ALBERT

Description: Two algorithms have been used in a hybrid scheme in order to obtain sea iceconcentration maps at 12 km resolution from 19, 37, and 85 GHz SSM/I data.The first one is an algorithm based on the polarization difference near 90GHz and the second one is the NASA Team algorithm which uses the 19 and 37GHz SSM/I channels. Ice concentrations are calculated using the 85 GHzchannels. In addition, the lower frequency channels are used to decidewhether the data points belong to the ice-free ocean or to the ice-coveredarea. This combination of high and low frequency channels eliminatesincorrect high ice concentrations caused by weather effects over the in factice-free ocean using the rather weather independent low frequencies whileretaining high resolution over ice with the high frequency. The estimationof proper tie points for the 85 GHz algorithm was a major task. Astatistical linear regression method for reference brightness temperatureestimation was applied in order to avoid misarranged guesses of the tiepoints. This method requires independent ice concentration reference datawhich were derived from aircraft dual-polarized passive microwavemeasurements at 19 and 37 GHz and optical line scanner images. ERS-2 SARimages were used to analyze the capability of the SSM/I to resolve featuressuch as the evolution of the marginal ice zone in the Fram Strait and theStorfjorden Polynya. Two different numerical atmospheric models were used toanalyze the effect of an increased resolution of ice data from 50 to 12 kmon the model results. It was found that the representation of the ice edgezone significantly influences the modelled atmospheric boundary-layertemperatures. The temperatures obtained with the high resolution ice dataagree significantly better with aircraft observed data.

Description: By means of satellite data, maps of sea ice concentration are preparedfor observation of the global distribution of sea ice, for improvement ofice drift models and for support of the navigation in polar regions. At thepresent time, only insufficient accuracies for concentrations can beachieved by the available algorithms - in particular because anarea-covering validation to the actual situation of sea ice is missing.Essentially for this validation, we have developed different airborne linescanner systems, which are prisented in this paper.

Description: A new low-cost high-resolution line scanner has been developed at theAlfred Wegener Institut for Polar and Marine Research in Germany. ForNDVI applications the Color Line Scanner (CLS) measures the solarradiation reflected by the ground surface in the spectral ranges of500nm to 570nm (green), 580nm to 680nm (red) and 720nm to 830nm(near infrared). With the red and near infrared spectral bands the NDVIcan be calculated in order to map vegetation. The line scanner supportsa resolution of 2048 pixels per line for each spectral band. During dataacquisition 50 lines per second are stored yielding a maximum spatialresolution of better than 0.5m. With DGPS and attitude measurements(INS or Vector GPS) it is possible to geo-reference the line scanner datainto a map format with an absolute accuracy of a few metres. Severalimages can be combined to cover large areas. After the determinationof mounting errors the geo-referencing into a map is carried outautomatically without manual adjustments. The CLS was first used as animaging NDVI sensor at Airborne Research Australia (ARA) the MajorNational Research Facility at the Flinders University of South Australiato investigate spatial and temporal variations of vegetation cover.

Description: The potential of Synthetic Aperture Radar (SAR) images recorded withthe First European Remote Sensing Satellite (ERS-1) for sea iceclassification is investigated. The classification strategy employed is aregion-based segmentation with subsequent interpretation of thesegments. The formation of homogenious regions allows for acharacterization by texture parameters besides backscatter coefficients.To avoid a subjective selection and characterization of trainingsegments two airborne line scanner systems, one for visible and theother for infrared digital images of the ice surface, are utilized forvalidation measurements. After the correction of the radiometric andgeometric distortion of the line scanners the image data are classifiedin a two-dimensional feature space. After the co-location of the linescanner images with the segmented SAR data it is possible to derivethe characteristics of different ice types from the radar signatures.

Description: Sea ice concentrations, derived from passive microwave satellite sensor measurements, provide today reliable ice information on global scal. The current operational Special Sensor Microwave/ Imager (SSM/ I ) operating as part of the U. S. Defence Meteorological Satellite Program ( DMSP ), was first launched in june 1987 into a polar orbit. Due to the different emissivities of water and different ice types it is possible to obtain ice concentrations and ice fractions using suitable algorithms. During the winter time, when the overlying snow on sea ice is dry, concentrations can be retrieved to an accuracy of +/- 7%. In the summer, when surface melt is predominant, melt features appear on the ice in form of puddles or melt holes. The extent of melt features depends very much on the host ice type. In general the ice can be calssified as first year ice or old ice.The amount of water accumulated in melt features could be as high as 90% on first year ice and possibly up to 50% on old ice can lead to significant algorithm retrieval errors of ice concentrations. In order to obtain accurate melt feature concentrations a line scan camera mounted on a helicopter can provide areal coverage. Preliminary analysis of the camera data indicates that water, ice, puddles and refrozen puddles could be quantitatively separated. On 7 an 8 August 1990, based on two line scan camera flights, the ice concentration was 67% with 14% melt puddles and 40% ice concentrations with 52% melt puddles, respectively.The rest open water. The SSM/ I ice concentration for the same locations wer 46% and 35% respectively. The SSM/I ice concentrations can be compared directly to the line scan camera values. The difference in ice concentrations observed of 21 and 5% respectively can be explained due to differences in areal coverage by the two sensors. The line scan camera covered only about 6% of the gridpoint area covered by the SSM/I which consist of an area of 625 Km2 , based on these measurements, the actual SSM/I ice concentration should be 86% and 65% respectively. The method of using a line scan camera shows great promise of providing a correction to the SSM/I algorithm under summer conditions. With further experiments the camera measurements will also provide a better understanding of the melt feature life cycle on old ice.