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.     

A microwave technique for mapping ice temperature in the Arcticseasonal sea ice zone

A technique for deriving ice temperature in the Arctic seasonal sea ice zone from passive microwave radiances has been developed. The algorithm operates on brightness temperatures derived from the Special Sensor Microwave/Imager (SSM/I) and uses ice concentration and type from a previously developed thin ice algorithm to estimate the surface emissivity. Comparisons of the microwave derived temperatures with estimates derived from infrared imagery of the Bering Strait yield a correlation coefficient of 0.93 and an RMS difference of 2.1 K when coastal and cloud contaminated pixels are removed. SSM/I temperatures were also compared with a time series of air temperature observation from Gambell on St. Lawrence Island and from Point Barrow, AK weather stations. These comparisons indicate that the relationship between the air temperature and the ice temperature depends on ice type     

Beaufort-Chukchi seas summer and fall ice margin data from Seasat:conditions with similarities to the Labrador Sea

The margin of the sea-ice pack of the Beaufort and Chukchi seas is examined using the microwave data from Seasat taken during the summer to early fall, July 4 through October 8, 1978, and the observations are compared to the analogous observations taken in LIMEX'87. The sensors used are synthetic-aperture radar (SAR), the Seasat-A scatterometer system, and the Scanning Multichannel Microwave Radiometer     

A review of sea ice and ocean modeling relevant to the Labrador andNewfoundland shelves

Numerical and analytical models of sea ice and ocean circulation are reviewed from a viewpoint of their application to the Labrador and Newfoundland shelves. Seasonal variabilities of sea ice and ocean circulation have not been reported in numerical models. The difficulty comes from the importance of both stratification and bottom topography. Sea ice should be included for seasonal cycle calculations. Individual mechanisms of variabilities with time scales ranging from a day to a month have been studied extensively using various models so that the coupling of these models will make it possible to test the predictability of sea ice and ocean circulation. The data collected during LIMEX'87 and '89 provide verification of the models     

The dielectric properties of brine in sea ice at microwave frequencies

The dc conductivity and complex dielectric constant at frequencies of 7.50, 9, 11, 30, and 40 GHz of 16 samples of sea water brine in equilibrium with sea ice with freezing temperatures ranging from -2.8degC to-25.0degC have been measured. The data is analyzed to yield parameters occurring in a Debye relaxation equation so that the dielectric constant of brine may be calculated throughout the microwave region of the electromagnetic spectrum.     

Synergy of active and passive satellite microwave data for the study of first-year sea ice in the Caspian and Aral seas

The paper discusses application of active and passive microwave data for assessment of time and space variations of first-year ice cover. The Caspian and Aral seas are chosen as main study areas. The Caspian Sea evolution is primarily climate driven, while for the Aral Sea there is a mix of anthropic and climate factors. We analyze ice cover conditions using a novel method that combines active and passive satellite measurements for ice discrimination. This method uses the synergy of simultaneous data from active (radar altimeter) and passive (radiometer) microwave instruments onboard the TOPEX/Poseidon (T/P) satellite, launched in 1992. The benefits, drawbacks, and potential of ice cover studies using the proposed method are discussed. We analyze in detail how this method is influenced by the difference in footprints of the T/P sensors and by the radiometric properties of ice and snow at different stages of ice cover evolution. In order to link the T/P-derived results to historical observations that end in the mid-1980s, long time series of passive microwave data from SMMR and SSM/I sensors have also been analyzed. Satellite time series of ice cover extent and duration of ice period have been obtained for the Caspian and Aral seas since 1978. A good agreement is obtained between historical and satellite data, with significant spatial and temporal variability of ice conditions. There is a marked decrease of both duration of ice season and ice extent during the winters 1998/1999-2001/2002. These satellite-derived time series of sea ice parameters are very valuable in view of the heterogeneous and mostly unpublished data on ice conditions over the Caspian and Aral seas since the mid-1980s.     

Field observations and model calculations of dielectric propertiesof Arctic sea ice in the microwave C-band

The complex dielectric constant of first-year and multiyear sea ice was measured during the Seasonal Ice Monitoring and Modeling (SIMMS) field experiments, conducted in the Arctic in the spring of 1992, 1993, and 1995. The dielectric constant was also computed based on an established dielectric mixing model by using different assumptions about inclusion shape. Computations were based on detailed measurements and observations of ice physical properties and crystalline structure. Comparison between measurements and model results was conducted to identify working models for first-year and multiyear ice. For first-year ice, models that employ the assumption of vertically oriented brine pockets are applicable to columnar ice and those with the assumption of randomly oriented brine pockets are applicable to frazil ice. The validity of the models are established only for ice temperatures less than -8°C. For multiyear ice, there is no need to account for air bubble shape. The coexistence of brine and air inclusions in multiyear pond ice makes it characteristically different from hummock ice. Best results for pond ice were obtained from a simple model that accounts only for volume fractions of inclusions, rather than their shape. Physical parameters that can be retrieved directly from the dielectric constant are salinity of first-year ice at temperatures below -15°C and density of multiyear hummock ice. Detailed measurements of permittivity and loss of first-year and multiyear ice are presented along with some insight into interactions between the dielectric constant and physical parameters     

Inversion of the surface properties of ice sheets from satellite microwave data

Electromagnetic models are used as the basis for a least squares inversion technique to estimate the dry snow zone surface properties of the terrestrial ice sheets from active and passive microwave satellite data. Retrieved parameters include grain size, density, layer thickness, and accumulation rate. The prime motivation is to provide information of direct value to the Cryosat altimeter mission. The derived parameters can be used to convert from elevation change to snow mass change. They can also be used to predict geophysical retracking errors in altimeter data and to estimate the resulting uncertainty in the altimeter elevation measurement. With this technique, snow accumulation rate can also be estimated using passive microwave data. These data can then be compared to historical European Remote Sensing (ERS) satellite altimeter data in order to assess the impact of interannual variability in accumulation rate on the significance of rates of elevation change. The technique is in the preliminary stages of assessment, but is demonstrated using ERS-2 altimeter data in conjunction with spatio-temporally colocated Special Sensor Microwave/Imager (SSM/I) and QuikSCAT data. It is planned to apply the technique ultimately to Cryosat.     

基于图论的人脸图像数据降维方法综述

近几年基于图论的降维方法越来越得到人们的关注,本文针对人脸识别中的核心问题即对高维数据进行降维的目的,首先介绍了有关图论的基本概念,通过总结各种人脸图像降维的方法,将这些方法统一到图嵌入框架中。然后结合线性与非线性的角度分析了各种算法的优缺点,得出了非线性图嵌入算法在挖掘人脸图像中的非线性特征以及在数据降维方面均优于传统的方法。最后针对现有的构图方式所存在的问题对今后的研究与发展方向进行了讨论。     

On the utility of SeaWinds/QuikSCAT data for the estimation of the thermodynamic state of first-year sea ice

The thermodynamic state of sea ice is important to accurately and remotely monitor in order to provide improved geophysical variable parameterizations in sea ice thermodynamic models. Operationally, monitoring the thermodynamic state of sea ice can facilitate eased ship navigation routing. SeaWinds/QuikSCAT (QuikSCAT) dual-polarization [i.e., horizontal (HH) and vertical (VV)] active microwave data are available at a sufficiently large spatial scale and high temporal resolution to provide estimates of sea ice thermodynamics. This analysis evaluated the temporal evolution of the backscatter coefficient (/spl sigma//spl deg/) and VV/HH copolarization ratio from QuikSCAT for estimating sea ice thermodynamics. QuikSCAT estimates were compared against RADARSAT-1 synthetic aperture radar (SAR) imagery and the Canadian Ice Service (CIS) prototype operational ice strength algorithm. In situ data from the Collaborative Interdisciplinary Cryospheric Experiment (C-ICE) for 2000, 2001, and 2002 were used as validation. Results indicate that the temporal evolution of /spl sigma//spl deg/ from QuikSCAT is analogous to RADARSAT-1. The QuikSCAT /spl sigma//spl deg/ temporal evolution has the ability to identify winter, snow melt, and ponding thermodynamic states. Moreover, the copolarization VV/HH ratio of QuikSCAT could provide a second estimate of the ponding state in addition to identifying the drainage state that is difficult to detect by single-polarization SAR. QuikSCAT detected thermodynamic states that were found to be in reasonable agreement to that of in situ observations at the C-ICE camp for all years. Operational implications of this analysis suggest QuikSCAT is a more effective and efficient medium for monitoring ice decay compared to RADARSAT-1 and can be utilized to provide more robust modeled ice strength thresholds.     

Surface roughness characterizations of sea ice and ice sheets: case studies with MISR data

This work is an examination of potential uses of multiangular remote sensing imagery for mapping and characterizing sea ice and ice sheet surfaces based on surface roughness properties. We use data from the Multi-angle Imaging SpectroRadiometer (MISR) to demonstrate that ice sheet and sea ice surfaces have characteristic angular signatures and that these angular signatures may be used in much the same way as spectral signatures are used in multispectral classification. Three case studies are examined: sea ice in the Beaufort Sea off the north coast of Alaska, the Jakobshavn Glacier on the western edge of the Greenland ice sheet, and a region in Antarctica south of McMurdo station containing glaciers and blue-ice areas. The MISR sea ice image appears to delineate different first-year ice types and, to some extent, the transition from first-year to multiyear ice. The MISR image shows good agreement with sea ice types that are evident in concurrent synthetic aperture radar (SAR) imagery and ice analysis charts from the National Ice Center. Over the Jakobshavn Glacier, surface roughness data from airborne laser altimeter transects correlate well with MISR-derived estimates of surface roughness. In Antarctica, ablation-related blue-ice areas, which are difficult to distinguish from bare ice exposed by crevasses, are easily detected using multiangular data.     

A Bayesian classification model for sea ice roughness fromscatterometer data

For sea ice in the Baltic Sea, surface scattering can be regarded as the dominant scattering mechanism at C-band. In this paper, a new statistical method is introduced for making statistical inferences about the underlying ice surface roughness on the basis of one-dimensional (1D) scatterometer data y. The central parameter in the hierarchical model applied in the context is a mixture parameter p, which indicates the degree of surface roughness in ice surface. Several questions related to the occurrence of different ice classes on a transect can be solved with the aid of the posterior distribution [p|y]. An empirical approximation for the posterior distribution is computed by using Markov Chain Monte Carlo methodology. The efficiency of the suggested approach is investigated by analyzing a C-band HH-polarization helicopter-borne HUTSCAT scatterometer data. The results provided by the statistical model show good agreement with a video-based ice type classification     

我国渤海域微波雷达海杂波功率对数正态分布模化

对流层波导可以使电磁波在较小的衰减下实现超视距传播,也可能造成雷达盲区,引入干扰,降低雷达、无线电通信系统的性能.因此,海上对流层波导参数的实时探测反演具有重要的意义.利用雷达海杂波反演对流层波导（refractivity from clutter, RFC）是当前国内外研究热点.雷达海杂波作为RFC输入的主要信息,其功率在传播路径上分布的随机特性严重影响反演效果,本文利用2011-06-2011-08和2012-07-2012-11渤海域对流层波导测量数据开展传播路径上雷达海杂波功率分布特性研究,有助于反演.比较测量数据和分布模型的概率密度函数表明,传播路径上的微波雷达海杂波功率不服从雷利分布,而服从对数正态分布.     

Retrieval of land and sea brightness temperatures from mixedcoastal pixels in passive microwave data

A technique is presented to separate uncontaminated land and sea brightness temperatures from mixed coastal pixels in 37-GHz vertically polarized passive microwave data from the Special Sensor Microwave Imager (SSM/I) instrument. Combining a mathematical model of the instrument response over several neighboring footprints with a GIS representation of the coastline yields a relationship between land and sea brightness temperatures and radiation measurements made at the satellite. Inverting this relationship allows separate land and sea brightness temperature values to be derived for each mixed coastal pixel in the original image. The technique has been successfully applied to 37-GHz vertically polarized SSM/I imagery for test areas covering the Gulf of Aden and the British Isles. Errors in the retrieved brightness temperatures were estimated to be of the order of 1-2 K     

Analysis of surface roughness and morphology of first-year sea ice melt ponds: implications for microwave scattering

Variations in wind forcing over summer first-year sea ice (FYI) melt ponds occur at hourly to weekly scales and are a significant contributor to microwave backscatter (/spl sigma//spl deg/) variability observed from spaceborne synthetic aperture radar (SAR) platforms (e.g., ENVISAT-ASAR and RADARSAT-1). This variability impairs our ability to use SAR to derive information on summer sea ice thermodynamic state and energy balance parameters such as albedo and melt pond fraction. The surface roughness contribution of FYI melt ponds in the Canadian Arctic Archipelago to like-polarized, C-band /spl sigma//spl deg/ estimates is analyzed through a spectral and statistical analysis of surface wave height profiles for varying wind speeds, upwind fetch lengths, and melt pond depths. A unique derivation of melt pond surface wave height spectra is presented based on digital video of melt pond surface wave trains. Significant scale surface roughness was observed even at wind speeds of 3 m/spl middot/s/sup -1/, resulting in small perturbation model estimates of /spl sigma//spl deg/ (HH) ranging from -5 dB at 20/spl deg/ incidence to -22 dB at 50/spl deg/ incidence. Results from a multivariate linear regression analysis show that 53.5% of observed variance in /spl sigma//spl deg/ (HH or VV) can be explained by wind speed, upwind fetch from melt pond edges, and melt pond depth, with no appreciable difference in the relative contribution of explanatory variables. Modeled omnidirectional /spl sigma//spl deg/ as a function of wind speed and incidence angle for 100-m transects collected throughout the melt pond season act to elaborate the role of fetch and depth, as well as the modulating effect of hummocks, on /spl sigma//spl deg/.     

Fusion of satellite active and passive microwave data for sea icetype concentration estimates

Young first-year sea ice is nearly as important as open water in modulating heat flux between the ocean and atmosphere in the Arctic. Just after the onset of freeze-up, first-year ice is in the early stages of growth and will consist of young first-year and thin ice. The distribution of sea ice in this thickness range impacts heat transfer in the Arctic. Therefore, improving the estimates of ice concentrations in this thickness range is significant. The NASA Team Algorithm (NTA) for passive microwave data inaccurately classifies sea ice during the melt and freeze-up seasons because it misclassifies multiyear ice as first-year ice. We developed a hybrid fusion technique for incorporating multiyear ice information derived from synthetic aperture radar (SAR) images into a passive microwave algorithm to improve ice type concentration estimates. First, we classified SAR images using a dynamic thresholding technique and estimated the multiyear ice concentration. Then we used the SAR-derived multiyear ice concentration to constrain the NTA and obtained an improved first-year ice concentration estimate. We computed multiyear and first-year ice concentration estimates over a region in the eastern-central Arctic in which field observations of ice and in situ radar backscatter measurements were performed. The fused estimates of first-year and multiyear ice concentration appear to be more accurate than NTA, based on ice observations that were logged aboard the US Coast Guard Icebreaker Polar Star in the study area during 1991     

Bayesian approach with the maximum entropy principle in image reconstruction from microwave scattered field data

Microwave imaging is of great interest in medical applications owing to its high sensitivity with respect to dielectric properties. It allows detection of very small inhomogeneities. The image reconstruction employing the microwave inverse scattering consists of reconstructing the image of an object from the scattered field measured behind the object. This reconstruction runs up against the nonuniqueness of the solution of the inverse scattering problem. The authors propose to solve the ill-posed inverse problem by a statistical regularization method based on the Bayesian maximum a posteriori estimation where the principle of maximum entropy is used for assigning the a priori laws. The results obtained demonstrate the power and potential of this method in image reconstruction.     

Determination of the age distribution of sea ice from Lagrangianobservations of ice motion

A procedure for monitoring the local age distribution of the Arctic sea ice cover is presented. The age distribution specifies the area covered by ice in different age classes. In the authors' approach, a regular array of grid points is defined initially on the first image of a long time series, and an ice tracker finds the positions of those points in all subsequent images of the series. These Lagrangian points mark the corners of a set of cells that move and deform with the ice cover. The area of each cell changes with each new image or time step. A positive change indicates that ice in a new age class was formed in the cell. A negative change is assumed to have ridged the youngest ice in the cell, reducing its area. The ice in each cell ages as it progresses through the time series. The area of multiyear ice in each cell is computed using an ice classification algorithm. Any area that is not accounted for by the young ice or multiyear ice is assigned to a category of older first-year ice. The authors thus have a fine age resolution in the young end of the age distribution, and coarse resolution for older ice. The age distribution of the young ice can be converted to a thickness distribution using a simple empirical relation between accumulated freezing-degree days and ice thickness, or using a more complicated thermodynamic model. The authors describe a general scheme for implementing this procedure for the Arctic Ocean from fall freeze-up until the onset of melt in the spring. The concept is illustrated with a time series of five ERS-1 SAR images spanning a period of 12 days. Such a scheme could be implemented with RADARSAT SAR imagery to provide basin-wide ice age and thickness information