Bi-temporal LiDAR-based estimates of stand variables reduce growth modelling costs in even-aged Larix olgensis A. Henry plantations

Key message We developed a dynamic stand-level growth and yield model using bi-temporal UAV LiDAR data to support forest management and planning. This approach aims to reduce the cost and intensity of ground-based inventories by using low-cost UAV LiDAR as an alternative data source. Context Growth and yield models, as the principal tools for simulating and forecasting stand development dynamics, support sustainable long-term planning of forest resource management. Aims This study developed a dynamic growth model for even-aged tree stands based on bi-temporal field measurements and UAV LiDAR data, using Larix olgensis A. Henry—a common afforestation species in Northeast China—as the focal species. Methods The field measurements and the LiDAR scans were done in 2019 and 2023. We evaluated the suitability of using LiDAR-based estimates for stand variables (stand age, dominant height, basal area and number of trees per hectare) as modeling data. Models based on field measurements provided the benchmark. The growth and yield models consisted of three simultaneously fitted equations: the first for the dominant height, the second for the number of trees per hectare, and the third for the stand basal area. The LiDAR-predicted values of the variables were obtained by using the k‑nearest‑neighbors (kNN) imputation method. Models based on LiDAR-estimated stand variables were very similar to those based on field measurements. The models were used to simulate stand development from 2019 to 2023, using either field data or LiDAR-based variables as predictors. Results The simulation results for dominant height were the best when both the models and the inventory data consisted of LiDAR-based estimates. For the stand basal area and number of trees, the field inventory yielded better simulation results than using LiDAR estimates in the area-based inventory approach. Conclusion The results indicated that predicted LiDAR-based variables can be used to develop stand-level growth and yield models for plantation forests, in addition to using LiDAR in forest inventory.