Mapping Wetlands in Zambia Using Seasonal Backscatter Signatures Derived from ENVISAT ASAR Time Series
VerfasserSchlaffer, Stefan ; Chini, Marco ; Dettmering, Denise ; Wagner, Wolfgang In der Gemeinsamen Normdatei der DNB nachschlagen
Erschienen in
Remote Sensing, Basel 2016, Jg. 8, H. 5, S. 402
ErschienenMDPI 2016
Published version
DokumenttypAufsatz in einer Zeitschrift
Schlagwörter (EN)wetlands / SAR / time series / Fourier analysis / local incidence angle / radar altimetry
URNurn:nbn:at:at-ubtuw:3-1600 Persistent Identifier (URN)
CC-BY-Lizenz (4.0)Creative Commons Namensnennung 4.0 International Lizenz
 Das Werk ist frei verfügbar
Mapping Wetlands in Zambia Using Seasonal Backscatter Signatures Derived from ENVISAT ASAR Time Series [44.52 mb]
Supplementary File 1 [37.68 mb]
Zusammenfassung (Englisch)

Wetlands are considered a challenging environment for mapping approaches based on Synthetic Aperture Radar (SAR) data due to their often complex internal structures and the diverse backscattering mechanisms caused by vegetation, soil moisture and flood dynamics contributing to the resulting imagery. In this study, a time series of >100 SAR images acquired by ENVISAT during a time period of ca. two years over the Kafue River basin in Zambia was compared to water heights derived from radar altimetry and surface soil moisture from a reanalysis dataset. The backscatter time series were analyzed using a harmonic model to characterize the seasonality in C-band backscatter caused by the interaction of flood and soil moisture dynamics. As a result, characteristic seasonal signatures could be derived for permanent water bodies, seasonal open water, persistently flooded vegetation and seasonally flooded vegetation. Furthermore, the analysis showed that the influence of local incidence angle could be accounted for by a linear shift in backscatter averaged over time, even in wetland areas where the dominant scattering mechanism can change depending on the season. The retrieved harmonic model parameters were then used in an unsupervised classification to detect wetland backscattering classes at the regional scale. A total area of 7800 km2 corresponding to 7.6% of the study area was classified as either one of the wetland backscattering classes. The results demonstrate the value of seasonality parameters extracted from C-band SAR time series for wetland mapping.