Research On Habitat Classification Of Red-crowned Crane In Zhalong Wetland Based On Remote Sensing Technique

Biodiversity is vital for wetland ecosystem standing and developing. Waterfowls, a kind of birds relying on wetland ecologically, are the major component of wetland biodiversity. Therefore the status of wetland habitat will directly affect the waterfowls'surviving. In order to protect wetland biodiversity, protecting their habitat has been one of the hot spot in nowadays research.Zhalong Natural Reserve in Heilongjiang Province is famous national nature reserve, which protects some rare waterfowls such as red-crowned cranes and their habitat. Zhalong Natural Reserve posseses of the largest, relative original and intact bulrush wetland in China at present. However, with the change of natural environment, high-intensity man-made interference and the change of climate in this region, the living space for red-crowned cranes has been threatened severely by both of the abrupt decreasing of wetland area and the degradation of wetland's ecological functions in a large scale. For the purpose of protecting red-crowned cranes, it's necessary to understand and master the spatial characteristics and the changes of the red-crowned cranes'habitat in detail from the view of their living.The nest-site selection of red-crowned cranes'habitat during breeding period in is affected by various natural and human factors. Based on the habitat characteristics of red-crowned cranes'during breeding period in Zhalong Wetland, high spatial resolution ALOS images and other data, the paper mainly discusses the four habitat factors including terrain, water bodies, vegetation and external interference sources. With using different classifying methods on remote sensing, the classification of red-crowned cranes'habitat during breeding period in Zhalong Wetland has been accomplished eventually. The major works in this paper are as follows:1. Describing the habitat characteristics of red-crowned cranes'habitat during breeding period in Zhalong Wetland. The four main determined habitat factors are terrain, water bodies, vegetation and external interference sources by considering the actual characteristics in the study area and the project's conditions.2. Determining the classification principle of red-crowned cranes'habitat considering the physical characteriscs of Zhalong Wetland and the spectral analysis of ALOS images. Based on this classification principle, the interpretation symbols for the landscape type and the landscape classification system on red-crowned cranes'habitat in the study area were established. With the interpretation symbols and classification system, the landscape map of red-crowned cranes'habitat in 2008 was completed by visual interpretation method.3. DEM establishment. The elevation lines and elevation points are extracted from relief map with scale 1:10,000 based on MapGIS 6.7. At the same time, lakes'boundary as the control parameters are also extracted from ALOS images for DEM establishment. In order to depict the whole terrain characteristics, DEM grid map, slope and aspect map derived from it were generated by elevation interpolating using irregular triangulation network method.4. Representative sample area was chose to do experiment using classification methods based on remote sensing. Supervision classification method, object-oriented information extraction method and decision tree based on expert knowledge method were mainly discussed. Through accuracy evaluation based on error matrix, results indicated that both object-oriented information extraction method and decision tree based on expert knowledge method showed their own advantages. So time can be saved and efficiency can be improved combining the two methods.5. Water bodies extraction. Through analyzing the spectrum characteristics of water bodies with other objects for ALOS image, water bodies could be automatically extracted by using threshold method and decision tree method. By comparing with water bodies by visual interpretation, the accuracy of automatic results could reach 85%.6. Realizing the vegetation information extraction and bulrush sub-classification. The NDVI grid image has been accomplished by using the modeling function based on ERDAS 9.1. And based on the NDVI grid image, bulrush sub-classification was completed by using statistic analysis method. A qualitative description about the relationship among bulrush, terrain and slope was made. On the theory basis of the habitat characteristics of nest-site selection, combined with various data sources, another qualitative description about the rationality and validity of bulrush sub-classification was done.7. Based on the conclusion of experiments about remote sensing classification, combined with all kinds of existing remote sensing data, the habitat classification map of red-crowned cranes'habitat during breeding period in Zhalong Wetland was eventually accomplished by using object-oriented information extraction method, decision tree based on expert knowledge method, also including necessary visual interpretation method.Finally, the shortages in this paper were listed and further improvement were discussed which will be further rearched in future.