Multiyear Ice

MYI concentration map 10 Dec, 2017
MYI concentration map 10 Dec, 2017

Sea ice in the Arctic can be classified into several types, the main ones being

  • young ice (YI): thin (up to 30 cm thick) new ice; includes a few sub-types; can be smooth or rough
  • first-year ice (FYI): formed during one cold season; thickness above 30 cm,; surface can be levelled, rough or with ridges
  • multiyear ice (MYI): ice that has survived at least one melt season;topographic features generally smoother than FYI

The physical properties of sea ice differ significantly for the different ice types. Therefore. knowledge of the sea ice type is important for a number of activities incuding marine navigation and modelling of the ice-ocean-atmosphere system, and also for remote sensing of other quantities related to sea ice such as the depth of the snow layer on top of it.  With the recent accelerated decrease of MYI in the Artic, mapping this ice type on adaily basis has become important for many applications

The MYI data on this  site are from a new satellite-based retrieval of sea ice type in the Arctic which can in principle distinguish YI, FYI and MYI.

The retrieval method uses data from active and passive microwave instruments (radar scatterometer and radiometer, respectively), but then in addition applies several correction schemes to account for the effect of melt-refreeze processes, snow metamorphosis and sea ice drift on the sea ice type retrieval. Both air temperature and sea ice drift data are used in the corrections.

As the retrieval does not work during the melt season, the data record usually spans October to May. Even during the winter season, the reitrieval can be problematic - see details here

 

 

Retrieval Algorithm and Data Sources

Retrieval algorithm

The retrieval algorithm has two main steps:

  1. A constrained optimisation technique that uses different sets of microwave satellite data  and the probability distributions of radiometric signatures of different ice types whose concentration is to be retrieved. This is the Environment Canada Ice Concentration Extractor (ECICE) [Shokr et al. 2008, Shokr and Agnew 2013]. The input data used here are microwave radiometer data at 18 and 37 GHz (horizontal and vertical polarisation), and microwave scatterometer data at 5.3 GHz
  2. Two correction schemes for the output from ECICE  in order to  correct for anomalous radiometric backscatter observations. Such anomalies are caused by snow wetness and metamorphism. This happens when air temperature rises to approach the melting point. Air temperature is used in one correction scheme and ice drift is used in the second [Ye et al. 2016a, 2016b].

 Main input

The passive microwave (i.e., radiometer) data are from either AMSR-E on the Aqua satellite by NASA (until 2011) or from AMSR2 on the GCOM-W1 satellite by JAXA (since 2012).

The active microwave (i.e., scatterometer) data are from ASCAT on the Metop-A satellite. Until 2015, ASCAT data regridded to the common 12.5 km North polar stereographic grid ("NSIDC grid") by Ifremer/CERSAT were taken (ASCAT-A 12.5km Arctic Sea-Ice Backscatter Maps) after that , near-real time ASCAT data have been regridded by ourselves.

Data for the correction schemes

For the correction scheme,  surface temperature data and sea ice drift data are needed.

The surface temperature data are meteorological reanalysis data from the ECMWF (ECMWF), namely ERA-Interim data - for each day, the daily maximum of the 2m air temperature is taken.

The sea ice drift data are either to the to following sources

  • National Snow and Ice Data Center, NSIDC: Polar Pathfinder Daily 25 km EASE-Grid Sea Ice Motion Vectors, Version 3 [Tschudi et al. 2016, doi:10.5067/O57VAIT2AYYY], based on  active and passive microave and visible/infrared data as well as buoy data.
  • Ocean and Sea Ice Satellite Application Facility, OSISAF: Low Resolution Sea Ice Drift product (OSI-405), based on active and passive microwave data

Retrieval Details

Because of different availability of the various input data sets, the following table is a detailed overview for which season which sources were taken, along with references to specific parameter settings for the correction schemes.

Retrieval Details
Season Radiometer Scatterometer Ice Drift Correction Parameter Set Remarks
2009/2010 AMSR-E ASCAT (Ifremer) NSIDC A
2010/2011 AMSR-E ASCAT (Ifremer) OSISAF A
2012/2013 AMSR2 ASCAT (Ifremer) OSISAF A
2015/2016 AMSR2 ASCAT (Ifremer) OSISAF A
2016/2017 AMSR2 ASCAT OSISAF A coast correction
2017/2018 AMSR2 ASCAT OSISAF A coast correction

Coast correction

Directly at a coast, the signal of the scatterometer or radiometer can contain contribution from the land ("land contamination"), resulting in erroneous high MYI concentration in pixels directly at the coast. This can lead to large errors later in the season. Therefore, during the first 10 days of the season  retrieved (end of September), a zone that is one pixel wide, along all coast is cleared of MYI unless  connected to MYI offshore. This is done before applying the correction schemes.

Correction Parameter Set

The correction schemes use a number of empiric parameters as documented in Ye et al. The specific parameters used for the retrieval of the listed data are

Set A:

Temperature correction:

  • T1 (threshold for beginning of warm episode): -2°C
  • T2 (threshold for end of warm episode): +1°C
  • DeltaCM (min change of MYI concentration to be considered): 10%
  • maximum duration of warm episode: 10 days

 

 

 

Caveats, retrieval problems

While the MYI  data retrieved from active and passive microwave data and corrected using temperature and ice drift data are considerably more realistic than data retrieved simply from radiometer data, some problems still remain:

  • Weather influence on the surface of the MYI (including the snow cover) can cause considerable day-to-day fluctuation of MYI concentration - a moving average or several might reduce them, but is currently not applied.
  • In the Eastern Arctic (Kara and Laptev Sea), unrealistically high MYI concentrations occur in March and April. They are probably caused by small areas of spurious MYI caused by rough young ice (possible with wet snow) in the previous autumn. 
  • We are grateful of any feedback on unexpectued features and possible problems of the data, see email links below.

How to cite

Please help maintaining this service by properly citing and acknowledging if you use the data for publications:

Ye, Y., M. Shokr, G. Heygster, and G. Spreen (2016) Improving multiyear ice concentration estimates with ice drift. Remote Sens., 8(5), 397, doi:10.3390/rs8050397.

Ye, Y., G. Heygster, and M. Shokr (2016) Improving multiyear ice concentration estimates with air temperatures. IEEE Trans. Geosci. Remote Sens., 54(5), 2602–2614, doi:10.1109/TGRS.2015.2503884.

References

Lavergne, T., Eastwood, S., Teffah, Z., Schyberg, H. and L.-A. Breivik (2010), Sea ice motion from low resolution satellite sensors: an alternative method and its validation in the Arctic. J. Geophys. Res., 115, C10032, doi:10.1029/2009JC005958

Shokr, M., A. Lambe and T. Agnew (2008), A new algorithm (ECICE) to estimate ice concentration from remote sensing observations: An application to 85-GHz passive microwave data. IEEE Trans. Geosci. Remote Sens.,  46(12), 4104-4121.

Shokr, M., T. Agnew (2013), Validation and potential applications of Environment Canada Ice Concentration Extractor (ECICE) algorithm to Arctic ice by combining AMSR-E and QuikSCAT observations, Remote Sensing of Environment 128, 315-332, doi:10.1016/j.rse.2012.10.016.

Tschudi, M., C. Fowler, J. Maslanik, J. S. Stewart, and W. Meier (2016), Polar Pathfinder Daily 25 km EASE-Grid Sea Ice Motion Vectors, Version 3. (Arctic). Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. doi:10.5067/O57VAIT2AYYY.

Ye, Y., M. Shokr, G. Heygster, and G. Spreen (2016a), Improving multiyear ice concentration estimates with ice drift. Remote Sens., 8(5), 397, doi:10.3390/rs8050397.

Ye, Y., G. Heygster, and M. Shokr (2016b,) Improving multiyear ice concentration estimates with air temperatures. IEEE Trans. Geosci. Remote Sens., 54(5), 2602–2614, doi:10.1109/TGRS.2015.2503884.

Contact

For questions regarding the data please contact Gunnar SpreenChristian Melsheimer, or Georg Heygster.
For questions related to the website and data-browser please contact Malte Gerken.

Institute of Environmental Physics, University of Bremen, Germany.