from SSM/I brightness temperatures (19H, 19V, and 37V) using modified multiple layer perceptron neural networks. Learning data for the neural networks were extracted from ice maps derived from Okean and ERS satellite imagery to capitalize on the stability of active radar multiyear ice signatures. Evaluations of three learning algorithms and several topologies indicated that networks constructed with error back propagation learning and 3-20-1 topology produced the most consistent and physically plausible results. Operational neural networks were developed specifically with January learning data, and then used to estimate daily multiyear ice concentrations from daily-averaged SSM/I brightness temperatures during January. Monthly mean maps were produced for analysis by averaging the respective daily estimates. The 14-year series of January multiyear ice distributions revealed dense and persistent cover in the central Arctic surrounded by expansive regions of highly fluctuating interannual cover. Estimates of total multiyear ice area by the neural network were intermediate to those of other passive microwave algorithms, but annual fluctuations and trends were similar among all algorithms. When compared to Radarsat estimates of multiyear ice concentration in the Beaufort and Chukchi Seas (1997-1999), average discrepancies were small (0.9-2.5%) and spatial coherency was reasonable, indicating the neural network's Okean and ERS learning data facilitated passive microwave inversion that emulated backscatter signatures. During 1988-2001, total January multiyear ice area declined at a significant linear rate of À54.3 Â 10 3 km 2 yr À1 (À1.4% yr À1 ). The most persistent and extensive decline in multiyear ice concentration (À3.3% yr À1 ) occurred in the southern Beaufort and Chukchi Seas. In autumn 1996, a large multiyear ice recruitment of over 10 6 km 2 (mostly in the Siberian Arctic) fully replenished the previous 8-year decline in total area, but it was followed by an accelerated and compensatory decline during the subsequent 4 years. Seventy-five percent of the interannual variation in January multiyear sea ice area was explained by linear regression on two atmospheric parameters: the previous winter's (JFM) Arctic Oscillation index as a proxy to melt duration and the previous year's average sea level pressure gradient across the Fram Strait as a proxy to annual ice export. Consecutive year changes (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001) in January multiyear ice volume were significantly correlated with duration of the intervening melt season (R 2 = 0.73, À80.0 km 3 d À1 ), emphasizing a large thermodynamic influence on the Arctic's mass sea ice balance during summers with anomalous melt durations.

The polar bear inhabits mainly the Arctic sea ice, where it moves and hunts. The study of polar bear movements aimed mostly at determining the bear sea sonal habitats [1] and evaluating the size of its home ranges [2]. However, the ice where the polar bear lives is drifting, and the ice drift have a substantial influence on the trajectory of animal relocations. The sea ice motion has been earlier assessed only qualitatively, but the rate and direction of sea ice drift should be mea sured quantitatively to evaluate correctly the animal movements on ice. A technique of quantitative mea surement has been elaborated in this study.To study the polar bear movements, a collar with a built in radio transmitter of the Argos satellite system is usually mounted onto an adult female. The sea ice drift in Arctic is estimated using autonomous buoys, satellite images, and modeling. Data assimilation improves the accuracy of estimation.The object of our study was one out of three female polar bears that we have tagged on the island Alexan dra Land of the Franz Josef Land archipelago [3]. This animal moved throughout the Barents Sea, which is an area of high sea ice dynamics. The data transmitted from the satellite collar on the bear provided informa tion about the accuracy of coordinates, which were used to plot the trajectory with a standard 24 h time step of vertices, which corresponded to the animal position at 12:00 UTC. Erroneous locations were excluded manually from the trajectory. Position coor dinates were calculated using the weighted mean of locations (±12 h), accuracy class, and time interval from the noon. The resultant trajectory has been con structed using 165 locations from Several datasets of ice drift have been used from OSISAF [4], CERSAT [5], and NSIDC [6] providers. To verify visually these datasets, an independent eval uation of ice drift used the data on the concentration and brightness temperature measured with an AMSR E microwave radiometer [7].The ice drift is evaluated from the satellite data by the method of maximum cross correlation [3]. Each out of four 36 and 89 GHz channels of both polariza tions is treated independently with different sizes of the search areas and with overview of the ice edge posi tion, surface wind, and sea level pressure.To localize the female polar bear, the vectors of sea ice drift have been constructed using spatial interpola tion by the method of ordinary kriging [8] with a max imum distance of 200 km. Variogram is approximated by Bessel function yielding the least error values as compared to other functions.The OSISAF dataset estimates ice drift for a 2 day interval, CERSAT reflects ice movement for 2, 3, and 6 days according to AMSR E and for 3 and 6 days according to ASCAT, the combined ASCAT and SSM/I data for 3 and 6 days, and the combined Quik SCAT and SSM/I data for 3 and 6 days. The average daily ice drift is withdrawn from the original NSIDC set in the uniform grid.To match the data on sea ice drift and the data on the polar bear movements, the animal locations wer...

No abstract

No abstract

No abstract