Construction Of Height To Crown Base Of Larix Olgensis Based On Mixed Effect And Quantile Regression

Height to crown base is an important index to reflect the characteristics of the tree crown.It is of great significance to accurately predict height to crown base for forest management and increasing stand production.This study was based on data from 323 fixed plots of Larix olgensis plantations in Heilongjiang Province.Among the five commonly used height to crown base foundation models,the one with the best fitting effect was selected.Individual tree factor,stand factor and competition factor were introduced into the basic model to construct the generalized height to crown base model.Then,the mixed effect model and quantile regression model were further extended.The prediction ability of the models was evaluated and compared by the ‘leave one out’ cross-validate.In addition,four sampling designs and different sampling sizes were used to correct the height to crown base model.Select the sampling scheme which can take into account the model prediction accuracy and sampling cost.The conclusion and significance of this study are as follows:(1)The Logistic equation had high prediction accuracy and biological significance,and was selected as the height to crown base’s basa model of Larix olgensis.With the introduction of the basal area of the stand and the average dominant height as covariables,the generalized height to crown base model can achieve the best fitting effect.(2)Considering the random influence of plot on height to crown base,the generalized height to crown base model was used to construct a single-level mixed effect model of height to crown base of Larix olgensis.The mixed effect model produced better fit and test results than the generalized height to crown base model.(3)We fitted a quantile regression model with nine different quantiles using the generalized model of height to crown base as the primary baseline.The best fitting was obtained when the quantile was 0.5 which was not different from the generalized model.Three nonlinear quantile height to crown base prediction models were constructed from the combination of 3 quantiles,5quantiles,and 9 quantiles.Among them,three-quartile regression model’s mean absolute error and mean absolute percentage error were the lowest.(4)In order to improve the prediction accuracy of mixed effect model and quantile regression model,four different sampling designs were used to correct the two models,The sample size ranges from 1 to 18 strains.Scheme Ⅳ(drawing average trees)was obtained as the best sampling correction scheme.Considering the prediction accuracy and sampling cost of the model,the sample size of 5 trees is suitable.(5)The newly developed Larix olgensis height to crown base models had good prediction ability.The corrected mixed effect model has slightly better prediction accuracy than the other models.The results of this research provided a basis for accurately predicting the height to crown base of Larix olgensis in Heilongjiang Province.