Non-linear scenic beauty model of scenic Platycladus orientalis plantations based on in-forest color patches

We applied the theory of landscape spatial pattern analysis in landscape ecology to evaluate the quality of landscape by analyzing the structure of color patches in photos. The relationships between SBE grades and 14 patch indexes were studied.Six indexes among them, i.e.,SPLIT, SHDI, DIV, SIEI, ACOV and LPI,which influence scenic degree significantly, were selected by using correlation analysis. Results showed that the relationships between DIV, SIEI, ACOV, LPI and SBE grades were in multiple ways rather than merely in a linear way; SPLIT and SHDI changed in a positively linear trend with the decrease of SBE grade.Based on the six indexes, the scenic Platycladus orientalis plantations could be classified into four types by using cluster analysis. The indexes were used as the dependent variables to establish linear and non-linear models. Comparison showed that fitting accuracy of the non-linear model was higher than that of the linear model. The relationships between SBE value and landscape index could be well reflected by quadratic polynomial regression models.