Modelling single-tree diameter increment of Scots pine and Norway spruce using multi-source tree, stand, soil, and climatic data

In light of the observed changing conditions of forest growth, it is important to develop up-to-date forest growth models at the level of entire ecosystems, individual stands, and single trees. The aim of the study was to determine how multi-source variables explain the radial growth of Pinus Sylvestris and Norway spruce trees. Models for Scots Pine were developed based on over 100 000 annual observations of tree ring width from 6749 trees across 301 plots located in Poland, and for Norway Spruce, based on over 15 000 observations, 1090 trees across 57 plots. The analyzed period spanned from 2005 to 2022. Data from airborne laser scanning was used to calculate tree height, various competition indices, crown volume, and relative sunlight exposure. Other variables included: diameter at breast height, stand stock and age, climatic variables describing vegetation season and soil characteristics. Random forest models were developed. Cross-validation was applied, selecting one year of observations and random 20% of plots as test data per iteration. Models trained on all data reached around 70% explained variability, however after cross-validation explained variability dropped to around 18% and 7% for tree ring width of Scots Pine and Norway Spruce respectively. Despite the use of an extensive set of explanatory variables, big part of the variability in growth was determined by other factors and remained unexplained. The obtained results indicate the need for further research in the field of modeling diameter growth of individual trees.