The Analysis And Extraction Of Soil Conservation Practices Information Based On Hi-Resolution Remote Sensing Images

In the aspect of the extraction of information on land use and Soil and Water Conservation Practices, previous studies were mainly based on TM and other middle-resolution Remote Sensing image data. Limited by the resolution, small pieces of the distribution of soil and water conservation Practices such as terracing, orchard, grass and plantation land and dam land can not be extracted effectively. The rapid development of Remote Sensing technology is resulting in the increasing resolution of remote sensing images. So the information of Soil and Water Conservation Practices extracted on high-resolution Remote Sensing images has become feasible. With the improving cost performance of high-resolution Remote Sensing images, Soil and Water Conservation Monitoring using these high-resolution images as an information source is very important, especially on the hilly and gully areas of the Loess Plateau.This study was performed on the Liulin Zhen Ma Zhuang watershed in the Yanan Baota District. Based on the technology of RS and GIS, the information of Soil and Water Conservation within this area was extracted. Use was made of visual interpretation and a semi-automatic classification method, which is a combination of supervised classification and visual drawing. This method was applied on the IKONOS high-resolution Remote Sensing image shot of 2001 and supplied maps with the information on the Soil and Water Conservation Practices of the actual situation in 2001 in Ma Zhuang Watershed. Then the map precision of the actual Soil and Water Conservation Practices information extracted by the semi-automatic method was appraised. This was done by using the map of the actual information on Soil and Water Conservation Practices, extracted by visual interpretation as a reference value. Furthermore, in order to study the law of the accuracy of information extraction from remote sensing images with different resolutions, the 1m IKONOS image was resampled to a different resolution image. The Soil and Water Conservation Practices information from these images was again extracted with help of semi-automatic methods. Then the precision of these maps extracted by semi-automatic methods from different Resolution Images was appraised. Following are some results:1) Supervised classification was used to extract the information from Remote Sensing images and supplied the preliminary classification map. The preliminary classification was post-processed with help of the visual drawing and other post-processing methods, resulting in maps of Soil and Water Conservation Practices information. The total precision of this map can be up to 85.87%. Except for the cartographic accuracy of forest land which was 76.73%%, the cartographic accuracy of other land-use types are all above 97%. This shows that the used method of high-resolution Remote Sensing images for semi-automatic extraction of Soil and Water Conservation Practices information are feasible.2) Maps from different resolution images do not appear to have certain laws about the accuracy of information extraction. The true reasons behind the relation between information extraction and resolution images need more discussion and work.