The Landscape Change Based On Maximum Likelihood Classification Of Ningxia Hongsipu Irrigation District

Hongsipu is the main irrigation Area of the Yellow projects in Ningxia,and is a national key construction project. Its essence is that developmented oasis in the desert. It is not only fundamentally improved the local ecological environment, but will have a important strategic significance to economic development and social prosperity in southern Ningxia after the completion of irrigation development. In order to be Scientific and rational for the region's agricultural development planning, layout and planting crops.The paper usesd the remote sensing classification techniques which a linear spectral mixture decomposition combined with maximum likelihood classification, researchedthe landscape changes of land cover of Hongsipu, grasping of land-cover information in the development of prestate to faster and better, to promote the local agricultural economic development.It is the main study area of Hongsipu Irrigation, collected TM remote sensing data and the natural environment,socio-economic in the extensive, using a variety of remote sensing information technology to extract information.It extracted sensing information to Ningxia Hongsipu Irrigation 1989,1999,2003 and 2008 by RS, GIS software. After the Hongsipu remote sensing image are geometric correction and radiometric correction,using the traditional supervised classification methods-maximum likelihood classification, it is taking into account the impact of mixed pixel to classification, through analysis and comparison,It decomposed mixed pixel decomposition by a linear mixed model, to obtain the area percentage of each surface feature,, on this basis, Selecting training samples to classify.it has higher classification accuracy.it can meet To meet the needs of practical application. Completion of remote sensing classification, Application the principles and methods of landscape ecology, using FRAGSTARTS analysis software.According to research needs, this paper selected the relevant landscape pattern index and the index of patch characteristics of landscape elements, and use the relevant index which was selected analysis the landscape pattern and dynamics in different years, that is, it analysised the landscape patterns and dynamics at 1989,1999,2003 and 2008, four times of Hongsipu, separately from the landscape level and landscape-scale, and analysis the conversion features of each type of the landscape through the transfer matrix,the final analysis of the hallway structural characteristics of the man-made ditch and the impact of landscape pattern. The results show that:The results The results show that:(1) mixed pixel phenomenon is particularly serious in Hongsipu Irrigation, it can decompose mixed pixel well using linear spectral mixture decomposition, it can be used in the classification and extraction of surface information widely.In the decomposition, how to choose pure end-member is the key.(2) principal component transform can remove the redundant information between the bands, compress the data dimension, suppress the noise, remove the correlation between bands and choice the end-member using its Scatter of the band, the physical meaning of clear,The end-member to choice is relatively pure.(3) Hongsipu Irrigation Area from 1989 to 2008, growing role in human disturbance, the landscape pattern change significantly, the number of patches increased, fragmentation increases, reducing the number of fractal dimension, landscape shape becomes more rules and simple. In 1989, the main landscape types of grass, to 2008 is mainly sandy, cultivated land increased year by year, which is due to human activities directly affect the outcome, it cultivated grassland to cropland for a large number, grassland and Shrubs are part of the desert, part of the desert shrub, in part converted into farmland.(4) In 1989-2008 years, the region's network structure of canals is growing complexity, a direct impact on the structure type of landscape. the landscape diversity index and evenness index were increased,the richness of landscape element types increases with the ditch network structure complexity...