Sea Ice Detection

by Kathrin Rösner

While icebergs, glaciers and ice sheets are formed on land, sea ice is just frozen ocean water. The amount of sea ice in the oceans increases during winter and decreases during summer while some sea ice exists all year in some regions. During some part of the year, approximately 15% of the oceans are covered by sea ice. [Source: http://nsidc.org/seaice/intro.html ]
table of contents

Why is sea ice important?

Sea ice is important because of various reasons, and the influences of sea ice aren't restricted to the polar regions in which sea ice usually occurs. In fact, sea ice has big influence on the global climate system since its bright surface reflects much sunlight back into the atmosphere, keeping the temperature low in sea ice areas. If much sea ice melts, the affected areas absorb more solar energy and have therefore higher temperatures that melt even more ice. As a result, the polar regions are extremely sensitive to global climate changes and are good to detect and measure climate changes.

Additionally, the salt that the sea ice emits influences the ocean's global water circulation. Changes in the sea ice coverage can have an impact on the ocean circulation and can even lead to global climate changes.

Of course, sea ice can cause problems for ships, for example for routes through the Northwest Passage or for oil ships that are travelling through sea ice areas. The sea ice detection as described below is used to deal with this problem. [Source: http://nsidc.org/seaice/intro.html , more info available at http://nsidc.org/seaice/environment/global_climate.html]

table of contents

Remote Sensing

Remote sensing is defined as getting information about an object without physical contact. For the purpose of detection sea ice in order to improve navigation, active remote microwave sensing is usually used. This means that a sensor on a satellite or on an aircraft emits microwave radiation (therefore "active"). The ground reflects the microwaves - depending on the kind of reflecting material, the reflected radiation has different properties. The sensor system measures this and computes an image of the surface. Usually the term "remote sensing" refers to the usage of electromagnetic radiation from space or from an aircraft. Passive remote sensing refers to measuring radiation that an object emits naturally. But since this energy is relatively low, passive remote sensing cannot be used for measuring sea ice in a detail that is neceessary for ship navigation. In contrast to that, active remote sensors observe radiation that the sensor itself generated. Therefore, this article focusses on active sensors. [Source: Remote sensing of Snow and Ice]
table of contents

Sea Ice Detection Sensors

1. Aerial photography and electro-optical systems that operate in the visible and near-infrared region have problems with cloud coverage.
   This is an important drawback because a series of temporal consecutive images of a very large area is needed for measuring and recording climate changes. Therefore, the most common sea ice detection techniques are based on microwave radiometry since they are not dependent on daylight or on a cloud-free sky. Additionally, the microwave radiometry is usable for global coverage. [Source: Remote sensing of Snow and Ice]

2. Real Aperture Radar (RAR) (on board of satellite OKEAN)
   Problem: two objects in the azimuth (= along-track) resolution will be measured as a single, larger object if they are closer to each other than the radar beam-width.
   Real aperture azimuth resolution = H * wavelength / .(length of the antenna * cos(incidence angle) )
   In order to obtain a better azimuth resolution, a shorter wavelength is needed or a longer antenna. Long antennas are in the best case expensive and in the worst case not long enough, since a needed antenna length of 4.6km is possible (calculation with nominal values: wavelength = 5cm, platform height = 800km, incidence angle = 30°). On the other hand, shorter wavelength are subject to higher attenuation due to clouds and atmospheric conditions. Therefore, the resolution is very dependent on the height, and RAR is usually used for airborne systems though there are some spaceborne systems like the satellite OKEAN.
   The SAR was developed in order to overcome these restrictions. [Source: http://earth.esa.int/applications/data_util/SARDOCS/spaceborne/Radar_Courses/Radar_Course_II/real_aperture_radar_azimuth_resolution.htm ]

3. Scatterometry (onboard of satellite QuickScat)
   Just like SAR, scatterometers are active remote sensors that can be used on airborne or spaceborne systems. But a scatterometer places less emphasis on high spatial resolution than a SAR, and focusses on high radiometric resolution. Usually, but not always, a scatterometer uses more than a single antenna in order to determine the angular dependence of the backscattering coefficient more precisely. The QuikScat is an example for a satellite that carries a scatterometer onboard. [Source: Remote sensing of Snow and Ice]

4. SAR
   SAR stands for 'Synthetic Aperture Radar'. It is an active microwave remote sensor. That means it is a remote sensor that actively emits microwave radar and obtains its images by measuring the backscatter. It can be airborne or spaceborne. Please see the next paragraph for a detailed description of its concept.

How does SAR get its pictures?

Beaufort Sea, north of Alaska. Picture taken by RADARSAT. The dark lines that are merged with the sea ice are leads that are covered with no or with thin ice. The brighter colours come from multiyear ice floe. [Source: http://nsidc.org/seaice/study/active_remote_sensing.html] (Image courtesy of NASA JPL, University of Alaska–Fairbanks.)

A SAR, no matter if it is airborne or spaceborne, emits microwave radiation. This radiation gets reflected from the ground in different ways, depending on the reflecting material. It is easy to distinguish between open water and sea ice since the microwave emissivity of sea ice differs significantly from that of open water. SAR measures the backscattered power of the reflection and creates an image like image above.
Since a longer antenna improves the resolution of the image, and long antennas are quite expensive, the SAR uses the movement of the vehicle and complex processing techniques in order to simulate a larger antenna. The SAR emits several impulses while travelling over a specific object. Due to the movement of the vehicle relative to the ground, the frequency of the reflected impulses are shifted. This is called doppler effect. During a complex calculation, the SAR processor uses reference doppler-shifted frequencies and the actual doppler-shifted reflected radar waves in order to match the reflected waves with the correct object. Thus, the SAR processor creates an image that looks like it was derived from the data of a far larger antenna than the SAR actually had. The term 'synthetic aperture' in SAR refers therefore to the distance that the vehicle moved while its radar antenna received the reflections of a single object.
[Source: http://www.asf.alaska.edu/about_sar/faq.html#saperture], more info available from this NASA website

Techniques for improving SAR

1. Polarimetry
   Radar waves are polarized, and different materials reflect just radar waves with a specific polarization, some materials even change the polarization of the incoming radar wave while reflecting it. Therefore, emitting radar waves with different polarizations lead to different images with different included information which can be combined. An example for this can be seen in the images on the right side, which shows the west coast of Newfoundland on April 1994 in HH (leftmost) and HV (middle) polarization. Both images were obtained by the Spaceborne Imaging Radar-C mission that carried a SAR on board of a space shuttle. The same polarizations can be obtained by the spaceborne RADARSAT - systems. [Source: http://www.radarsat2.info/application/ice/cs_seaice.asp]

2. Classification
   It is possible to classify data like the one from the above mentioned images. This was done in the leftmost picture by using Maximum Likelihood classification. The rightmost image is the result of a H/A/a Maximum Likelihood classificator after five iterations. [Source: http://www.radarsat2.info/application/ice/cs_seaice.asp ]

3. Interferometry.
   The basic idea is comparing two different SAR images of the same area from slightly different locations, obtaining the phase informations. The phase differences between the images are considered to be totally due to the geometrical differences. With this assumption, the exact geometric surface can be determined.
   This technique optimizes the precision of a SAR – geometrical data and bulk translation of solid surfaces are accurate up to the order of a centimetre, the topography is precise up to the order of a meter. But this technique requires hi-tech on the side of the SAR-operating system and on the side of the user who processes the SAR-data. [Source: Remote Sensing of Snow and Ice]
   See the InSar fact sheet of the U.S.Geological Survey or an Physics Today article named 'InSar, a tool for measuring Earth's surface deformation' from Matthew E. Pritchard, Department of Earth and Atmospheric Sciences at Cornell University, USA (July 2006)

Problems of SAR

1. Imaging radars like SAR are subject to a specifying viewing geometry. This geometry causes a number or geometric and radiometric interferences in areas with a considerable relief. These disturbances can be corrected if a proper digital model of the area is available.

2. SAR uses coherent radiation. This causes a characteristic noise called “speckle”. Speckle worsens the radiometric resolution. A list of literature about speckle reduction can be found at http://www.gi.alaska.edu/~rgens/teaching/literature/sar_speckle_filtering.html.

Technical Data for SAR and Scatterometer (spaceborne)

[Source: Remote Sensing of Snow and Ice]
table of contents

Evaluation – what should we use for sea ice detection?

• OKEAN RAR, ERS SAR and RADARSAT SAR:
  Both RAR and SAR images are usable for sea ice detection, but high-resolution ScanSAR images are “particularly suited”
  Source: Comparison of sea ice signatures in OKEAN and RADARSAT radar images for the northeastern Barents Sea by V.Y. Alexandrov, S. Sandven, K. Kloster, L.P. Bobylev, and L.V. Zaitsev. In: Canadian Journal of Remote Sensing, Volume 30, Number 6, December 2004 , pages 882-892. Free full issue (46 MB) and an abstract are available.

• Scatterometer
  Some sea ice edge detection algorithms for the QuikScat underestimated the ice edges. This (link) paper proposes and evaluates a new sea ice edge detection algorithm for the QuikScat, validates it with RADARSAT-1 images (amongst others) and comes to the conclusion that a reliable ice edge detection seems to be impossible by better resolutions than 2.225km.
  “Automatic detection and validity of the sea-ice edge: an application of enhanced-resolution QuikScat/SeaWinds data” by Haarpaintner, Tonboe Long and Van Woert
  National Ice Center, Washington, DC, USA; July 2004. In: IEEE Transactions on Geoscience and Remote Sensing; Volume 42, Issue 7, pages 1433- 1443

Links

• Official website for RADARSAT-2
  http://www.radarsat2.info/

• Scientific SAR user guide
  Alaska Satellite Facility, Geophysical Institute, University of Alaska Fairbanks
  http://www.asf.alaska.edu/about_sar/

• Report on ice hazards
  ICE HAZARD TEAM REPORT 2001 BR>Committee On Earth Observation Satellites Disaster Management Support Group

• University of Hamburg, Germany.
  Overview of their remote sensing research.
  Good overview of Sea Ice remote sensing and many other related topics. (english website)
  http://www.ifm.uni-hamburg.de/~wwwrs/res.html#sat_remotesensing
  

• Canada Centre for Remote Sensing
  http://ccrs.nrcan.gc.ca/
  (technical specifications of RADARSAT-1 and -2, application of different SAR systems, tutorial on radar remote sensing)

• National Snow and Ice Data Center, University of Colorado at Boulder
  http://www.nsidc.colorado.edu/
  Overview of sea ice & remote sensing: http://nsidc.org/seaice/study/active_remote_sensing.html

• National Ice Center, USA
  http://www.natice.noaa.gov/

• Sea Ice Remote Sensing Group, Goddard Space Flight Center, NASA
  Overview of their projects, including AMSR-E sea ice animations and the current data set of Satellite Microwave Radiometers (SMMR and DMSP SSM/I)
  http://polynya.gsfc.nasa.gov/seaice_projects.html

References