the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Dec 2025
Mapping sea ice concentration using Nimbus-5 ESMR and local dynamical tie points
Abstract. As part of the European Space Agency's Climate Change Initiative the one channel US satellite microwave radiometer Nimbus-5 ESMR (N5ESMR) level 1 data have been reprocessed to estimate global sea ice concentration from 11 December 1972 to 16 May 1977. The full data set is available in the CEDA Archive: DOI: 10.5285/8978580336864f6d8282656d58771b32 at a grid resolution of 25 x 25 km² and a daily timestep (Tellefsen et al., 2025). A new methodology using locally and seasonally variable algorithm coefficients called tie points has been used to calculate the sea ice concentration in both first-year ice and multi-year ice in the Arctic and the radiometrically distinct ice types A and B in Antarctica. Validation of sea ice concentration using Arctic sea ice charts from the US National Ice Center shows an overestimation of open-water SIC of up to 20 % in some places and an underestimation of SIC in sea ice covered regions near the ice edge. Validation also shows that local dynamical tie points (LDTP) improve the mapping of sea ice concentration for different types of ice, while estimates of the extent of sea ice are identical to the previous processing of the same data in Kolbe et al. (2024). A new set of quality control (QC) filters has been developed that discards far fewer data points (57.7 % reduction) than the filters in the previous processing. The data set therefore closes significant gaps in the sea ice concentration and sea ice extent record compared to the earlier data record. Of the 1.6 billion data points recorded by the satellite, 23.0 % have been discarded. 1136 days during the 1616-day period from 1972 to 1977 are covered (at least partly), which gives estimates of the mean monthly sea ice minimum and maximum extent in the Arctic and Antarctica during this period, except for some dropouts in 1973 and 1975.
- Preprint
(4835 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2025-660', Anonymous Referee #1, 23 Dec 2025
The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2025-660/essd-2025-660-RC1-supplement.pdfCitation: https://doi.org/
10.5194/essd-2025-660-RC1 -
RC2: 'Comment on essd-2025-660', Anonymous Referee #2, 19 Jan 2026
Review of “Mapping sea ice concentration using Nimbus-5 ESMR and local dynamical tie points” by Tellefson et al.
Summary
This paper presents an improved method to derive sea ice concentrations from the Nimbus-5 ESMR sensor using local dynamical tie points and improved quality control measures. The improvements in quality control allow considerably more viable data than the previous method yielding a more complete time series. The local tie points reduce biases in concentration and provide a higher quality product.
General Comment
This is an incremental but valuable improvement to the previous ESMR sea ice concentration product. Though only a single-channel instrument and some issues with quality, ESMR is a valuable resource because it can extend the passive microwave record by about 7 years. The previous version of the product was very helpful to make the ESMR data useful for sea ice studies. This version makes substantial improvements, yielding more complete data and less biased fields. The methodology is sensible and explained clearly. It is good to see that the data and software are only available. The manuscript is acceptable with only minor revisions in response to the comments below.
Specific Comments (by line number):
13-15: This sentence is a bit unwieldy and confusing. It talks about some dropouts in 1973 and 1975 – I presume these are monthly averages. In the same sentence, it discusses minimum and maximum extents – I assume here these are the lowest annual months (Mar and Sep in the Arctic). I assume the dropouts in 1973 and 1975 refer to months in general, not min/max months, but it isn’t clear. A rewrite – maybe separating out the mean months and the mention of min/max months into two sentences.
17-18: There are many, many papers that address the importance of sea ice extent as an indicator for climate change. Listing only four is fine, but I would include “e.g.”. Also, the order of the citations seems random. Not sure what the standards are for this journal, but usually they are listed in temporal order.
19: It is not only “Long observational records”, but “Long and consistent” records that are key. Without consistency, the observational record is not going to be useful.
23: “a more recent drop” instead of “the more recent drop” – using “the” presumes a familiarity with the Antarctic record that a reader may not have.
27: “from an earlier Nimbus satellite program instrument” – since SMMR on Nimbus-7 has long been a part of the standard PM record, you can’t just say “data from the Nimbus satellite program”. Actually, I think you could rework this paragraph and the next to be a bit clearer. In the next paragraph (29-24), you discuss the PM sensors – SMMR, SSMI, SSMIS. Perhaps mention that first and then say, “Recently, experimental satellite microwave data from the Nimbus-5 and Nimbus-6 program have been used…”
57: I think many readers may not be familiar with a “Dicke microwave radiometer”. I think just “passive microwave radiometer” may be sufficient, but otherwise, “Dicke” should be explained.
58-59: It might be worth noting that other PM radiometers in the standard time series (SMMR, SSMI, SSMIS) are conically-scanning. So, their incidence angle and IFOV are constant, in contrast to EMSR.
72: The NIC charts used Nimbus-5 ESMR, so the comparison with this product are not completely independent. I think the comparison is useful because (I believe) the ESMR data used in the NIC charts were manually analyzed (maybe just looking at TBs), so the analysis methods are independent. But I think it is good to emphasize the lack of complete independence and explain why the comparison is still useful - either here or in the results section where the comparisons are made.
100-101: Why was 310 K chosen? As you note, ice won’t have a TB > 273.15 K. I can see some buffer to account for instrument noise, but 310 K seems rather arbitrary. Why not 300 K? Or even 280 K?
147-148: “where MYI SIC is underestimated”. Do you mean “where SIC is underestimated in MYI regions”? In other words, I think you mean total SIC is underestimated due to the presence of significant MYI? I guess the total SIC is underestimated because the MYI is underestimated, but I think the point here is that total SIC is biased low because the MYI TB signature is not accounted for.
149: This paragraph is discussing the new method, in contrast to the old method discussed in the previous paragraph, correct? I think a transition phrase would be helpful, “In our new method, the Tp,ice…” Also, add “value” after “Tp,ice(i,j,t)”.
Figure 3: Though the figure is pretty clear, a couple suggestions: (1) the MYI in the map seems to be a different color than the legend – the blue in the MYI legend actually looks more like blue shade of open water in the map. (2) the contrast of the FYI red with the underlying TB color scale could make it difficult for some people to distinguish between the two (the red isn’t too far off from the dark orange). I can make things out, so maybe not a big deal, but changing the color scales – either of the FYI/MYI boxes or the TB values – would be clearer. For example, you could use a viridis (or similar) scale for TB and then the red/blue for the FYI/MYI would stand out clearly. Or just change the FYI/MYI colors to be more distinct from the rainbow/jet TB scale. You could also consider colored box outlines instead of solid boxes to mark the FYI/MYI areas.
Figure 4: Are the “whiskers” the min and max values? And the box represents one standard deviation around the mean rSD. I guess it is a bit confusing because it is a plot of statistics around an SD value. And often a boxplot is used to denote median and quartiles, though I assume that is not the case here because it states “mean rSD”.
255-261, Figure 9: There is really a significant difference near the ice edge – e.g., the Odden feature. It makes sense that it is related to thin/new ice, but I’m surprised by how large it is. Since, NIC used ESMR, perhaps it is due to how the analysts assessed the TB values back in the 1970s – maybe the manual analysis effectively used a lower TB threshold? Another possibility is simply that analysts were being very conservative in where they drew the ice edge. This has been an issue in other comparisons – the analysts want to err on the side of too much ice vs. missing ice. In other words, the fact that that V1.1 shows less ice than NIC may not be a low bias by V1.1, but a high bias by NIC.
Figure 9: Another add feature I see is that that while V1.1 is low relative to NIC right near the ice edge, it appears that in several regions (e.g., Bering Sea), beyond that low bias area, there is a “rim” of red shades, indicating a high bias in V1.1. How does this happen? Presumably this is well into open water.
256: missing “)” in the list of references.
Citation: https://doi.org/10.5194/essd-2025-660-RC2
Data sets
ESA Sea Ice Climate Change Initiative (Sea_Ice_cci): Nimbus-5 ESMR Sea Ice Concentration, version 1.1. NERC EDS Centre for Environmental Data Analysis, 25 March 2025. Rasmus Tage Tonboe et al. https://dx.doi.org/10.5285/8978580336864f6d8282656d58771b32
Model code and software
NIMBUS 5 ESMR processing code Emil Haaber Tellefsen https://github.com/EHTellefsen/N5ESMR-SIC-processing
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 243 | 89 | 20 | 352 | 26 | 19 |
- HTML: 243
- PDF: 89
- XML: 20
- Total: 352
- BibTeX: 26
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1