Operational Service for Mapping the Baltic Sea Landfast Ice Properties
Round 1
Reviewer 1 Report
The authors provide a well written overview about a new sea-ice community service with a focus on making information about land-fast sea ice available to th broader public in an easy way. The manuscript provides an overview about recent studies in the area and about the need for further and especially readily available information to the public. In general, the manuscript is in good shape and is within the scope of the journal and should be published. However, of course there are some remarks and questions that remain and that this reviewer would like to see adressed before publication.
Introduction
General: remove first two paragraphs completely or at least start with third paragraph (L45) as it gives a proper intro to the area and work. First two paragrpahs read like a repetition of the abstract just in a less concise way.
L49: ‘length’ sounds inappropriate, suggest to use ‘duration’
L57: Stick to SI Units: change 120cm to 1.2m
L64: Seems like there is an’the’ missing regradless of THE amount
L65/66: To avoid repeating maybe change to: Most important navigational parameters comprise…
L75: Could you highlight the Quark and Aaland Sea in Figure 1? At least this reviewer is lacking the specific geographic knowledge the authors seem to expect
L116: so areas with backscatter returns in a certain range get assigned the thickness estimated from the ice chart which in turn relies on measurements? This part is unclear to me. Although the authors provide a reference I feel this should be clarified for a potential reader without having to look up additional sources
L122: MLP is not explained in the text, while I assume the authors mean multi-layer perceptron? Some additional information on the NN would also be nice: How many layers does it feature with how many neurons per layer? Is it a shallow or deep one? Activation functions used and so on?
L132: Maybe the targeted reader group is more familiar with the unit NM (I assume a nauical mile) but I suggest to stick to SI Units everywhere and give this Information as an addition: e.g. 1852m/1.852x10³m (1 nautical mile)
L135-137: Could the authors explain this relationshipa bit more detailed?
L149: SNT and LFI ARE estimated...
Datasets and Processing
L176-180: Just to clarify, so this is an additional correction to the one mentioned in L172? Maybe say so as it is currently not really clear
L212: ‘equation’ instead of ‘question’
L235: Suggest to write: “model uses atmospheric forcing data from the ECMWF... “
L283: swap numbers around to match SIE and SIV, respectively
L300: These min/max values at a station are estimated how? Maybe I misread it, but this is observations now, correct? Are the measured at one point several times? Or are we talking about min/max of the time-series?
L310: While probably familiar to most readers, abbreviation EO is not explaine din the text I think – please double check!
BALFI
L364: Could the authors also explain in more detail (as they did for temporal cc) the use of temporal phase correlation as well as optical flow? Especially I am unfamiliar with the term optical flow and how that is computed.
Figure 5: It appears to me the font size on this figure especially inside the boxes is way to small to read on a printout?
Figure 6: I find the numbers hard to read on this Figure as well.
L407: Again I would stick with SI unitsfor consistency
Figure 7: Interestingly, the arc shaped area of thicker (50cm) sea ice does not really correlate with the ice deformation. Any suggestions? Maybe there is more in the manuscript on that later.
L427-429: While I appreciated concise and short manuscripts as I also try to keep my writings as short as possible, I also think potential readers appreciate the offer of small bits of additional information to put studies in to place without having to check into more detail into the given references. So I - and potentially other readers as well – would appreciate more details on which texture parameters were use? GLCM? Also about the random forest? Whats the number of trees used on based on which training data? This can be said in one sentence offering some beneficial additional information without going into too much of detail extending the manuscript to unreasonable length.
L460: Minus ‘of’
L471: ArcGIS is not really ‘freely’ available though…
Results
L500: ‘except’? Instead of ‘expect’ - maybe consider to put the following paragraph in here before the part about themean difference… but that is only a suggestion out of personal taste
L557: First time I see NM shown in some other unit here.
L559: Please double check abbreviation IS to be explained before? - Have to admit I was a bit dissapointed by the large offset from all I read about the model beforehand in the introduction but you explanations seem sound.
L574: Did you check this assumption by looking into any meteorological data availabl to you at FMI?
L596: Could this be related to the grouping of pixels of BALFI data? Do single pixels match better?
Discussion
L651: As a general suggestion and personal taste I always encourga e authors to reintroduce abbreviations in a new chapter for ease and convenience of readability
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Operational Service for Mapping the Baltic Sea Landfast Ice Properties by Mäkynen et al., describes an operational service system for mapping the Baltic Sea landfast ice characteristics using a combination of SAR data, a thermodynamic sea ice model and Copernicus Marine Environment Monitoring Service products. The paper makes a solid contribution to operational ice services around the world and provides a solid reference for the production chain description of how the BALFI products are constructed. My main criticisms of the manuscript are in its organization within the Introduction, Abstract and Datasets and Processing sections and the possible incorrect use of ice chart and SAR datasets (incorrect matching of years). Other than these organization issues, I recommend that the manuscript be published subject to major revisions (listed below).
Major Concerns
One could make a strong case that the Introduction should begin with the 3rd paragraph (L45) which describes the Baltic Sea icescape which then flows into an introduction of the CMEMS products at starting at L101 and then finally evolves into your first two paragraphs (L1-44) which link the BALFI products to the aforementioned descriptions. The way the Introduction is currently laid out is confusing and it seems like the science questions are not driving the objectives of the manuscript – which seems to me is really ‘to test the accuracy and quality of the BALFI products with available in-situ and ice chart data for the ice season 2019-2020’. Finally, given the somewhat circular use of CMEMS products towards the development of BALFI products, each of which use SAR data, I suggest the authors consider introducing Figure 5 (production chain) earlier if possible … or at least refer to this processing chain Figure 5 in the Introduction somewhere. In short, the Introduction section is largely confusing to read through until the reader is presented Figure 5.
Secondly, I believe the authors may have a mix-up between Figure 4 (ice chart) and Figures 6,7,8 etc (produced from SAR image analysis). I was confused as to why the ice chart shows lots of thin ice types between 63 and 65oN latitude … but Figures 6,7,8, etc does not (only open water) – or am I missing something here? Figure 4 appears as though it may be the March 5 ice extent in 2019 (see top right corner of figure) whereas the caption reads March 5, 2020. Given the much smaller extent of landfast ice in 2020 as compared to 2019 (apparently; Figure 4 versus the other figures) I would suggest that analysis be re-done with the 2019 ice chart and SAR data (ie. colder winter in 2018-19 compared to 2019-20), showing much more ice and many more thin ice types. As a result of this apparent discrepancy, I do not have confidence in the results of Figure 9 and the comparison of BALFI product estimates and ice and snow station data thereafter. Perhaps this is the reason why the results have large discrepancy between BALFI and in situ obs.
I would be happy to review these results again after clarification.
Minor Comments
There are a few sentences which do not make sense as currently stated:
- Abstract L15-16. We have developed operational products on these LFI 15 parameters based on synthetic aperture radar (SAR) imagery and existing products and prediction models on the Baltic Sea ice properties. What do you mean by ‘developed … existing products and prediction models’?
- Introduction L42-43. “A CMEMS downstream service is targeted to provide value-added ocean products or information or derived products developed from information provided by CMEMS”. The sentence reads very awkwardly and has too many ‘ors’ and ‘informations’ in it.
L147 - In this study we present a method …
L132 – spell out nautical mile first this first time NM is introduced.
L180 – you do not need degrees symbol twice here (θ0 of 30°) … just one of them.
Figure 2 caption. I’m confused, the caption reads that the mosaic is from both 2015 and 2019/2020 ice seasons?
L378-379 It is not entirely clear what is meant by ‘thresholding a temporal second smallest value of the cross-correlation in the SAR HH- and HV-polarization mosaics’., besides just citing the Karvonen paper. While I, a supposed expert in the field, hardly, can decipher what is implied here, the average reader is not, despite the reference to the Karvonen paper. This procedure could be more fully described with another sentence or two.
L390. In addition to the radar data for SIT and SNT estimation on Landfast ice you could cite some of the snow on FYI work from the Canadians using SAR data such as Yackel, Barber, Gill and others for saline ice using SAR. Also, S. Farrell has been using ICESAT-2 for snow thickness retrieval and I. Lawrence has been using multi-frequency radar altimeter data for snow thickness (DUST). In general, Section 3.2 seems overly biased to SIT and there is minimal mention of SNT.
Section 3.3. There is a recent paper using SAR for landfast sea ice roughness from SAR that the authors may be unaware of Estimation of Level and Deformed First-Year Sea Ice Surface Roughness in the Canadian Arctic Archipelago from C-and L-Band Synthetic Aperture Radar. SM Cafarella, R Scharien, T Geldsetzer, S Howell… - Canadian Journal of Remote Sensing, 2019.
L740 …. if the required input data
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Please find attached my review.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
The paper describes an operational system named BALFI aimed at monitoring the landfast sea ice extent and thickness of the Baltic Sea. BALFI is operated by using the SAR imagery provided by Sentinel-1 and Radarsat-2 sensors, complemented by the predictions of a sea ice thermodynamic model from FMI and the CMEMS products for the Baltic Sea. A large number of users have been using the BALFI service for several purposes among which recreational activities and transportation activities of persons and goods.
BALFI outputs have been validated with available in situ data.
The paper is well written and a comprehensive description of the product is provided.
I do not have particular comments to improve the paper. Therefore, I recommend the paper to be published as it is.
Author Response
Thank you for your very positive review on our paper. We have made several changes and edits to the paper based on the comments by three other reviewers.
Round 2
Reviewer 2 Report
I wish to thank the authors for their replies to my comments. I am satisfied with the revised version and it is now ready for publication. It will make for a valuable contribution to Ice Services.
