Analysis And Realization On The Change Of Land Utilization In Hetao Irrigation Area Based On TM Remote Sensing Image

Hetao area,located in the middle reaches of the Yellow River,is not only one of the three oversize irrigation area in China,but also the largest diversion irrigation area in Asia.Its role and status play a decisive role.In order to precisely know well about the change of land utilization in the Hetao irrigation area from 1990 to 2010,this thesis not only classifies and statistically analyzes the land utilization of the Hetao irrigation area in 1990,2000 and 2010 using the TM remote sensing images,but also provides a scientific basis for the land resources management and rational allocation of irrigation water and the main contents are as follows:(1)The splicing and cropping method of TM remote sensing image in Hetao irrigation area were studied,and we visually interpreted the panorama image after splicing and establish the land classification system of irrigation area.The three remote sensing images contain three images,and each image includes seven bands respectively.According to the abundant information recorded in the TM remote sensing image and the large region for the Hetao irrigation area,the SURF algorithm with rotation and scaling invariance was selected to extract the feature points of the images.Shearlet transform algorithm was used to fuse the images after matching.The complete map of the Hetao irrigation area was then obtained after splicing.The region of interest was then extracted by the polygon irregular cutting algorithm.The optimal exponential method was selected to obtain the optimal band combination of each image because it is simple,efficient and convenient to calculate According to the type of land utilization in Hetao irrigation area.(2)Three kinds of supervised classification methods were adopted,including maximum likelihood method,neural network method and support vector machine method.And the classification method was then optimized by combining hierarchical classification and support vector machine method.Support vector machine classification theory is used to study the case of small sample..Moreover,it shows good applicability in practical remote sensing image classification application.Hierarchical classification is based on different classification strategies while the traditional single classifier is not targeted for different recognition accuracy of different types of objects.Therefore,support vector machine and hierarchical classification were combined to optimize the land classification method.The results show that the support vector machine behaves globally better than the maximum likelihood method and neural network method.Meanwhile,the support vector machine and the hierarchical classification are combined to give full play to their respective advantages and the classification effect is remarkably improved.The overall classification accuracy of the stratified classification and Kappa coefficient reached86.10% and 0.81,respectively,for 1990,93.66% and 0.99 for 2000,and 92.80% and 0.90 for2010.(3)Three The land classification results of the Hetao irrigation area was statistical analyzed by means of land use dynamic degree.In order to obtain the quantitative analysis information of land use utilization from 1990 to 2000,the dynamic degree of land utilization can quantitatively describe the rate of land utilization change,which has a positive effect on predicting the future trend of land utilization change.Therefore,dynamic degree method was used to obtain the results of land utilization change in the Hetao irrigation area from 1990 to 2010,and the overall land utilization dynamic degree was 2.08% from 1990 to 2000,2.83% from 2000 to 2010,1.13% from1990 to 2010.According to the changes in the past two decades,the fastest rate of change was the residential area,up to 13.80%,while the slowest rate of change was the cultivated land,up to0.97%.