Deep learning of Ecosystem Service Indicator Maps for Agile and Precise forest planning

Simulation-optimization computations for forest management planning face challenges in terms of information provided by modern
forest inventories. Current planning and decision support systems cannot fully utilize the vast number of decision alternatives formed by
small spatial units, simulated alternatives and their complex spatiotemporal interactions. We develop intelligence in the simulationoptimization-
framework by adding a component that learns from the uncertainties in the data and successes of previously solved forest
planning problems. The developments based on novel deep learning methods are expected to streamline the simulation-optimizationframework
with reduced time complexity to allow Agile & Precise planning based on a very high number of computation units. The
developments promote the effectiveness of planning of sustainable forest management by enabling a more detailed, optimal allocation
of forest resources to produce various forest ecosystem services in parallel.