Current issue: 54(5)
Under compilation: 55(1)
Despite the numerous studies on year-to-year variation of tree growth, the physiological mechanisms controlling annual variation in growth are still not understood in detail. We studied the applicability of data-driven approach i.e. different regression models in analysing high-dimensional data set including continuous and comprehensive measurements over meteorology, ecosystem-scale water and carbon fluxes and the annual variation in the growth of app. 50-year-old Scots pine stand in southern Finland. Even though our dataset covered only 16 years, it is the most extensive collection of interactions between a Scots pine ecosystem and atmosphere. The analysis revealed that height growth was favoured by high water potential of the tree and carbon gain during the bud forming period and high water potential during the elongation period. Diameter growth seemed to be favoured by a winter with high precipitation and deep snow cover and a spring with high carbon gain. The obtained models had low generalization performance and they would require more evaluation and iterative validation to achieve credibility perhaps as a mixture of data-driven and first principle modeling approaches.