SMOS thin sea ice thickness¶

SMOS L3 Sea Ice Thickness (AWI/ESA, v3.6) and University of Bremen SMOS and SMOS/SMAP Thin Sea Ice Thickness

Available

At a glance¶

Note on period: The AWI v3.6 Arctic record starts on 2010-10-15, with 2010–2024 fully reprocessed in v3.6. Southern Hemisphere data also go back to 2010 (v3.3 operational since April 2024, v3.6 operational from April 2026); commissioning-phase data from January to May 2010 are available on request only. Northern Hemisphere fields exist only from 15 October to 15 April and Southern Hemisphere fields from 15 April to 15 October. The University of Bremen products are retrieved from October to April in the Arctic and from March to September in the Antarctic.

Overview¶

Daily thickness of thin sea ice derived from ESA SMOS L-band (1.4 GHz) brightness temperatures, which are sensitive to the thickness of thin ice (the AWI product description says thickness information can be derived up to about 1.5 m). Radar altimeters such as CryoSat-2 have large relative errors over thin ice with small freeboard, so SMOS complements them over newly forming ice, ice edges and polynyas. The main product is AWI's L3 (v3.6), produced under an ESA project, which is also the SMOS input to the merged CS2SMOS and CS3SMOS thickness products. The University of Bremen produces separate products with a different empirical method: a SMOS-only product and a SMOS/SMAP product that adds NASA SMAP (v300).

What changes with the successor¶

SMOS L3 SIT v3.3 → SMOS L3 SIT v3.6¶

Transition: 2025-10

AWI changed the SMOS L3 sea ice thickness processing to compute thickness per swath first, and improved the RFI filter and auxiliary weather data. The whole period 2010–2024 was reprocessed; v3.6 became the operational version with the Arctic season starting on 2025-10-15, and the Southern Hemisphere has been processed with v3.6 since April 2026.

Sources: AWI — SMOS Sea Ice Thickness product page (version history, v3.6 improvements) · AWI SMOS ReadMe-first Technical Note v4.0 (2025-10-15) · AWI SMOS PDD v5.0 (2025-04-24) · AWI notice — 2025/26 season start and switch to v3.6 (2025-10-17)

Available variables¶

Variables from the AWI v3.6 product description document (PDD v5.0, Table 6). Variable names in the University of Bremen SMOS and SMOS/SMAP files are not given on the official pages, so they are not listed.

instantaneous mean accumulated maximum minimum static — instantaneous: value at the time stamp · mean/accumulated/maximum/minimum: over the indicated interval · static: does not change with time

Thickness (AWI v3.6 daily files)¶

Brightness temperature and auxiliary quantities¶

Data counts and quality indicators¶

The full variable list is in the official documentation.

How to get the data¶

• ESA SMOS dissemination server — L3_SIT_Open collection — HTTPS / FTPS (ftps://smos-diss.eo.esa.int/), with monthly index files
  • Download requires a free ESA EO Sign In account. When checked on 2026-10-13, monthly indexes covered 2010-04 to 2026-10.
• AWI FTP — sea ice product root — Anonymous FTP No login needed
  • AWI states that access is anonymous. The v3.3 quicklook path on the AWI wiki (smos/v3.3/nh/) suggests a smos// layout, but the v3.6 folder path is not confirmed on the official pages, so check the FTP listing.
• University of Bremen data server — SMOS and SMOS/SMAP thin ice thickness — HTTPS / anonymous FTP (data.seaice.uni-bremen.de, user anonymous) No login needed
  • The products are in the smos/ (SMOS only) and smos_smap/ (SMOS/SMAP) folders. Subfolder layout and file format are not documented on the official pages, so check them directly.
• ESA Earth Online — SMOS L3 Sea Ice Thickness catalogue — Web information (DOI, product description, ATBD, read-me note)

Example code¶

Opens an AWI v3.6 Northern Hemisphere daily file (SMOS_Icethickness_v3.6_north_.nc), applies the checks recommended in the read-me note (uncertainty above 1 m, saturation ratio above 95%), and looks at thickness in the Sea of Okhotsk.

import glob
import xarray as xr

f = sorted(glob.glob("SMOS_Icethickness_v3.6_north_*.nc"))[-1]
ds = xr.open_dataset(f).squeeze()

sit = ds["sea_ice_thickness"]                       # m, daily mean
ok = (ds["ice_thickness_uncertainty"] <= 1.0) & (ds["saturation_ratio"] <= 95) & (ds["land"] == 0)
sit = sit.where(ok)

# Sea of Okhotsk (roughly 44–60°N, 135–160°E) — latitude/longitude are 2-D variables
box = (ds["latitude"] > 44) & (ds["latitude"] < 60) & (ds["longitude"] > 135) & (ds["longitude"] < 160)
print(str(ds["time"].values), "mean thin-ice thickness (m):", float(sit.where(box).mean()))
print("cells with RFI ratio above 50%:", int((ds["RFI_ratio"].where(box) > 50).sum()))

Practical tips¶

• CS2SMOS (../cs2smos/) and CS3SMOS (../cs3smos/) in the same tab merge this SMOS L3 thickness with CryoSat-2 (and Sentinel-3) altimeter thickness by optimal interpolation into weekly (7-day) L4 fields. Use the merged products for the full thickness range in one map, and this daily L3 for day-to-day changes during freeze-up or for thin ice itself. CS2SMOS v206 used SMOS v3.3 as input, and CS3SMOS v300 uses v3.6.
• CryoSat-2-only thickness (../cryosat2-sit/) has large relative errors over thin ice (below roughly 1 m), whereas SMOS saturates over thick ice. The AWI product description says thickness information can be derived up to about 1.5 m, but the maximum retrievable thickness depends on ice salinity and temperature, so read values with saturation_ratio above 95% as "at least this thick".
• The retrieval assumes 100% ice concentration, so cells containing open water (ice edges, leads) are biased thin, and this effect is not included in the uncertainty. In seasonal ice zones with lower concentration such as the Sea of Okhotsk, interpret the thickness together with a concentration product (e.g. ../amsr2-asi/, ../osisaf-sic/).
• Radio-frequency interference (RFI) is still a problem in the Arctic, especially along the Russian coast (Barents, Kara and Laptev Seas) since 2023. Use RFI_ratio and nPair to find poorly sampled cells. RFI is negligible in the Antarctic.
• Because v3.6 uses an ERA5 climatology instead of daily weather, it is robust to outages of auxiliary data servers but cannot capture inter-annual variability in air temperature and wind; the average effect is about 5 cm. Do not mix results from v3.3, and even though the ESA catalogue citation still reads "Version 3.3", state the version you actually used (v3.6 in the file name) in publications.
• Antarctic thickness is more uncertain because snow is handled with Arctic parameterizations and Southern Hemisphere bulk salinity has not been fully evaluated. Iceberg A-23A in the Weddell Sea (about 75.3°S, 37.3°W) appears as anomalously thin ice, which is not yet corrected.
• The University of Bremen products use different algorithms from AWI (the empirical method of Huntemann et al. 2014, and the SMAP-combined method of Paţilea et al. 2019). Bremen recommends the RFI-filtered SMOS version or the SMOS/SMAP product over the raw SMOS one, and notes that with about 40 km resolution only larger areas of thin ice are retrieved correctly. Do not mix values from the two institutes in one series.

Notices and change history¶

References for your paper¶

Dataset citation¶

ESA SMOS Level 3 Sea Ice Thickness (ESA catalogue DOI) — DOI 10.57780/sm1-5ebe10b

@misc{esa_smos_l3_sit,
  author    = {{European Space Agency}},
  title     = {{SMOS} Level 3 Sea Ice Thickness, Version 3.3},
  year      = {2023},
  publisher = {European Space Agency},
  doi       = {10.57780/sm1-5ebe10b}
}

Reference papers¶

AWI SMOS thin sea ice thickness algorithm (requested by AWI) — DOI 10.5194/tc-8-997-2014

@article{tiankunze2014smos,
  author  = {Tian-Kunze, Xiangshan and Kaleschke, Lars and Maa{\ss}, Nina and M{\"a}kynen, Marko and Serra, Nuno and Drusch, Matthias and Krumpen, Thomas},
  title   = {{SMOS}-derived thin sea ice thickness: algorithm baseline, product specifications and initial verification},
  journal = {The Cryosphere},
  volume  = {8},
  pages   = {997--1018},
  year    = {2014},
  doi     = {10.5194/tc-8-997-2014}
}

Operational application and validation in the Barents Sea (requested by AWI) — DOI 10.1016/j.rse.2016.03.009

Antarctic thin sea ice thickness (for Southern Hemisphere data, requested by AWI) — DOI 10.5194/essd-16-3149-2024

University of Bremen SMOS method — DOI 10.5194/tc-8-439-2014

University of Bremen SMOS/SMAP method — DOI 10.5194/tc-13-675-2019

Acknowledgements¶

AWI asks users to cite the three papers above (Tian-Kunze et al. 2014, Kaleschke et al. 2016, and Kaleschke et al. 2024 for Antarctic data) and to acknowledge that the production of SMOS sea ice thickness data was funded by the ESA project SMOS & CryoSat-2 Sea Ice Data Product Processing and Dissemination Service and that data from DATE to DATE were obtained from AWI. Copy the exact wording from the ReadMe-first Technical Note. For the University of Bremen products, cite Huntemann et al. (2014) or Paţilea et al. (2019).

Information on the datasets used in this study was compiled with the help of the Korean Climate Data Navigator (KCDN) (Sim, 2026; https://doi.org/10.5281/zenodo.23300923).

Links¶

• AWI — SMOS Sea Ice Thickness product page (version history, documents)
• AWI CS2SMOS-PDS wiki (operational news)
• ESA — SMOS L3 Sea Ice Thickness catalogue
• University of Bremen — Thin Ice Thickness (SMOS, SMOS/SMAP)
• This site — CS3SMOS merged sea ice thickness (SMOS v3.6 input)
• This site — CS2SMOS merged sea ice thickness (SMOS v3.3 input, updates ended)
• This site — CryoSat-2 sea ice thickness
• This site — University of Bremen ASI AMSR2 sea ice concentration