Please find my comments in attachment.

Review of 

A high-resolution pan-Arctic meltwater discharge dataset from 1950 to 2021

by Igneczi and Bamber

General

This paper presents a new daily, 250 m resolution, 71-year runoff dataset for the Arctic, partitioning between runoff from ice and tundra. Its main strengths are the high temporal and spatial resolution, long time series and consistent source data treatment. Its main weaknesses are using only a single model product and the lack of detailed (regional) evaluation. This makes it hard to judge whether the (sometimes significant) differences that are found when comparing with previous products represent real improvements. See major comments below.

Major comments

l. 117: If you go from 90 to 250 m resolution, how do you deal with fractional ice cover?

l. 141: Was the RCM forced hourly by ERA5? Usually this is every three hours. How do you assess non-glaciated runoff from regions not covered by the MAR domain?

Section 4.1: Although I appreciate that the authors prefer consistency in their calculations, it would be good to show/discuss the potential impact of solely relying on surface routing over e.g. the Greenland ice sheet.

Please briefly discuss how findings in this recent paper, which shows that meltwater is stored in the glacial Greenland system for significant amounts of time, could affect your results: https://www.nature.com/articles/s41586-024-08096-3

l. 260: What albedo product was used in MAR? Albedo does typically not vary smoothly, making it harder to downscale as a function of elevation. Allowing albedo only to increase with increasing elevation may not be a valid assumption in many areas. How are the albedo corrections used in the final runoff product?

Section 4: Although useful, a bulk evaluation does not necessarily align with the bulk of the applications and users, which may well predominantly use dingle basin timeseries. 

l. 374: "We propose...". This and later hypotheses can be -at least partly- confirmed or rejected by comparing the runoff products in elevation bins: is the difference indeed deriving from the lower elevations which are better resolved? Same for non-Greenland ice. 

l. 392: I find it unlikely that resolution is the only/leading explanation for the large differences in Greenland tundra runoff. This can be relatively easily checked by comparing total tundra area, the depth of the seasonal snow cover and rainfall. This can also be used to provide a more robust answer to the question why tundra runoff outside Greenland agrees better (although the variability is again more different).

IN general, I miss a direct comparison between the non-downscaled and downscaled products. Where/when do the differences occur, and can it be objectively assessed whether the downscaling improves upon the original products? It presumably does, but unless it is somehow quantified this remains speculative.

Minor comments

l. 25: warmed -> increased (my strong preference!)

l. 159: This equation holds for runoff from land ice, please specify.

l. 340: If find the reasoning for distinguishing runoff from above and below the snow line hard to follow. Why is it relevant? Figure 6 suggests that the large majority of runoff comes from below the snow line. Interpretation?

l. 410: What is meant by "its overall uncertainty"? I presume you mean the uncertainty in runoff?