The forest is a major carbon sink in the terrestrial ecosystem, accounting for nearly 72% of terrestrial carbon storage in woody biomass and soil. Determining the amount of biomass in a forest stand is necessary for property managers to make informed decisions about the value and use of their forest land. The aim of this study is to create machine learning models for forest biomass determination based on NFI plot data, ALS data with a minimum point density of 4 points per square meter and other freely available cartographic materials. Forest biomass models have been developed for parameters such as above-ground biomass (AGB), below-ground biomass (BGB), trunk biomass (SB), branch biomass (BB) and stump-root biomass (SRB). The dataset created from NFI, ALS and other datasets is randomly divided into training dataset (80% of data) and 20% is left for model validation. The coefficient of determination for all developed models ranges from 0.76 to 0.85, and RMSE values from 1.44 (SRB) to 35.05 (AGB)

Janis Ivanovs, Andis Lazdins
Latvian State Forest Research Institute Silava

---

 
ID Abstract: 58