The influence of forest tree species composition on the forest height predicted from airborne laser scanning data – A case study in Latvia

Abstract

Airborne laser scanning (ALS) is used to predict different forest inventory parameters; however, the ALS point cloud properties depend on various parameters such as the type of ALS scanner employed, flight altitude and scanning angle, forest stand structure, forest tree species composition, vegetation season, etc. This study used national coverage high-resolution ALS data with minimum point density of 4 points per square meter in combination with field data from the National Forest Inventory (NFI) to build forest stand height models for forest stands dominated by 6 most common tree species in Latvian mixed forest stands, viz. Pinus sylvestris L., Betula pendula Roth, Picea abies (L.) Karst., Populus tremula L., Alnus incana (L.) Moench and Alnus glutinosa (L.) Gaertn. for the various ALS scanners employed and at different growing seasons. The selected NFI plots are divided into modelling and validation datasets in a ratio of 3 : 1. The results show that for a universal forest stand height model, the RMSE value is 1.91 m and the MAE is 1.41 m. For the forest stand height models, which are stratified by scanner, individual tree species and seasons, the RMSE value is within the limits of 1.4 m for forest stands dominated by Scots pine in leaf-on canopy condition to 3.8 m for birch in leaf-off canopy condition.

Key words: forest inventory, airborne laser scanning, phenology, large scale forest mapping