GIS-BASED SUITABILITY EVALUATION OF THE DIFFERENTIAL RADAR INTERFEROMETRY METHOD FOR LANDSLIDE MONITORING

In recent years differential radar interferometry (D-InSAR) has proven to be a powerful remote sensing technique to detect and measure deformation of landslides with an accuracy of a few millimeters.

However, as a consequence of the inclined imaging geometry, areas with a topographic relief, where landslides usually occur, appear heavily distorted in the radar image. Thereby slopes inclined towards the radar appear shortened and in extreme even can cause an overlapping of different radar signals (layover effect); slopes oriented away from the radar seem stretched or even can be shadowed by a steep mountain (shadowing). These effects limit or even prohibit the use of a radar image for interferometric applications.

Besides these geometric distortions, the land cover has great influence on the applicability of differential radar interferometry. For example vegetation-free areas such as buildings, roads and rocks show a high stability in their backscattering properties, whereas areas covered by vegetation, especially forests, have varying backscattering properties at different times. Areas with a high constancy of their backscattering properties are better suited for the D-InSAR-technique.

To date prior to an investigation using D-InSAR these limiting effects usually are only estimated, sometimes leading to disappointing results when the actual radar images are analyzed. Therefore in the present work a GIS routine was developed, which based on freely available DEM data not only accurately predicts the areas in which layover and shadowing will occur, but also determines the measurable percentage of the movement for a given radar acquisition geometry. Additionally land cover classification data is used to assess the influence of the land cover to D-InSAR deformation measurements.

Thus by using this new GIS application, in future it is possible to evaluate the usability of D-InSAR landslide deformation measurements in a certain region quite accurately before the expensive actual radar records are ordered.