[Remote sensing estimation of urban forest carbon stocks based on QuickBird images]

Urban forest is one of the positive factors that increase urban carbon sequestration, which makes great contribution to the global carbon cycle. Based on the high spatial resolution imagery of QuickBird in the study area within the ring road in Yiwu, Zhejiang, the forests in the area were divided into four types, i. e., park-forest, shelter-forest, company-forest and others. With the carbon stock from sample plot as dependent variable, at the significance level of 0.01, the stepwise linear regression method was used to select independent variables from 50 factors such as band grayscale values, vegetation index, texture information and so on. Finally, the remote sensing based forest carbon stock estimation models for the four types of forest were established. The estimation accuracies for all the models were around 70%, with the total carbon reserve of each forest type in the area being estimated as 3623. 80, 5245.78, 5284.84, 5343.65 t, respectively. From the carbon density map, it was found that the carbon reserves were mainly in the range of 25-35 t · hm(-2). In the future, urban forest planners could further improve the ability of forest carbon sequestration through afforestation and interplanting of trees and low shrubs.