Stein Sandven

Multisensor approach to automated classification of sea ice image data

A multisensor data fusion algorithm based on a multilayer neural network is presented for sea ice classification in the winter period. The algorithm uses European Remote Sensing (ERS), RADARSAT synthetic aperture radar (SAR), and low-resolution television camera images and image texture features. Based on a set of in situ observations made at the Kara Sea, a neural network is trained, and its structure is optimized using a pruning method. The performance of the algorithm with different combinations of input features (sensors) is assessed and compared with the performance of a linear discriminant analysis (LDA)-based algorithm. We show that for both algorithms a substantial improvement can be gained by fusion of the three different types of data (91.2% for the neural network) as compared with single-source ERS (66.0%) and RADARSAT (70.7%) SAR image classification. Incorporation of texture increases classification accuracy. This positive effect of texture becomes weaker with increasing number of sensors (from 8.4 to 6.4 percent points for the use of two and three sensors, respectively). In view of the short training time and smaller number of adjustable parameters, this result suggests that semiparametric classification methods can be considered as a good alternative to the neural networks and traditional parametric statistical classifiers applied for the sea ice classification.

Arctic sea ice and Eurasian climate: A review

The Arctic plays a fundamental role in the climate system and has shown significant climate change in recent decades, including the Arctic warming and decline of Arctic sea-ice extent and thickness. In contrast to the Arctic warming and reduction of Arctic sea ice, Europe, East Asia and North America have experienced anomalously cold conditions, with record snowfall during recent years. In this paper, we review current understanding of the sea-ice impacts on the Eurasian climate. Paleo, observational and modelling studies are covered to summarize several major themes, including: the variability of Arctic sea ice and its controls; the likely causes and apparent impacts of the Arctic sea-ice decline during the satellite era, as well as past and projected future impacts and trends; the links and feedback mechanisms between the Arctic sea ice and the Arctic Oscillation/North Atlantic Oscillation, the recent Eurasian cooling, winter atmospheric circulation, summer precipitation in East Asia, spring snowfall over Eurasia, East Asian winter monsoon, and midlatitude extreme weather; and the remote climate response (e.g., atmospheric circulation, air temperature) to changes in Arctic sea ice. We conclude with a brief summary and suggestions for future research.

Satellite earth observation in operational oceanography

The role and contribution of satellite data in operational oceanography is reviewed, with emphasis on northern European seas. The possibility to observe various ocean parameters and processes by existing satellite sensors, such as optical instruments, infrared radiometers, passive microwave radiometers, and active microwave systems (altimeter, scatterometer, SAR) is discussed. The basic parameters are: sea-surface temperature observed by infrared radiometers, ocean colour by spectrometers, sea-surface elevation by altimeters, and surface roughness by active and passive microwave systems, which can be used to derive surface wind and waves. A number of ocean processes can be derived from synoptic mapping of the basic parameters of larger sea areas, such as current patterns, fronts, eddies, water mass distribution, and various water quality parameters (chlorophyll, surface slicks, suspended sediments). The suitability of existing satellite data to fulfil the operational requirements for temporal and spatial coverage, data delivery in near-real-time, and long-term access to data is discussed in light of the fact that optical/infrared data in northern Europe are severely hampered by frequent cloud cover, while microwave techniques can provide useful data independent of weather and light conditions. Finally, the use of data assimilation in oceanographic models is briefly summarised, indicating that this technique is under development and will soon be adopted in operational oceanography.

Eddy-Related Winter Convection in the Boreas Basin

Abstract In March 1989 a weakly stratified chimney containing Deep Water with temperature of - 0.41°C and salinity of 34.91 ppt (parts per thousand) at the surface was observed outside the ice edge in the northern Greenland Sea. The chimney was about 20 km in diameter and had fairly uniform potential density (sigma-theta = 28.056) from the surface to at least 500 m. Below this depth the potential density only increased to 28.077 near the bottom of 2800 m. Surface cooling of only a few tenths centigrades is sufficient to initiate vertical convection in the chimney. We propose that the mechanism responsible for exposing deep water masses to the surface in this case is a cyclonic eddy characterized by strong doming of the isopycnals. This mechanism can occur in areas where the layer of surface and intermediate water masses, which usually reside in the upper few hundred meters in winter, is thin and weakly stratified. At the surface the water is cooled by the heat flux due to air temperatures of about -20°C and 10 - 15 ms −1 . northerly winds which is typical for winter conditions in the northern Greenland Sea. Surface cooling can then drive the convection to several hundred meters in order of 10 days, the characteristic life-time of mesoscale eddies in the area.

Barents Sea seasonal ice zone features and processes from ERS 1 synthetic aperture radar: Seasonal Ice Zone Experiment 1992

Sea ice features and processes in the Barents Sea were studied during the Seasonal Ice Zone Experiment 1992 (SIZEX92), a dedicated ERS 1 satellite synthetic aperture radar (SAR) field campaign carried out in March 1992. SIZEX92 was based around the research vessels Polarsyssel and Hakon Mosby. In situ oceanographic, meteorological, and ice measurements were made, coordinated with low-altitude aerial observations and ERS 1 SAR data acquired in near real time. Fifty-eight low-and full-resolution SAR scenes were obtained during SIZEX92, which provided a validation data set for ERS 1 SAR backscatter under winter conditions in the Barents Sea. Analysis of SIZEX92 data has provided geophysical insights, including a better understanding of Barents Sea ice edge freezing processes and ice edge development in response to atmospheric forcings. In particular, areas of new ice formation were found to be related to bathymetric features through their influence on the circulation of Arctic and Atlantic water masses. ERS 1 SAR image sequences revealed rapid, mesoscale variations in new ice areas and the ice edge in response to wind conditions. In addition to geophysical insight, SIZEX92 demonstrated some of the technical capabilities and limitations of ERS 1 SAR to identify new ice areas, the ice edge, ice floes, and ice types. Mapping new ice areas using satellite SAR data may be considered the most promising new application. The limitations of ERS-1 SAR-derived ice classification and ice motion estimates in the marginal ice zone are identified.

Cover ERS-1/2 SAR monitoring of dangerous ice phenomena along the western part of Northern Sea Route

The Northern Sea Route ( NSR) as a part of the Arctic Ocean is very important for sea transportation to the Siberean coastal and river settlements, as well as for future transportation between Europe and the Paci® c Ocean countries. Recent opening of the gigantic oil and gas deposits on the Siberean shelf will require the build-up of a specialeet for eA cient marine oil and gas operations in this area. However, round-the-year navigation as well other marine operations (® shery, mining, oil and gas reconnaissance, etc. ) at the diA erent parts of the Arctic Ocean is a very complicated problem and an optimal choiceof the concrete sea route directions depends on numerous environmental factors. Environmental conditions which determine the high-latitude navigation in the North Pole area depend on seasonal and mean-annual distribution of water masses in the Arctic Ocean. An important factor which inuences marine operations in the Arctic Ocean is the presence of round-the-year ice. This is a complicated regional and global-scale process which depends on the location and properties of basic oceanic massif ice regime, on patterns of seasonal and mean-annual distribution of the basic massif s spurs, on the behaviour of the local ice massifs which in each area of the Arctic Ocean have their own characteristic features of the seasonal and mean-annual variability. The location and drift of the basic oceanic ice massifs spurs are variable para- meters. These spurs consist of the residual ® rst-year ice, and the second-year and multi-year pack iceoes. The thickness of this ice can be more than 3m and its invasion in the NSR area creates an especially dangerous situation for marine operations. Control of this phenomenon by all-weather remote sensing instruments is therefore extremely desirable. Another signi® cant factor which creates the favourable inuence for the ice navigation in the Arctic Ocean is the round-the-year presence of the vast recurring

Mesoscale eddies and chimneys in the marginal ice zone

Abstract A review of observation and modeling of mesoscale eddies and chimneys in the marginal ice zone is presented. Deep water formation, which is important for the global climate, occurs only in a few locations of the world ocean mainly at high latitudes. Some of these locations are associated with eddies in or near the marginal ice zone of the Arctic as well as the Antarctic. Deep water formation is not a well understood process, investigations so far indicate that it can occur intermittently in winter time in narrow chimneys of very low stability which have a horizontal extent of order 10 km and are located in the upper few hundred meters of the water column.

Satellite radar ice monitoring for ice navigation of a tanker convoy in the Kara Sea

The international EU project Arctic Demonstration and Exploratory Voyage (ARCDEV) aimed to investigate the conditions for navigating a tanker vessel from Europe through the Kara Sea to the northwestern Siberian gas and oil fields during ice covered winter conditions in April-May 1998. This winter faced the most difficult ice conditions for 30 years. In order to perform such ship navigation operations, an escorting icebreaker support as well as detailed information on the ice conditions along the possible routes are required. The ARCDEV convoy was supported by near real-time ice information derived from both ERS-2 and RADARSAT ScanSAR synthetic aperture radar (SAR). It is demonstrated how satellite based microwave radar technology practically can support ice navigation operations to improve the safety and efficiency of the navigation.

The Fram Strait acoustic tomography system

The deep and wide Fram Strait between Greenland and Spitzbergen is the main influx and efflux gate to the Arctic Basin. Although major resources are invested in measurements of current and temperature here (http://asof.npoar.no), the flux estimates still have significant deficiencies and errors. Our objective is to build, test, validate and use an innovated integrated observing and modeling system, including acoustic tomography, for improved monitoring of volume, heat and freshwater transports in the Fram Strait. As part of the DAMOCLES‐IP, (=Developing Arctic Modeling and Observing Capabilities for Long‐term Environmental Studies ‐ Integrated Project) a first step acoustic tomography system is to be installed in the Fram Strait between East Greenland and West Spitzbergen in August 2008. The first step tomography system consists of one acoustic source near the Svalbard shelf and one receiver array in the middle of the Fram Strait. An extended acoustic system serving both tomography and glider navigation i...

Identification of oil spills based on ratio of alternating polarization images from ENVISAT

We propose here a method to identify surface film in SAR images using the Alternating Polarization ratio images from ENVISAT. This ratio is lower in polluted areas than in non-polluted areas due to the difference in relative contributions of the non-Bragg scattering to the total radar signal.