OBJECT-BASED IMAGE CLASSIFICATION AND WEB-MAPPING TECHNIQUES FOR FLOOD DAMAGE ASSESSMENT

摘要

The importance and power of geospatial information technology was proven during the recent (June 2008) Midwest floods in some low lying areas along the river courses. This paper studies the use of the Landsat satellite images to assess the flood damages in nine of the counties in southern Indiana, and the development of Web-based mapping and reporting based on the Google Earth API technique. The Landsat images acquired on June 11, 2008 became the most relevant data as the peak flooding was observed on June 10 and 11 in this area. The availability of June 12, 2007 Landsat imagery over the same area enables us to observe the changes and accurately map the flood extents. To do so, both temporal images were classified using object-based image classification method to avoid the inherent problems of the traditional pixelbased methods. In this technique, segmentation produces homogeneous regions or objects with spectral, spatial and texture features. This substantially improves the classification results. The results are then made available on the Internet through Google Earth API (Application Programming Interface) released by Google in June 2008. We embed a number of functionalities of visualization and query of Google Earth into our own webpage using GE API and script language. Its availability on the Web helps decision makers to respond and take necessary actions against such natural hazards. Based on this study, it is found that most of the damages were caused to the standing crops at the early stage of their growth. Crops Data Layer 2007 (CDL) of USDA was used to assess the crops damage, assuming the similar crops in the same areas during year 2008. The results show that an average about 15-16percent of corn and soybean, 5percent of wheat and 5percent of pasture and hay were affected by the flooding. Comparison with Indiana Department of Transportation (INDOT) roads 2005 data shows that about 5percent of the roads (mostly county roads) were affected by the flooding.