HPC MSU

Publication Abstract

Stressed Vegetation Identification by SAR Time Series as an Indicator of Slope Instability in Mississippi River Levee Segments

Hasan, K., Aanstoos, J.V., & Mahrooghy, M. (2013). Stressed Vegetation Identification by SAR Time Series as an Indicator of Slope Instability in Mississippi River Levee Segments. 2013 IEEE Applied Imagery Pattern Recognition Workshop. Washington, DC: IEEE. DOI:10.1109/AIPR.2013.6749307.

Abstract

Surface vegetation reflects various characteristics of the soil on which it grows. Vegetation type and growth rate differences were observed between recently cracked surfaces and stable soil on earthen levees along the lower Mississippi River. We attempted to directly characterize the levee surface beneath the vegetation cover using X-band synthetic aperture radar from TerraSAR-X. Due to its short wavelength, however, most of the backscatter is from the vegetation rather than the soil. Hence a time-series of the SAR imagery was made over the time when vegetation growth and biomass were at their lowest and the multi-temporal radar backscatter pattern was used to identify healthy and stressed vegetation growing over stable and unstable (subject to slump slides) levee segments. Field data showed that vegetation was most vigorous over healthy or repaired levee surfaces and poorest over areas with surface cracks. Three time series of the HH and VV and HH-VV images were made for 6 dates over a 7 month period. Correlation with field data and analysis revealed that imagery from October through February was most effective in identifying the target vegetation differences. In this time span the differences in vigor of the vegetation were greatest between healthy levee surfaces and those with cracks on them (which could lead to potential levee failures). The 3 time series were classified independently using field derived training polygons and the stressed vegetation class was extracted from it. More than 90% of the known slump slides were identified by the classification but a significant number of ‘false positives’ resulted.