|The manuscript has been substantially improved. However, the rationale of comparing snow column BC mixing ratios (BCC) and near-surface atmospheric concentrations of BC (BCS) is still unclear (i.e. page 12973, lines 6-11 in original manuscript; page 16, paragraphs following line 327 in revised manuscript). There is still a general lack of focus, as was pointed about by the other referee. The former is a critical shortcoming of the study because the authors hinge the second part of the paper and key conclusions on the proposition that biases in BCC and BCS somehow provide evidence for a missing source of BC in the model without sufficiently justifying this. For instance, the authors argue that deficiencies in emissions inventories are not likely the primary explanation for the model under-prediction of BCC (page 17, line 345). Instead the authors consider a lack of an additional source, and more specifically, a soil source of BC to be a more plausible explanation of model biases (page 18, line 368). It is still not clear how the authors come to this conclusion.|
The use of comparisons of BCC and BCS biases to argue for missing processes (or emissions) in the model is misguided since both quantities are affected by fundamentally different processes, none of which are accounted for in the analysis in the manuscript. The importance of different processes for explaining differences in BCC and BCS biases is clearly evident from results presented in the manuscript. In particular, Fig. 7 shows that deposition and atmospheric concentrations of BC have very different spatial distributions. Furthermore, section 3.3.1 (page 19) provides a useful discussion of differences in atmospheric concentrations and deposition fluxes. However, the analysis of BCS and BCC does not account for any of this.
Generally, concentrations of BC in the snow (BCC) strongly depend on aerosol wet removal by precipitation and snow microphysical processes during the formation of the snowpack. In contrast, atmospheric near-surface concentrations (BCS) are probably more sensitive to local emissions and near-surface atmospheric processes, which occur on very different time scales than processes that affect BCC. Comparisons of biases in BCC and BCS in the manuscript are based on averages of these quantities over the same short (3 month) time period, without accounting for impacts of precipitation nor any other processes on analysis results. These are much serious problems than the sparsity of data which is mentioned as possible source of analysis uncertainty (page 16, line 334).
A fundamental re-think of the analysis approach is necessary. What information about model processes (or their lack thereof) can be derived specifically from an analysis of BCS and BCC? How do model biases in deposition rates and other processes affect the results? How plausible is a soil source of BC for explaining differences in simulated and observed BCC concentrations? Is it possible to constrain potential soil BC emissions by considering soil fraction and dust mobilization processes in model sensitivity simulations for a range of plausible conditions? Is there any evidence that regions with more persistent and widespread snow cover are less affected by soil BC emissions than regions with more exposed soil?