| Global change is closely linked to human life.So more and more experts have begun to study global change because their significant change attracts more and more people’s attention.The study of global change and sustainable development of human society needs high precision global land cover data’s support.And the Globel Land30,who developed by the experts and scholars in our country,just supports the high resolution of the global land cover data to research and monitor the global change.Due to the difference of politics,economy,geography,culture,religion in different country,the spectral characteristics,morphological characteristics and geometrical features of artificial cover in different countries are not same as each other.In this way,they used many ways to extract artificial cover in the process of developing the product,but with a variety of artificial cover’s difference in spectral,shape,texture,and remote sensing image has the phenomenon of different objects which have the same spectrum and the same objects have different spectrum,some wrong is inevitable.This paper summaries the local induction and formation of inspection rules of GlobelLand30’s artificial covered data from artificial covering type,geographical factors,human factors,economic factors and urbanization to overall and then feedbacks the errors to improve the Globel Land30 accuracy.In this paper,we find a law from examining the artificial cover based on the knowledge that the extraction error of the artificial cover and the cultivated land is more prone to appear in the process of GlobelLand30’s artificial cover extraction.DMSP/OLS night light data is a kind of remote sensing image data and it describes the night lights’ brightness,and its sensor can detect the small artificial lights at night,even traffic lights,etc.According to this feature,this paper extracts the town’s contour based on the SVM region growing method,so that we can eliminate the influence of towns surrounding a large number of cultivated lands.We get the study area from dealing with the TM image based on DMSP/OLS night light data extracted from the urban contours,and analysis the six TM bands spectral characteristics,and build a decision tree by adding texture features.TM6 band is thermal infrared band,the heat island reaction causes the artificial cover’value in the TM6 band is higher than the other and has some certain separabilityIn this paper,we add TM6 band on the basis of NDBI to calculate the surface radiation artificial TRNDBI(Thermal Radiation Normalized Difference Building Index)to extract the artificial covered.And then by comparing the TM4 and TM5 band values greater than 30 as a condition of a judge.Using the DMSP/OLS nighttime light data,TM4,5,6 bands and contrast to build decisiontree to extract the artificial covered.In the last we use confusion matrix and Kappa coefficient for precise analysis.After verification,the result is as follows: the overall accuracy of confusion matrix is 87.7%,the Khat is 81.4%.In the end of this paper,we analysis urban expansion rationality by taking The Beijing-Tianjin-Hebei integration as a example,and note the unrationality and the step we should do. |
PDF Full Text Request