would support this.
362 -"despite significantly higher air temperatures observed at TT1" -this is not shown and needs to be supported by additional results -perhaps a histogram of temperature differences between each site in different conditions.
388 -what fraction of periods were excluded? 423 -"Similar to heat advection, peak vertical turbulent heat fluxes coincided with peak Vcomponent at the centerline." -to what extent is this due to the correlation between mean wind speed and vertical fluxes? Please discuss.
424 -"Correlation coefficients R(w´T´,UT) were high between TT1-TT2 and TT2-TT3 station pairs with a slightly higher value for stations closer to the centerline." It is unclear how this relates to the data presented in Table 1. Please revise.  509 -The steep moraine sides are likely to play a role in the sheltering of the site closest to the glacier margin, especially considering the sharp slope transitions and short distances involved. Thus, the flow hitting the glacier edge may not be well developed and still be affected by lee-side flow separation etc, reducing its ability to influence the stable glacier boundary layer. This may be worth discussing further here.
528 -as the study only presents data from 5 days, it would be more meaningful to say "during five days that displayed a distinct disruption of down-glacier flow during a three week period in summer 2018." Or similar. 541 -'induced by strong westerly winds' -while this makes sense, the origin of the flow is still speculative so please revise.
552 -'At the peripheral areas stronger exposure' -shouldn't this be 'weaker exposure'.
552 -As wind direction is not presented for TT1 it is impossible to assess if the 'preservation of a very-shallow low-level katabatic jet' is supported by the results. Figure 1 shows the WD is aligned at all levels at TT3 during disturbed situations -in order to support a katabatic jet at TT1 the wind direction would need to be maintained down-slope. The BL could still be decoupled at TT1 because of the strong thermal stratification, but this does not necessarily mean that a katabatic jet will exist at TT1. Please revise.
575 -"the frequency of such flows at other glaciers is not known" -this comment highlights that fact that the frequency of these flows has not been presented in the current study. This would be an easy and useful addition to the results.
Editorial comments: