Title | Evaluation of c-band SAR for identification of flooded vegetation in emergency response products |
| |
Author | Brisco, B; Shelat, Y; Murnaghan, K; Montgomery, J; Fuss, C; Olthof, I; Hopkinson, C; Deschamps, A; Poncos, V |
Source | Canadian Journal of Remote Sensing vol. 25, issue 1, 2019 p. 73-87, https://doi.org/10.1080/07038992.2019.1612236 |
Year | 2019 |
Alt Series | Natural Resources Canada, Contribution Series 20200473 |
Publisher | Taylor & Francis |
Document | serial |
Lang. | English |
Media | paper; on-line; digital |
File format | pdf; pdf |
Released | 2019 05 24 |
Abstract | A synthetic aperture radar (SAR) data set of the Peace Athabasca Delta, Alberta, was used to evaluate approaches to flooded vegetation mapping. A primary objective was to identify how to add a flooded
vegetation layer to the Emergency Geomatics Service (EGS) SAR-derived flood products. Field data were used to identify non-flooded and flooded vegetation. A combination of statistical analyses and box plot visual inspection was used to evaluate the
magnitude-only images, the polarimetric and compact polarimetric parameters/decompositions, and the coherence products for mapping flooded vegetation. This paper provides some background on the use of SAR for flood mapping, describes the data and
processing methods, and presents the results of this comparison. To some degree all polarizations and techniques were effective for mapping flooded vegetation due to the increased backscatter intensity and the phase shift from the double bounce
scattering. In particular, all polarization combinations, the HH/HV ratio, Shannon entropy, and the m-chi double bounce produce good separation. The water/vegetation interface remains coherent when flooded, also allowing flooded vegetation to be
identified with seasonal coherence. These results demonstrate that the use of multi-mode RADARSAT Constellation Mission data for providing a flooded vegetation layer to EGS flood maps is possible. |
GEOSCAN ID | 327319 |
|
|