This project aimed to analyse the changes of tropical forests, and at first it summarised the trend of deforestation and restoration, then analysed the changes of forest structure and function from the view of forest patches, at last clustered and classified the different forest ecosystems for similarity analysis.
Trend Analysis
Line chart is commonly used method for trend analysis. This project used this graph to count the areas of different forests for years for 1960, 1979, 1986, 2000, 2005, 2010 with R. Then separated the whole changing process into two stages: deforestation stage and restoration stage. At last, combining the land use history of study area, I analysed human influence on forest changes in these two stages.
Difference Analysis
First of all, I collected and calculated five landscape indices (patch area, shape index, contiguity index, core area index and Euclidean nearest neighbour index) in patch level and then transformed data for each group (24 subsets) to test normality and homogeneity of variance. In the following step, I compared the differences of landscape indices between different groups used Analysis of Variance (ANOVA).
Similarity Analysis
Firstly, Principle Component Analysis (PCA) was used to reduce the
dimensions of variables (the selected landscape indices in class-level), and in
this case, 11 variables were reduced to 5 principle components to describe each
forest ecosystem in a specific year. After that, I used clustering with
distance based on the principle components. 24 communities (subsets) could be
clustered and ordered with respect to the similarities of forest function and
spatial structure.