Robust flood area detection using a L-band synthetic aperture radar: Preliminary application for Florida, the U.S. affected by Hurricane Irma
Abstract
Urgent crisis response for a hurricane-induced flood needs urgent providing of a flood map covering a broad region. However, there is no standard threshold values for automatic flood identification from pre-and-post images obtained by satellite-based synthetic aperture radars (SARs). This problem could hamper prompt data providing for operational uses. Furthermore, one pre-flood SAR image does not always represent potential water surfaces and river flows especially in tropical flat lands which are greatly influenced by seasonal precipitation cycle. We are, therefore, developing a new method of flood mapping using PALSAR-2, an L-band SAR, which is less affected by temporal surface changes. Specifically, a mean-value image and a standard-deviation image are calculated from a series of pre-flood SAR images. It is combined with a post-flood SAR image to obtain normalized backscatter amplitude difference (NoBADi), with which a difference between a post-flood image and a mean-value image is divided by a standard-deviation image to emphasize anomalous water extents. Flooding areas are then automatically obtained from the NoBADi images as lower-value pixels avoiding potential water surfaces. We applied this method to PALSAR-2 images acquired on Sept. 8, 10, and 12, 2017, covering flooding areas in a central region of Dominican Republic and west Florida, the U.S. affected by Hurricane Irma. The output flooding outlines are validated with flooding areas manually delineated from high-resolution optical satellite images, resulting in higher consistency and less uncertainty than previous methods (i.e., a simple pre-and-post flood difference and pre-and-post coherence changes). The NoBADi method has a great potential to obtain a reliable flood map for future flood hazards, not hampered by cloud cover, seasonal surface changes, and "casual" thresholds in the flood identification process.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFMNH23E2830N
- Keywords:
-
- 1910 Data assimilation;
- integration and fusion;
- INFORMATICS;
- 1922 Forecasting;
- INFORMATICS;
- 4313 Extreme events;
- NATURAL HAZARDS;
- 4331 Disaster relief;
- NATURAL HAZARDS