Circumpolar polynya regions and ice production in the Arctic: Results from MODIS thermal infrared imagery for 2002/2003 to 2014/2015 with a regional focus on the Laptev Sea - Revision 01
Preußer, A., et al.
I thank the author team for their careful consideration of the reviewer's comments and concerns.
Reading the rebuttal letter doesn't leave anything open except one thing:
I wrote: Page 6, Line 14/15: I understand that the authors mention March here as this month contains the spring equinox. However, November is almost as close to the winter solstice as February is. Could it be that in November the cloud coverage is the problem?
You replied: Including the months of October and April would be problematic since the amount of suitable clearsky and nighttime MODIS scenes decreases with increasing amounts of solar radiation.
My comment (again): In the 1st version of the manuscript you wrote:
Because of the restriction to nighttime scenes, a less frequent IST coverage is present in the beginning (November) and at the end (March) of each winter-season.
Winter solstice is on December 21, right? Therefore sun elevations and hence day lengths are similar on November 21 and January 21, on October 21 and February 21, and so on. From that I'd expect that the number of nighttime cases is similar for the month pairs December/January, November/February, and October/March. Therefore I felt that mentioning November and March in one sentence explaining the limitation to fewer valid cases does not fit and was suggesting in my former review that clouds might have had a larger impact in November - which is something found in the literature as well.
The manuscript has improved substantially and is almost ready to go.
However, I still have a concern about the usage of polynya area and the associated ice production. The fact that previous studies used similar regions as are used in this study should not prevent the authors from pointing out potential shortcomings of those definitions.
- I strongly suggest therefore that the authors clarify their understanding of a polynya (in the context of how WMO defines it) and their usage of polynya area.
- I suggest further that the authors think one more time about the way they include the results of regions WNZ and KAR into their analysis and results and state even more clearly that their results of particularly these two regions are heavily influenced by ice formation in a MIZ rather than in a polynya.
These two concerns plus the specific comments led me to the suggestion that the authors should have the chance to for a minor revision which are to the authors' discretion, i.e., there is no need for me to have another review.
Specific comments / suggestions / typos:
P4, L31: Please check meaning of "inhibit". I would go for "contain" or "exhibit".
P5, Figure 2, Caption, 3rd line: "'IST' denotes to ..." Please either change "to" to "the" or omit completely.
P6, Lines 6-12: I suggest to add the information that the approach assumes a sheet of ice being present, i.e. that is not able to include the frazil / grease ice stadium.
P6, Line 23: The authors could motivate the usage of the median instead of the mean in one additional sentence.
P6, Line 31: "completely uncovered pixels" means that there are no MODIS data available?
P8, Line 16: "weighted average" Does this averaging use the same weights as mentioned above in Lines 11-12?
P8, Line 26: Please define "POLA".
P10, L9: How is Q_ice computed? I assume it is computed per pixel and it is computed as a function of the daily composite TIT of that respective pixel? I guess an additional sentence clarifying the procedure would help.
P10, L24: "areal extent of each pixel" Would it be fair to assume that this is 4 km^2?
P10, L25: "extrapolated to daily rates" One could also write instead: "multiplied by 86400 to obtain IP / day."
P11, Line 3: "stable cloud cover": In line 5 you write "persistent" instead of stable. I suggest to use "persistent" throughout.
P11, Lines 6-9 + Figure 4: While these 3 lines are a step into the correct direction I would suggest to take into account two more important aspects here:
i) Starting early in the freezing season might imply that the definition of a polynya is simply not yet fulfilled. Hence what is termed "POLA" is not POLA because it is not related to a polynya by definition. A look at Figure 4, November, reveals this clearly: In the eastern Kara Sea TIT frequencies are low because the thick ice has arrived. Polynyas at Islands and along the coast can be delineated. Towards the southwest TIT frequency first increaes and then decreases again towards the Kara Strait. The fact that TIT frequency is low close to the Kara gate is due to the fact that sea ice is still absent there (on average) in November and that the Kara Strait is presumably still open. Kern et al. (2005, Geophys. Res. Lett.) stated:
"In each season the PSSM (e.g. polynya area retrieval) analysis starts once the entire Kara Sea is ice covered or an ice bridge has formed in the Kara Strait so that the remaining open water area inside the Kara Sea can be regarded as a polynya. The such defined starting date varies from mid-November to the end of December."
Because of this the results of Kern et al. (2005) and Kern (2008) are based on the period January to April - to ensure that what is targeted in the conclusions and interpretations is in fact associated with a polynya.
Please note also that the results of Kern (2008) are based on coarse resolution (25 km and 12.5 km, Backus-Gilbert interpolated to 12.5 km and 5 km) SSM/I data with the respective limitations detailed already by the authors.
ii) The same applies to the IP. If one would choose 2016 (we are at the beginning of November right now) more or less the entire Kara Sea is still ice free. The same would apply (currently) to regions WNZ, STO, BSH, CHU, half of ESS, and NOW. The entire ice production happening in the Kara Sea would be included into an estimate of the ice production which according to your paper is associated with ice production in a polynya. What one would in fact include is open ocean ice production. Yes, I agree that the advance of the ice egde can be rather quick and the time between still open water and a closed sea ice cover of thickness > 0.2 m can be a few days only and the associated IP will be like noise. I tend to say though that freeze-up of region KAR and WNZ takes longer.
One would also assume that the entire ice production in these still large open water areas is due to thermodynamics. This is certainly a fair assumption for TIT < 0.2 m. However, one cannot consider effects such as rafting (due to wind but also due to tidal forcing) and one cannot consider the fact that a substantial part of the sea ice might be of type pancake ice which has a substantial dynamic formation component.
I guess the same applies to region WNZ where particularly the southern part might not be ice covered at all during November through March (at least when I browse through sea-ice concentration maps of the period 2002/03 through 2014/15 I'd say that in half of the winters region WNZ was ice free most of the time). For the region SVA only the southeastern part is influenced by MIZ activities. The Whaler's Bay polynya north of Svalbard, which is known as a sensible heat polynya is included into region SVA, however.
I suggest to, in the context of Figure 4 and its interpretation, A) describe how the TIT frequency is computed (does this include open water cases?) and B) stress that areas with low TIT frequency can be both sea ice or open water.
I am asking for A) because I am quite surprised to see the fracture event occurred in Feb 2013 in the Beaufort Sea (see Beitsch et al., 2014) to pop up in the TIT frequency map of FEB while the fact that large parts of the Kara Sea where essentially ice free in February 2013 seems not to have added to the TIT frequency. Therefore I am curious to learn from which measure the TIT frequency is computed (all native clear-sky IST cases - or all cases where a TIT value is present no matter whether it is native or interpolated) and what is the basis to make it a relative measure.
P11, Lines 21/22: Please stress that the average TIT is computed only from pixels with TIT < 0.2 m and not from the fraction (COV4) of the open ocean area. This is important to avoid mis-interpretation of your results. Perhaps you could put the TIT values into a separate table together with the POLA?
Page 13, Line 8: Please check usage of "compliment". Perhaps "complement" would fit better?
Page 13, Line 10: "this" --> "these"
Page 16, Line 11 to Page 17, Line 9: I am glad the authors do mention here the fact that freeze-up has an impact on both POLA and IP particularly during Nov./Dec. To my opinion it is not only the freeze-up which has an impact but also the fact that some of the regions simply remain ice free has an impact. I'd tend to say that the decreasing trend of IP in region WNZ, for example, is simply a cause of a general decrease of the sea-ice cover in that region. In Line 6 on page 17, I suggest to refer to region SVA instead of region Storfjorden polynya.
Page 17, Line 7: The fact that region NOW stands out that much is again caused by a substantial fraction of IP during freeze-up. This is indicated partly by the high IP values in the fjords which become covered with fast ice in December/January (see Preusser et al., 2015, Remote Sensing, 7).
Page 19, line 11/12: Instead of "the net effect ... here" the authors could write: "would require a theoretical study where an artifical polynya is investigated using several different spatial resolutions." as this would point towards a future direction of research.
Page 20, Lines 3-4 versus Line 7: I find that the notions of "suggests a southward shift of the fast-ice edge" and "shape and location of the fast-ice edge did not vary significantly" contradict each other and could lead to a misunderstanding. Perhaps the authors could reformulate this part. A suggestion would be: "... and (2) we observe opposing negative / positive IP-trends along the coasts of the Laptev and Kara Seas which could be due to changes in fast-ice extent. Decreasing ..."
Page 20, Lines 11-17: This is a quite global statement which could be strengthened. I suggest to i) refer to when in particular 2m-air temperatures have been increasing (which season?), to ii) stress that delayed freeze-up and longer open water seasons may easily lead to increased POLA and IP which could counterbalance smaller IP caused by warmer air temperatures, and to iii) be more specific how and why a downward trend in sea-ice volume can have an influence on POLA and IP.
Page 22, Line 11: "it has to noted" --> "it has to be noted"
Page 24, Figure 12: The annotation in the upper left corner of the image and the y-axis annotation says: "Ice Export Area" while the caption speaks of "Ice Area Export". Which is correct?
Page 26, bullet (1): You could add that with this spatial resolution you can go much closer to the coast and land spill over effects are efficiently mitigated.
Bullet (2): Given the concerns formulated above I suggest the authors stress that the IP mentioned here is not necessarily from ice production in a polynya but includes open ocean ice production.
Further in this bullet you could formulate more concisely instead of "Compared to ...": "Our estimate of the average total ice production exceeds that of Iwamoto et al. (2014) by about 50%. We note that differences ... "