Study On Vegetation Dynamics Monitoring In Li River Basin Based On Remote Sensing Technology

For Li River basin, especially the forest vegetation is not only producers and defenders of the good ecological environment and tourism resources,but also defenders of runoff. There is a necessity to study on the the dynamic monitoring of vegetation growth in the whole Li River, in order to better protect advantageous ecosystems and environmental conditions along Li River. As is well known to us, it is optimal to use remote sensing for dynamic monitoring of vegetation. To calculate spectral characteristics of different vegetation types in the three kinds of remote sensing image, seven kinds of vegetation index, principal component analysis, MNF transform, asselled cap transformation(Only to the TM image) were needed. Gray-level co-occurrence matrix were applied to extract for texture characteristic in ALOS and ZY-3images. Decision tree models of three kinds of images were established on spectral characteristics,texture characteristic and data on the spot. Vegetation information of Li River was extracted in1986,1998,2007and2012. Land transfer matrix has been established to analyze changes of vegetation dynamic in Li River for nearly30years. The main research conclusions are as follows:(1) By analyzing the three data sources in the decision tree model, we can draw conclusions:Greenness Index obtained from Tasselled cap transformation of TM can effectively distinguish vegetation area with non-vegetation area, which is the same to ALOS and ZY-3in the range of green light,in Li River basin.(2) All the vegetation information extraction from the three data sources is good. In particular, ZY-3has a total classification accuracy of91.96%and Kappa coefficient of0.9029, which shows that on the basis of spectral features and texture features, using the decision tree algorithm to select the appropriate vegetation information extraction is feasible.(3) Different approaches has their own advantages, ZY-3is best suited to coniferous forest and broad-leaved forest, ALOS is best for shrubbery and the others two plants,and TM is best for bamboo forest.(4) Along Li River,the four kinds of the main vegetation, Shrub forest, broad-leaved forest, coniferous forest to calculate spectral characteristics of different vegetation types in the three kinds of remote sensing image and Bamboo, have increased from1986to2012 In the same area and during the same time, Cultivated land, construction land, bare land,5kinds of the main vegetation types, and water have decreased;In Li River, shrub area was cut down from1986to1998,but grew sharply from1998to2007,and kept stable from2007to2012;The broad-leaved forests area increased from1986to1998,kept stable from1998to2007and decreased slightly from2007to2012;The coniferous forests area kept increasing from1986to2007,then began to decrase from2007to2012; Others increased sharply from1986to1998, decreased sharply from1998to2012.