Enabling stochastic programming for improved forest planning under uncertainty: moving from theory to practice

Forest planning faces many uncertainties, yet existing decision support systems (DSS) seldom incorporate techniques to address them. This study explores how stochastic programming (SP) functionality could be added to forest DSSs to account for data uncertainty, aiming to investigate the added value of such functionality. An SP model was applied to a traditional long-term forest planning problem, and its quantitative performance was compared to deterministic optimisation. The user value of the DSS integration was explored in a workshop with potential users. The findings indicate that incorporating SP in a DSS is feasible both user-wise and for quantitatively improving the decisions, even if computational time and model complexity increase. Quantitatively, SP increased the total expected NPV by at least 2% compared to deterministic optimisation. Users involved in evaluating the SP integration acknowledged the benefits of using SP in forest planning, but expressed concerns about the increased complexity in problem specification and results interpretation. To enhance user adoption, the presentation of SP settings and outcomes should be done in a user-friendly manner, including intuitive visualisations and simplified summary statistics. This study underscores the potential of integrating SP in DSSs to improve long-term forest planning under data uncertainty.