Muhammad Junaid Ismail ¹ Tika Ram Poudel ² Akber Ali ¹ Lingbo Dong ^1*

• ¹ Key Laboratory of Sustainable Management of Forest Ecosystem, Northeast Forestry University, Harbin, China
• ² Feline Research Center of National Forestry and Grassland Administration, College of Wildlife and Protected Area, Northeast Forestry University,, Harbin, China

Tree attributes, such as height (H) and diameter at breast height (D), are essential for predicting forest characteristics, developing yield models, and serving as fundamental indicators of forest productivity and health. Measuring tree H is particularly challenging in uneven-aged forests compared to D. In this study, we collected data from 7,069 Larix gmelinii trees, sampled across 96 plots representing a wide range of stand densities, ages and site characteristics in Northeast China. To describe the H-D relationship, we evaluated fifteen nonlinear functions known for their capacity to model this relationship effectively. Model performance was assessed using root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R² ). Results identified the Ratkowsky model (M8) as the best performer, achieving the highest R² (0.74), the lowest RMSE (16.47%) and MAE (12.50%), and statistically significant regression coefficients (p < 0.05). Furthermore, the M8 model showed performance with the inclusion of stand-level variables, requiring no additional measurements, achieving an R² of 0.82% and an RMSE of 13.7%. Our findings suggest that incorporating stand-level variables representing plot-specific characteristics can further improve the fit of mixed-effects models. These advancements provide forest authorities with enhanced tools for supporting sustainable forest management.