Research On Remote Sensing Based Estimation Of Forest Carbon Stock In Shaoguan,Guangdong Province

With global climate change, the world is becoming more and more attention to environmental problems. As the world’s largest forest carbon stocks, the study on it is of great significance, and to study on the forest carbon stocks has become the current hot topic. In this paper, we taking Shaoguan area as the study area, select the elevation, slope, aspect, greenness index, humidity index, normalized difference vegetation index, ratio vegetation index of seven independent variable factors and sample unit area of carbon(carbon density) modeling. Use four different methods: multiple linear regression, K-nearest neighbor classification algorithm, artificial neural network algorithm and random forest algorithm which is very popular currently but rarely applied to forestry to build the model of carbon reserves estimation. Compare the accuracies of the 4 models, and then analysis the carbon storage image in Shaoguan with hot spot detection. And then use the model with the highest accuracy for the 2002 and 2012 Shaoguan carbon reserves. Compared with 2002, 2007, 2012 three data, analysis their variation tendency, the results show that:(1) by comparing the accuracy of four models we find that: the random forest algorithm precision is highest, the multiple linear regression is the lowest.(2) the distribution of the points with higher carbon stocks are at a higher altitude and steep places, the distribution of the points with lower carbon stocks are at a lower altitude and steep places. Higher carbon storage places mostly are forest parks, nature reserves; low carbon storage places are located in the town(3) 2002-2007, the carbon storage of Shaoguan area was reduced by 17.8%, the average carbon density declined, spatial aggregation weakened, and broken trend strengthened; from 2007 to 2012, carbon storage increased by 36%,the average carbon density increased, the spatial aggregation enhanced and the fragmentation trend decreases. The points of carbon stocks which had dramatic changes are located around cities and towns.The changes of carbon stocks are closely linked with the city development and policy.