Objectives The tree height-diameter model is a highly efficacious instrument in forest management. The purpose of this article was to establish the tree height-diameter model at the stand level in East China using the nonlinear mixed-effects method and provide an available prediction tool for forest stand tree height at a large regional scale. The final model was compared with the traditional nonlinear regression model to explore the performance of the nonlinear mixed-effects method in developing the tree height-diameter model.

Methods The nonlinear mixed-effects method was employed to construct a stand-level tree height-diameter model for the Each China region, utilizing data from the national forest resources inventory of seven provinces in the region. This involved the creation of dummy variables for the predominant tree species.

Results The nonlinear mixed-effects method performed well on both the training and testing sets. The performance of the training set of the final model in terms of R², root mean square error (RMSE), and mean absolute error (MAE) is 0.6785, 2.3967 m, and 1.8326 m, respectively. The performance of the validation set of the final model in terms of R², RMSE and MAE is 0.6738, 2.3520 m and 1.7985 m, respectively. Compared with traditional regression methods, the mixed-effects model demonstrated an 8.13% increase in R² within the training set, accompanied by a 7.09% reduction in MAE and a 6.87% reduction in RMSE. In the validation set, R² exhibited a 7.75% increase, while RMSE and MAE exhibited a 6.61% and 5.97% reduction, respectively.

Conclusions The stand-level tree height-diameter model for the main tree species in East China was established using the nonlinear mixed effects method. The final model demonstrated superior fitting and prediction accuracy compared to the traditional regression model. The model has a relatively wide applicable coverage area, which can provide a basis for forest management practices.