Study On Monitoring Method For Ecological Construction Of Water And Soil Conservation Based On High-resolution Remote Sensing Images

Water and soil conservation is one of the key measures to solve the problems of the riverbed sediment, to keep it alive and improve the catchment eoc-environment of Yellow River. Rencent years, the practices in eco-construction of soil and water conservation on Loess Plateau were envisioned effective apprently. The monitoring of the process of the soil and water conservation areas is found essential to management, assessment and decision-making. However, the main methods of monitoring the catchment at present are all manual, including statistics and reporting to the higher level agent, which result in time consuming and expensive to sustain the practice that are dominated by people. In this thesis, the monitoring of ecological engineering construction in water and soil conservation based on remote sening images with high resolution is adopted to study the sample area along the Yiluo River, part of the hilly and valley region of Loess Plateau. After analyzing the basic information of the ecological engineering construction, the spatial and temporal scale, and monitoring index, the 1m IKONOS image and 2.5m SPOT5 image were introdueced as data sources, the methodology of RPC modeling and GCP optimitiaztion as modification to the interpretation, the AHP used in determining combination of multi-light specrum and the whole band wave, the virtue reality technique monitoring the classification and recognation, variable cluster allocation coeffiecents and its adaptation in spatial and imaged attributs in the research on water and siol conservation subtracting the least pathches of the area. The final stage is to subtract the information via human and computer interactive to interpret and monitor the classification in the study assisted with GPS, field identification and construction supervision.The research results show that the optimization of modification to deviation residue with detenninant points based on RPC modeling corrected by IKONOS image results obviously highest accuracy relatively. The AHP in analyzing the fusion of Pan-spectral and Multi-spetral IKONOS image is the best methods on Loess Plateau compare to other routine ways. Based on planar environment to find the classification signature, the classification accuracy of SPOT imagery with 2.5 meter resolution is 88.19%, while based on the virtual reality technology to define the classification signature; the classification accuracy comes to 95.48%. In the hilly and valley regions of Loess Plateau, using the SPOT5 classified image to extract the construction information of water and soil conservation, the optimical minimum patch by which to postprocess is 6*6 pixels. The monitor results indicates that the process of the project get on well, complete the 94% and 95% of the 2002 and 2003 plans. Through the study, in the mornitor information of water and soil conservation, the best measure is combining the SPOT5 with 2.5 meter resoulution from which the information is gained by auto-unsupervised classifying and IKONOS with 1 meter resoulution from which the information is extracted by manual interpretation. This method is fast, economic and practical and has a bright practical prospect.