Sub-pixel Mapping Of Flood Submerged Range Based On MODIS Data

As a natural phenomenon,flood has an important impact on people's production and life.Therefore,it is of great significance to monitor flood,especially flood submerged range.MODIS image has high temporal resolution and wide coverage.It is an important data for long-time and large-scale flood monitoring.However,due to the mutual restriction between spatial resolution and temporal resolution of remote sensing images,the spatial resolution of MODIS images is low and the phenomenon of mixed pixels is obvious.This seriously restricts the accuracy of flood submerged range mapping.Therefore,this paper proposes to use sub-pixel mapping technology to solve the mixed pixel problem in MODIS,and realize fine mapping of flood submergence.This paper mainly completes the following work:1.This paper analyzes the shortcomings of PPI endmember extraction algorithm and proposes an extraction optimization method,and carries out the accuracy verification.This optimization method combines PPI and high resolution remote sensing images.The experimental results show that the endmember spectrum extracted by the optimization algorithm has clear meaning,and the RMS Error is less than 0.025,accounting for more than 94.3%,which is higher than the PPI algorithm.2.In this paper,the sub-pixel level extraction of flood inundation area is realized based on SPSAM sub-pixel location algorithm.The overall accuracy of the two experimental results is more than 91%.Considering the influence of sub-pixel micro-topographic fluctuation on flood submergence area,this paper presents a D-MSPSAM algorithm.The algorithm uses DEM elevation information to assist in correcting the abnormal distribution of water body in sub-pixel mapping results.The corrected results preserve the spatial details of the water body well.Many of the misalignments caused by the approximation of spectral characteristics to water bodies have also been corrected.The positioning accuracy was further improved,reaching 91.40%in flood season and 92.47%in dry season.This proves the application potential of DEM data in flood submergence extraction,and provides a new idea for the improvement of sub-pixel mapping model.3.This paper batch processed the MOD09A1 image covering the entire Dongting Lake area from 2008 to 2017,and studied the interannual variation of the maximum flooding range and flooding frequency of large-scale and long-term floods.The results show that 2016 has the largest area and the highest risk.In the 10 years from 2008 to 2017,22.67%of the total area of the study area was covered by water without flood risk permanently;6%of the total area had frequent land-water replacement,which was an important area for flood monitoring.