see attached referee report. 

This paper describes a jökulhlaup model, based on R-channel drainage, taking into account effects of saline water on the dynamics.  It finds that the impact of salinity on melt-rates is limited and the biggest impact is due to the increased density of the fluid on hydraulics.

This short paper has interesting findings and is novel as no-one has looked into impacts of salinity on R-channel dynamics.  It has however a few shortcomings which need to be rectified before publication.

Major comments

==============

The somewhat surprising take-home (that the main difference compared to freshwater discharge is due to the higher density and not due to higher melt rates) should be elaborated a bit more.  I would state the following, possibly in the Abstract & Conclusion:

- salinity makes liquid water possible at sub-zero temperatures: fresh water would be frozen.  So this is really the biggest difference: liquid water versus no liquid water.

- once the equations are solved at the sub-zero temperature given by the melting point of the saline solution, it is actually pretty obvious that impact on melt is minimal: dilution of the water as it traverses the R-channel is minimal as melt is really small compared to discharge.  Even for a setting where there is much more potential energy available, this statement holds.  Thus it makes sense that density has the biggest impact.

Provide a sketch (as fig. 1) of the setup for an overview, that way also the coordinate system and orientation are defined.

With many of the equations in section 2 I struggle:

- Eq6 and the following three unnumbered eqs have some issues, see line-by-line comments

- my understanding is that the authors assume that water-temp equal to the local melting point as stated on l72 (i.e. temperature is dictated by pressure and salinity).  Thus \theta, \theta_i and \theta_b are equal and hence the last term in Eq7 is zero.  If temperature was treated as a free variable, as in Fowler 1999, then this term would be needed.

  --> this whole temperature treatment is confusing.  I had a quick look at the code and I think that temperature is indeed not a free variable.

- the melt rate $m$ can be solved for in Eq7.  However, I do not understand what is going on here.  As stated above, the last term should ==0.  Also there should be a pressure dependent term to capture the pressure melting point effects.  At least when a linear relation between temperature and pressure is assumed then it takes the form $rho_w c_t  c_t dP_w/ds*Q$ (e.g. Röthlisberger 1972, Eq 17); maybe the authors try to capture this with their last term of Eq7, but I don't think that can be done like this.  A term in similar spirit would need to be added for the salinity dependent effects, presumably featuring a $d\beta/ds$.  In fact, this is the quintessential equation to be stated as that is the novel one, all others are known.

- How Eq15, also featuring the melt rate, then ties in with this is also not clear to me.

The lake has only ever a simple geometry, thus I would recommend to just keep it simple and shorten that part.  A simple "refer to Kingslake 2015 for a treatment of more complicated lake hypsometries."

Line by line

============

12: nice Introduction

13: (any many others): I prefer incomplete lists of references to be prefaced by a "e.g.".  Many instances in the Introduction (l.15, 18, 24, 26, etc.)

30: I would prefer to use an SI units throughout instead of "psu".  At least give a translation (1psu=1kg/m3).

70: Cite the original too, Röthlisberger (1972).  I would write something like: "We construct a lake-drainage model in which the water flows through a subglacial channel (Röthlisberger, 1972; Nye, 1976).  We follow the implementation and notation of Fowler (1999) and Kingslake (2015).  We assume..."  I also like the term "R-channel" and would use that, but no strong feelings here.

74: I disagree with Fowler (1999) here and thus with this manuscript: the given definition is the "negative basic hydraulic potential".  I think it would be good to state the "negative" somewhere.  Similarly Eq2 gives the negative potential gradient.

95: the "(instead of Darcy-Weisbach)" I find confusing as D-W is also for freshwater.  Just delete?

97: I think 0.6m^{-1/3}s should be 0.06 (according to the code).  Also cite where these values are from (maybe Clarke 2003?) as Fowler 1999 does not provide those values.

Eq6 and following: I don't think those formulas should be stated with so many digits.  At most two significant digits are needed.  Also state that for the $\rho_b$ and $\Delta\theta_P$ the 3rd and 2nd order terms are not important.  Also, Eq6 is written with \beta as an argument, the others are without argument: should be consistent.

Eq6: is this correct?  I evaluate $\rho_b(0)=0.1$ but fresh water has a density of 1000kg/m3.

Eq6 / l109: I don't understand how the melting point depression can have a constant term.  That would mean that it is always depressed!  Similarly for the change in melting-point due to pressure.  Also, I guess these formula depend on temperature being given in deg-Celsius, that should be stated.

123: cite Röthlisberger here too, drop Kingslake.

132: why suddenly salinity in kgm^-3 instead of "psu"?

139-143: this can be abbreviated to just state Eq 8.  The reader does not need to see a derivation of a conservation equation.

115: doesn't line 72 state that these temperatures are always equal?

167: I wouldn't call this a "no flux" boundary condition as there is water flux through the boundary.  Just call it a Neumann BC.  Also, what does it mean?  What is meant with "We do not require the channel to exit at a terminus or end subaerially"?  Where does it end?  Or at least where does the simulation end and how do we know that there dN/ds=0?

174: reference the suggested Fig.1 (the sketch) here.

Fig.1: could the temperature as a function of distance also be plotted?  Maybe even the different components, i.e. press-melt-point term and salinity-melt-point term.

230: I think this is vice versa.

242: what are "hyperconcentrated sediment concentrations"?  What is this "threshold"?

253-259: I recommend to delete these lines as they don't add much

288: I concur with Reviewer 1: these N values do not make sense to me.  Is this caused by the Neumann BC on N?  Note that a, in my opinion, more natural and standard BC N=p_i (or p_w=0) leads to dN/ds not equal zero. See for instance in Röthlisberger (1972): Fig.5d shows a setup similar to here and shows p_w -> 0 at the outlet.

312: Excellent that the code is provided, thanks!  Ideally the authors should also provide a third script that, when run, will reproduce and write to disc all figures used in the publication.