Search

The site quality evaluation of plantations has consistently been the focus in matching tree species with sites. This paper studied the site quality of Chinese fir (Cunninghamia lanceolata) plantations in Lin'an District, Zhejiang Province, China. The site quality model was constructed using the algebraic difference approach (ADA) to classify the site quality grades. The rough set algorithm was used to screen out the key site factors affecting the site rank of Chinese fir plantations. Site quality classification models based on random forest were established, and the importance of key site factors was evaluated. The results are as follows. The random forest model based on the rough set algorithm had small scale and low complexity, and the training and testing accuracies of the model were 92.47% and 78.46%, respectively, which were better than the model without attribute reduction. The most important factors affecting Chinese fir growth in the study area were the slope aspect, slope grade, and canopy closure. The least important factors were the humus layer thickness, soil layer thickness, naturalness, and stand origin. The attribute reduction method proposed in this study overcame the subjectivity of traditional site factor selection, and the site quality classification model constructed improved the model accuracy and reduced the complexity of the algorithm. The methods used in this study can be extended to other tree species to provide a basis for matching tree species with sites and to improve the level of forest management in the future. [ABSTRACT FROM AUTHOR]

The site quality evaluation of plantations has consistently been the focus in matching tree species with sites. This paper studied the site quality of Chinese fir (Cunninghamia lanceolata) plantations in Lin’an District, Zhejiang Province, China. The site quality model was constructed using the algebraic difference approach (ADA) to classify the site quality grades. The rough set algorithm was used to screen out the key site factors affecting the site rank of Chinese fir plantations. Site quality classification models based on random forest were established, and the importance of key site factors was evaluated. The results are as follows. The random forest model based on the rough set algorithm had small scale and low complexity, and the training and testing accuracies of the model were 92.47% and 78.46%, respectively, which were better than the model without attribute reduction. The most important factors affecting Chinese fir growth in the study area were the slope aspect, slope grade, and canopy closure. The least important factors were the humus layer thickness, soil layer thickness, naturalness, and stand origin. The attribute reduction method proposed in this study overcame the subjectivity of traditional site factor selection, and the site quality classification model constructed improved the model accuracy and reduced the complexity of the algorithm. The methods used in this study can be extended to other tree species to provide a basis for matching tree species with sites and to improve the level of forest management in the future.