|Referee review for manuscript tc-2016-06-manuscript-version-3, submitted for publication in The Cryosphere|
"Surface Energy Balance Sensitivity to Meteorological Variability on Haig Glacier, Canadian Rocky Mountains” by S. Ebrahimi and S. J. Marshall
The revised manuscript is an improvement on the initial submission. In particular, the model now includes more dynamic feedbacks which has increased the confidence with which the results can be interpreted. The re-setting of the NARR results as an extension of the sensitivity analysis is good to see and improves the focus of the paper. The authors appear to have considered and addressed the main points in their response to my first review. However, it was difficult to assess the actual changes made to address each point as no text changes were given in the response and a marked up version of the manuscript was not supplied.
Also, the paper still contains many ambiguities of method and much inference that isn't always supported by robust results.
The main result appears to be an extremely large increase in temperature sensitivity when albedo is allowed to vary. While one would expect the sensitivity to increase, the magnitude of the increase here needs to be better supported by the validation of the albedo scheme against measurements and further discussion or analysis around the role of impurities to justify the low minimum value for snow albedo used. The choice of minimum values of albedo for snow and ice also impacts the conclusion that winter balance is of less importance, as with higher minimum albedo values the contrast between snow and ice is larger and this will increase the sensitivity. The authors also need to comment on the processes driving this sensitivity - to what extent the sensitivity is driven by decreases in snow albedo, earlier transition to ice and summer snowfall.
The use of parameterisation for the radiative fluxes that responds to various drivers is encouraging. There is still a need for a validation of the melt produced by the fully parameterised model against that driven with measurements if the results are to be trusted. It is also ambiguous which version of the model is being used for any particular run and this needs to be carefully detailed and explained.
The authors identify a series of significant feedbacks between air temperature, humidity and the incoming fluxes of short and longwave radiation. It would appear most of these are all connected through cloud cover, and it would be much clearer and insightful for the authors to explicitly examine changes in the frequency and attenuation of cloud cover alongside with air temperature and humidity, rather than inferring these from the sensitivity of incoming shortwave radiation in the analysis of the in-situ dataset. This should be possible as the authors already have a parameterisation of cloud cover that includes humidity. If not, then the variability in incoming longwave also needs to be included in Table 4.
In general, the paper is fairly well written though some of the text in the results and discussion is quite methodological and repetitive and perhaps is better suited to a methods section. Further work is needed to distill the main results from a rather large body of work and succinctly present them here.
The authors have managed to refocus their analysis but the new results and novelty of their results are often quite hidden. I would suggest re framing the discussion around 1. A summary of the important sensitivities and feedbacks that are observed on the Haig, 2. A discussion of the utility of the theoretical sensitivity based on mean summer conditions (good correspondence with full summers when feedbacks are omitted) 3. A discussion of the utility of exploring sensitivity to inter-annual variability with reanalysis datasets (mixed results). This structure would alert the reader to what is new and avoid some of the repetition.
204 - replace ‘profile method’ with ‘bulk aerodynamic method’. The profile method uses wind speeds and temperature at two heights.
244 - the symbol psi has a different case from the equation.
249 - please include an explanation of how the refreezing rate is calculated and what constraints are put on the volumetric water content.
278 - for consistency please state the model uses a variable timestep from 10 minutes to 1 hour to allow for stability of the subsurface temperature prognosis.
349 - the minimum value for snow is very low. For the same site, Marshall (2014) gives 0.4 as a minimum. This will have a large impact on the sensitivity. Other authors have used values around 0.5 (Oerlemenas and Knap, 1998) and further justification of this low value is needed.
405 - need to provide more discussion and evidence for the reasons for the decrease in albedo - i.e. is the increase in particulate concentrations documented?
419 - the contribution to melt should really be computed for melting periods only, as non-melting periods will bias these fractions towards the sensible heat flux (see Conway and Cullen, 2016). Please either show the fluxes for melting periods only or discuss only the contribution to the energy balance and not to melt.
423-430 - please make it clear that the feedbacks and NARR analyses are presented in the following two sections, rather than the current section.
434 and 439 - some more context is needed to justify conditions on the Haig being 'typical' of other mid-latitude glaciers. Please add either a table showing this or some references to papers with similar climatologies.
444 - need further analysis in order to justify that the JJA has more impact on melt? Table 3 shows an almost identical combined sensitivity.
442-448 - this is quite a confusing paragraph as the rationale for including/excluding months changes from the start to the end of the paragraph and results are introduced, but not referenced properly. Please either point to the figures/tables that justify these statements or move this text to the discussion.
453 - Are these perturbations calculated as the average of positive and negative deviations from the mean? Some of the text (e.g. 518) seems to suggest that only positive perturbations were considered, which is not ideal.
503 - I am not sure at this stage it is appropriate to transfer the calculations of net energy to melt, as in reality not all periods will have melting conditions. Perhaps it is better to state the increase or decrease in the net energy available. Along with this I would remove the melt column from table 3.
516 - please be consistent with the symbols used - the text uses vapour pressure while table 3 shows specific humidity.
604 - Please be explicit this is top of atmosphere solar variability.
620 - For comparing the relative importance of each variable - it would be more useful to present the individual sensitivities relative to 1 standard deviation perturbations in Table 3.
630 - The way these variables have been perturbed is not meaningful as they are physically unrealistic. For example - some of the standard deviation in vapour pressure will be due to increases in relative humidity, but you also increase incoming shortwave in this experiment - which as you noted earlier is likely to decrease with increased relative humidity. Thus, the experiment is contradictory. Please exclude these last two lines in Table 3.
677 - It is still ambiguous which variables are held constant at their measured values and which are parameterised for each run (in particular incoming longwave and surface temperature). Please provide a comprehensive table.
680 - why were changes in incoming longwave not examined?
764 - this result is likely to be strongly dependent on the choice of the minimum albedo of snow, which in this case does not differ much from an ice albedo. Either the sensitivity of this result should be tested, or a more thorough justification made for the very low value chosen here. (0.3)
776 - are the anomalies calculated with respect to the mean in-situ conditions? I suspect you took the daily anomalies of NARR from the NARR climatology, then applied these daily anomalies to the mean in-situ conditions - please clarify.
792 - it would be useful to see the standard deviation of relative humidity included here.
801 - this line needs more context to link it with the previous sentences.
970 - do the interannual anomalies in SWin and LWin from NARR correlate with the anomalies from the in-situ dataset? If not, it is hard to see how the NARR represents realistic interannual variability in these fluxes. This severely limits the inferences that can be made from model runs made with these anomalies.
1005 - this statement needs more justification. It would seem that the NARR based reconstructions performed satisfactorily in describing interannual variations in net energy flux, but that this is based on the accidental cancelation of errors in the radiative fluxes driving the model (Figures 5 and 6 from the original manuscript. The approach is worthy of further exploration, but a more thorough evaluation of the performance of the model, including biases and areas for improvement, is needed here.
1413 - needs a legend describing the colours. Also, box 4 could be better as a separate figure, as the colours indicate the change in Qn due to different forcing, while in the other boxes the colours show the response of different fluxes to the same forcing.
1415 - both black and red are listed as net radiation - please fix.
1416 - add 'please note the different y scales'.
1416 - please clarify in the caption which scenarios these figures relate to in Table 4.
1277 - net melt has decreased 7-8% while net energy fluxes have remained similar. It would be useful to show the fraction of time the surface is diagnosed as melting in each month to provide some justification here.
1388 - I am not sure this figure adds much as you cannot see the detail in the daily values over the 11-year period, and the results presented in the figure are not discussed in the text. I would suggest either removing the figure, or modifying the figure so it is readable and discussing the results further. If the figure is kept the size of each box needs to be expanded and the line weight reduced to make a more readable figure.
1394 - please use thinner lines on these figures. Also, as months and not day of year are discussed in the text, it would be good to have months as the x-axis label, or at the very least, further tick marks that are at monthly or 30 day intervals.
On panel d, the median + interquartile range would better present the seasonal variation of melt rate. As it is, the mean appears to be greatly influenced by individual large melt events (such as around day 230).
1430 - the y axis of figure 7a should be m w.e. in line with the text and Figure 7c.
1469 - 1495 - were these figures meant to be included?
Conway, J. P. and Cullen, N. J.: Cloud effects on surface energy and mass balance in the ablation area of Brewster Glacier, New Zealand, The Cryosphere, 10, 313-328, doi:10.5194/tc-10-313-2016, 2016.
Oerlemans, J. and Knap, W. H.: A 1 year record of global radiation and albedo in the ablation zone of Morteratschgletscher, Switzerland, Journal of Glaciology, 44, 231-238, 1998.