Sea ice cover in Isfjorden and Hornsund 2000-2014 by using remote sensing

A new dataset of sea ice cover for two fjords in Svalbard (Isfjorden and Hornsund) is established using radar and visible remote sensing data from multiple products for the period 2000-2014. Remote sensing data are used to manually identify the areal coverage of “fast ice”, “drift ice” and “open water” in the Fjords. The timeseries of percentage of the fjords covered by ice is examined in relation to an index referred to as “days of fast ice coverage” (DFI), which is the cumulative fractional ice covered area for a given year and fjord. A similar progression of ice cover is found for both Isfjorden and Hornsund, with both fjords appearing to exhibit a substantial reduction in sea ice cover after 2006. The authors examine such changes in relation to oceanic and atmospheric changes discussed in relevant literature.

(2) The discussion section sometimes draws conclusions relating observed sea ice changes to changes in ocean circulation that are not necessarily supported by the data presented. Including ocean and atmospheric data in the analysis would help to better place the results in context of the discussed literature.
(3) The authors do not discuss the impact of mixing and matching different products in the analysis. In particular, what is the impact of different resolutions on the results? It seems that a lower resolution product might give different estimates of sea ice cover. For example, an entire 500 m grid box from MODIS could be classified as 100% ice-covered when an analysis of ASTER data could reveal that it is perhaps 60% covered by ice. The authors should attempt to quantify the impact of these effects and include a discussion of this in the methods section.
(4) The shift from 2005 to 2006 from relatively high to relatively low ice cover occurs just as the shift from visible data to SAR data occurs. It seems possible that such a shift could influence the timeseries, but this has not been discussed. The possible impact of the inclusion of SAR data should be analyzed and discussed.
(5) Errors are discussed in the data section, but not discussed anywhere else in the paper. The influence of errors on interpretation of results should be mentioned in the results and/or discussion sections. Errors should be quantified where possible.

1.
Title: I think a clearer title would be "Sea ice cover in the Isfjorden and Hornsund Fjords, Svalbard (2000-2014) from remote sensing data" or perhaps, "A new sea ice cover record for the Isfjorden and Horsund fjords (2000-2014) from remote sensing data" 2.
P. 4044, Line 15: Change "concept" to "index". I don't think "days of fast ice coverage" is an appropriate name for this index. (See later comments).
P. 4054, Lines 12-15: The time series does not show the influence of the forcing event on thermodynamics, it only shows changes in sea ice cover. Rather the sea ice timeseries and previous work seems to suggest that the change in circulation has an influence on the sea ice cover. Also, the forcing that occurs on a "timescale of weeks" is not discussed in the previous sentence. Please discuss the forcing event.

21.
P. 4055, Lines 2-5: Again, without any ocean data, I don't think it is possible to draw the conclusions from Figs. 2 and 3 that sea ice-cover is less influenced by AW intrusion. Also, I am not sure I really see variations on shorter timescales at Hornsund. It is difficult to draw that conclusion from the limited length of the timeseries.

22.
P. 4055, Line 9: It is not clear what supports the statement that Hornsund can be used as such an indicator. 23.
P. 4055, Line 12: What does "will be" refer to? Is this part of the AWAKE-2 project? 24.