Fast Flooding Information Extraction in Emergency Response of Flood Disaster

Flood disaster is one of the natural disasters those have the highest occurrence frequency. Remote sensing plays an active and important role in the flood disaster management such as monitoring, evaluation, emergency response and so on. Affected by the characteristics of flood disaster, emergency response for flood disaster must be guaranteed by a high quality and speedy remote sensing data resources. It has a special requirement to the speed of flood disaster information extraction and updating. In operational works, not same as other ordinary scientific research, “data-information” flow is always a process need to be accomplished automatically. Considering that the ability of extracting flood information will definitively affect the following operational work such as disaster situation analysis, loss evaluation, decision-making support for salvations and so on, it is need to make a father research on the technology of fast automatic flooding information extraction. According to the water characteristics that reflect on the remote sensing images (including optical sensor and microwave radar), the method of extracting flooded area from remote sensing images that can be used currently was analyzed. It includes threshold method, NWDI method, NDVI method, and so on. In the aspect of implementation and application, the automatic abilities of these methods were discussed and they are all thought having some automation ability. Based on this discussion, two automatic/semi-automatic methods of extracting flooded area information fast from remote sensing images were designed: decision tree model method and programming method. Being affected by spectrums of objects, these two methods are all disturbed badly by mountain shadow. By researching on the model encapsulate technology and combining the terrain analysis function of GIS spatial analysis, affection of mountain shadow can be mitigated and the result data can be updated automatically. All the methods reported in this article integrated in a case-study system. They were thought having some good effect. Under the situation not having human-intervention, the flooded area can be extracted from remote sensing images directly.