Mathematical Biology Seminar



Tuesday, 01/29/2019, at 4:10 - 5:00








Development of next-generation forest biogeochemistry models



Nikolay Strigul



Associate Professor



Department of Mathematics and Statistics, WSU-Vancouver










Forest ecosystem models often reflect the dominant management paradigm of their time. Unlike early models focused on timber production, the current generation of models are primarily designed to inform ecological forest management. State-of-the-art forest biogeochemistry models represent complex software systems and are typically analytically intractable. Intensive computer simulations are needed to produce predictions. However, different models typically produce distinct results, despite structural similarity, while predictions from the same model may vary depending on the parameter values. Over the past two decades, a new generation of terrestrial biosphere model has blended principles from physical ‘big-leaf’ models and individual-based ‘gap’ models into new hybrid models. This was done to reduce biogeochemical uncertainties in forests through explicit inclusion of vegetation dynamics. Parallel research on the ED model approximated the first moment of a gap model using systems of partial differential equations. A more powerful model reduction technique followed in the Perfect Plasticity Approximation, or PPA. Inspired by the TASS tessellation scheme, this work showed that adding phototropism and crown plasticity to the SORTIE gap model reduced variation in canopy join height to a negligible level. This observation allowed the reduction of individual-based dynamics to cohort-level interactions, improving computational efficiency and providing analytical tractability. We aim to extend SORTIE-PPA by adding mechanistic representations of energetic and biogeochemical fluxes, and landscape disturbance dynamics, using hierarchical multiscale modeling, into the next-generation SORTIE-NG model. The model incorporates evolutionary optimality principles and allows for variation elsewhere in the trait space to represent phenotype plasticity and intraspecific genetic diversity. In doing so, we aim to bridge the divide between big-leaf, gap, and forest landscape models by integrating principles from each, focusing on salient dynamics for continental-scale biogeochemistry, demography, and diversity. Planned extensions to SORTIE-NG include the first directly incorporated machine-learning processes, creating a new class of pattern-based hybrid models. This is the joint work with Dr. Adam Erickson.