Classification and Quality Assessment of Individual Tree in Castanopsis carlesii Broad-leaved Forest in Mid-subtropical Zone

In 2022,the Castanopsis carlesii community in mid-subtropical zone was selected as the research object at the state-owned forest farm in Youxi County,central Fujian Province.By constructing a system of 11 indicators including DBH,tree height,crown width,live canopy ratio,crown dieback and crown density,fuzzy comprehensive evaluation was used to classify the quality of individual trees.Linear discriminant analysis and K-nearest neighbor algorithm were used to establish a prediction model for the classification results,and 10 fold cross validation was conducted to evaluate the accuracy.The results showed that in the single tree quality classification results,the proportion of trees with four evaluation levels of poor,medium,good,and excellent was 22.20%,30.19%,30.45% and 17.16%,respectively.The overall quality level was mainly good and medium.The cross validation analysis showed that the error rate and accuracy of the linear discriminant prediction model were 14.18% and 85.82%,respectively.The error rate and accuracy of the K-nearest neighbor algorithm were 11.84% and 88.16%,respectively.The confusion matrix analysis linear discrimination error mainly occurs between medium and excellent,while the K-nearest neighbor algorithm mainly occurs between medium and good.The K-nearest neighbor algorithm outperformed linear discriminant analysis in terms of accuracy,average precision,and recall.Therefore,the individual tree quality prediction model established by the K-nearest neighbor algorithm can quickly evaluate the single tree quality of the Castanopsis carlesii broad-leaved forest in the study area while ensuring a certain prediction accuracy.