Coastal polynyas in the Arctic basin from the A inter period (January to April) are characterized using ESA European Remote Sensing satellite (ERS)-1/2 Synthetic Aperture Radar (SAR) Precision (Precise Image, PRI) and Browse images. A SAR polynya algorithm is used to delineate open water, new ice, and young ice, and to define the size and shape of polynyas. In order to extract radiometric and contextual information in the ERS SAR PRI images. three different image classification routines are developed and applied. No in situ data have been available for verification of the polynya shapes and sizes, but one of the ice classification routines have been verified earlier using ground truth data. The SAR polynya algorithm is demonstrated to be able to discriminate between the polynya and the surrounding ice area for 85 analyzed cases. The results from the SAR algorithm are compared to ERS Browse images. passive microwave data (a recent Polynya Signature Simulation Method (PSSM), and the Bootstrap and the NASA Team ice concentration algorithms), and a numerical polynya model (NPM) forced by National Center for Environmental Predictions (NCEP) wind fields and air temperatures. The ERS SAR Browse images show a relatively good correlation with the ERS SAR PRI images (.88) while the Special Sensor Microwave Imager (SSM/I) Bootstrap and the NASA Team ice concentration algorithms both have low correlation coefficients (below .3). The PSSM calculates the polynya shape and size, and delineates open water and thin ice. For polynyas of all sizes it has a correlation of .69 compared to the SAR PRI images. For polynyas with widths greater than 10 km the correlation increases to .83. The NPM computes offshore coastal polynya widths, heat exchange. and ice production. Compared to SAR data, it overestimates the maximum size of the polynya by about 15% and has a correlation of .71 compared to the analyzed SAR PRI images. The polynyas in our main investigation area, located at Franz Josef Land, are found to be primarily wind driven. The surrounding large-scale ice drift and tidal currents have little effect on the polynya behavior. One overall conclusion from this investigation is that SAR images processed through the SAR polynya algorithm in combination with the NPM is a powerful tool for investigating and characterizing polynyas at various scales in the Arctic. (C) 2002 Elsevier Science Inc. All rights reserved.