| This paper takes SiChuan MianYang as example, and classifies the remote sensing imagesin the years of2002and2007with the method of the maximum likelihood classification and getsthe classification maps of remote sensing images. According to the main land types of MianYang,We divided MianYang's landscape types into five types: forest,farmland,construction land,water,others. Then using the landscape software for various types of FRAGSTATS3.3tocompute the landscape indexes of the year2002and2007.Based on these results, we analyze thedynamic changes of the landscape pattern and predict the landscape pattern's changing trend of2012with the Markov model. The main conclusions are as follows:(1)The processing work of the MianYang' remote sensing images mainly contains: first,rectifying the remote sensing images of MianYang, mainly contains Radiometric Correction andGeometric Correction; and then join the rectified remote sensing images together, in the softwareErdas imagine9.2, join the remote sensing images together by first configuring the centre imageand then make them together according to the sequence and the actual operation of the remotedata. And then using the AOI that had been made to cutting the remote sensing images, so that wecan get the images that can be classified. Based on the steps that had been taken and learningfrom the results ahead, we choose the fourth,third and the second band as the best bands. Inorder to totally improve the sharpness of the remote sensing images, we take haze reduction onthe images. We use the Maximum Likelihood Classifier to classify the images, and then use theclump,eliminate and the recode to process the classified images. And finally get the remotesensing images which their kappa are0.84and0.85, both can be actually used.(2)The analyzing results of the landscape pattern indexes show that the basic land type ofMianYang is forest both in2002and2007.The sequence of the landscape patterns'PLAND is notchanged; the total majority of the patches have increased110in5years, the trend of the patchdensity is rising, the overall fragmentation has increased; The patch shape is becoming more andmore ruled, but the changing degree is not fierce, the influence of human beings is getting strong.(3)The dynamic changes analysis of the landscape pattern shows that the changing degree ofthe forest is the most fierce, mainly source from farmland,construction land and others; the areaof farmland has decreased640.13km~2,the area of construction land has increased28.69km~2,thearea of water has increased47.25km~2,the area of others has decreased524.25km~2. Thechanging among forest and the other four landscape patterns is fierce.Because the government'sinitiative of planting more trees which is in order to improve the quality of the soil which wassalinity-alkalinity, all the measures that mentioned aboved are due to the traits of MianYang city'sland using properties which are the area of mountainous region and hills are huge, the poor land takes large part of the total farmland. Reusing the land which was part of the hilly land that canbe used for agriculture uses. Properly adjusting the space structure of the farmland and themountainous region to take advantage of the land better. Between2002and2007these five years,the influence of human beings is getting stronger, the construction land area is increasing and thefarmland area is deducing. The changing between those two types is a bit fierce.(4)The predicting result using the Markov model shows that the percentage of constructionland has increased0.08%comparing to2007's, the changing degree is not big, it has taken upsome area of the farmland which causes the deduce of the farmland area. The percentage of theforest has risen0.55%. The percentage of the farmland has decreased0.66%, mainly changedinto construction land and forest. The trend of the other is deducing. The area of the water hasincreased.We concluded some suggestions for the current problems that exist in MianYang's landscapepattern, hoping can supply referring advices for its improving development. |
PDF Full Text Request