Seersma, Jessica 1
1 University of Colorado at Boulder
Routine and recurrent flooding has been experienced in many areas of the world since the time of the first flood of record (God). Flooding can serve to replenish groundwater supplies, preserve wetland habitats, and provide irrigation. However, excessive or recurrent flood inundation destroys crops, causes the temporary and permanent displacement of residents, results in the incurrence of economic losses resulting from property damage and loss of operating revenue. Techniques that combine multi-temporal and multi-spectral optical passive sensor with active sensor derived synthetic aperture radar (SAR) data allow for an adequate representation of the landform and background information while providing information about flood extents during and just subsequent to floods. Fourteen weeks of sustained rainfall beginning in mid-September of 2000 resulted in extensive fluvial flooding in England, Wales, and part of Northern Ireland and Scotland.
The flood inundation area extents resulting from these floods that occurred along the River Severn extending from the southern portion of Worcester, Worcestershire to a downstream location southwest of Gloucester, Gloucestershire were evaluated through the use of a data fusion technique adapted from (Yonghua, 2007). A comparative analysis was performed to assess the overall effectiveness in the determination of the flood extents resulting from the data fusion technique using Normalized Difference Vegetation Index (NDVI), maximum likelihood classification, and histogram thresholding applied to optical imagery taken after the flood. ERS SAR images sampled to a resolution of 10 meters, and 20 meter resolution HRVIR SPOT 4 optical imagery were used to perform the analysis.
The fusion technique was applied to the ERS radar data taken after the flood and the SPOT imagery taken before and after the flood, respectively. In the fused images the distribution of water and non-water was non-homogenous with the exception of the river and floodplain areas, Figure 1. Therefore, there wasn’t a significant correlation in the spatial distribution of the water and non-water classified areas between the fused images and the images produced using only the SPOT imagery taken after the flood. Although, the boundaries of the flood inundation areas were not clearly defined in the fused image that was produced using the SAR data taken after the flood and the SPOT imagery taken before the flood, the boundaries were clearly defined in the fused image that was produced using the SAR data taken after the flood and the SPOT imagery taken after the flood. It was also shown that between 74% to 82% of the area classified as water using only the SPOT imagery taken after the flood were classified as water in the fused images, Figure 2.
God. New International Version Genesis 7: 11-12. Retrieved July 19, 2011, from Bible Gateway http://www.biblegateway.com/passage/?search=Genesis%207&version=NIV
Yonghua, S. X. (2007). A Study on Optical and SAR Data Fusion for Extracting Flooded Area. Institute of Electrical and Electronics Engineers, 3086 - 3089.