Detailed maps of the sea ice coverage for navigational purposes are today generated by the national ice services to a very high degree based on manual interpretation and processing of satellite images. It is mainly radar images that are used for this purpose, i.e. the high resolution synthetic aperture radar (SAR) images. This is a time consuming process for expert operators. In Denmark for the Greenland waters this is done by the Danish Meteorological Institute. Sea ice maps are mainly produced for the southern part of Greenland, where most of the ship traffic is today.
Increased activities are expected in the future in most part of the Arctic when the sea ice coverage will decrease due to the global warming. Hence, there will be an increased need for more sea ice maps covering larger areas than today and with more frequent updates. In order to provide ships with such information on a regular basis it is necessary to use automatic methods to estimate the sea ice coverage from e.g. SAR images. Some difficulties exist, however, because some ice types may be confused in the SAR images, and also in some cases open water is confused with ice. This is of course a disadvantage for the production of sea ice maps, because it makes the maps less reliable. Another type of satellite sensor exists, i.e. the microwave radiometer, which is able to discriminate more accurately between the ice types and the water. The disadvantage is, however, that the spatial resolution is much worse than the SAR images, i.e. 10-50 km.
The purpose of this project is to combine the two types of data, so that the advantages from both types are preserved, i.e. the high resolution from the SAR images and the good discrimination potential from the microwave radiometer data. Methods for estimation of sea ice coverage from both SAR images and microwave radiometer data must be reviewed, implemented and tested on SAR data from Greenland. Based on this a new method for combination of the two data types must be suggested, implemented and compared with the single sensor methods.
I samarbejde medDanmarks Meteorologiske Institut
30350 Remote Sensing. Experience with programming