Methods for analyzing SAR images

Yrjö Rauste

    Research output: Book/ReportReport

    10 Citations (Scopus)


    Methods for SAR (Synthetic Aperture Radar) image analysis have been studied. A Seasat SAR image of a test site in northern Sweden near Arjeplog has been analysed together with a Landsat Thematic Mapper image of the same area. The SAR image was rectified using a polynomial rectification method with a digital elevation model. A backscatter anomaly map was produced showing areas where the backscattering coefficient of the soil/vegetation deviates from the average backscatter within the image. Quantitative analysis of topography-induced variation in SAR images was carried out. The separability of four land cover classes (spruce-dominated mixed forest, pine dominated mixed forest, deciduous forest, and regenerated area) in Seasat data was studied. The high accuracy of the polynomial rectification of SAR images (RMSE less than a resolution cell) shows that polynomial rectification using digital elevation data is well suited for applications where SAR data is merged with images from other sensors. Field checks showed that the backscatter anomaly map, produced using Seasat SAR data and a digital elevation model, highlights outcrops and areas covered by boulders. Terrain topography can explain more than 65 per cent of the total variation of SAR image in land areas. The separability of land cover classes improved with increasing incidence angle and increasing amount of speckle reduction.
    Original languageEnglish
    Place of PublicationEspoo
    PublisherVTT Technical Research Centre of Finland
    Number of pages102
    ISBN (Print)951-38-3434-4
    Publication statusPublished - 1989
    MoE publication typeD4 Published development or research report or study

    Publication series

    SeriesValtion teknillinen tutkimuskeskus. Tutkimuksia - Research Reports


    • remote sensing
    • synthetic aperture radar
    • image analysis


    Dive into the research topics of 'Methods for analyzing SAR images'. Together they form a unique fingerprint.

    Cite this