Climate Smart Forecasting and Robust Optimization of Forest Management System under Uncertainty

对更多商品和服务的需求不断增加，与全球变化和自然干扰相互作用，从而增加了对森林资源的压力。另一方面，气候变化的程度和森林系统进程的结果存在固有的不确定性。因此，必须保护生态系统的完整性，以应对全球变化的影响以及相关的风险和不确定性。这对于为继续改进森林系统管理提供坚实基础至关重要。这项研究建议将建立一个独特的加拿大研究小组，以实现以下项目目标：1）通过开发配备生态系统服务指标的最先进的森林过程气候智能预测系统，分析加拿大森林系统，模拟人为气候变化和管理干预对森林恢复力的预期影响，并评估森林对社会的贡献（2）通过量化对森林扰动和气候变化影响的长期预测的不确定性，评估加拿大森林及其服务的复原力和复原力驱动因素;（3）制定决策分析方法，为加拿大森林适应不确定的未来条件和情景找到最佳和稳健的解决方案。该项目采用了广泛使用的基于过程的森林系统模型3 PG（生理过程预测）欧洲申请人开发的一个版本，用于预测生态系统货物的动态（如木材）和服务（如碳、水、生物多样性），以应对详细的管理干预措施（间伐、施肥、收获、栖息地树木保留、气候变化（政府间气候变化专门委员会）情景）和自然扰动（例如风、火、干旱、昆虫）。申请人特别采用贝叶斯推理系统，以同时校准3 PG以适应特定的现场条件（例如物种、管理、土壤、生产力），并分析预测不确定性（即多个情景的深度不确定性和模型参数的随机性）。这些不确定性将单独或组合地传播到森林决策的最终绩效。最后，先进的优化技术（遗传算法）将被应用于找到最佳的多目标鲁棒解决方案。将采用折衷规划方法（例如后悔最小化和最大加权风险价值）来处理多个目标和风险的期望不确定性。这项研究将培养3名博士，6名硕士和9名理学士学生，以支持加拿大林业基于过程的森林系统决策，以应对加拿大森林未来的预期和不确定变化，例如森林生长，大规模干扰，并采用多目标和稳健的森林管理策略。

The increasing demand for more goods and services interacts with global change and natural disturbances and, consequently, increases the pressure on forest resources. On the other hand, there is inherent uncertainty about the degree of changes in climate and the outcomes of forest system processes. Therefore, the integrity of ecosystems must be safeguarded in response to the impacts of global change and the associated risks and the uncertainties. This is crucial to provide a solid base for continued improvement in forest system management. This research proposal will establish a unique Canadian research team to pursue the following project objectives: 1)Analyze Canadian forest systems by developing state-of-the-art Climate Smart Forecasting system of forest processes equipped with the indicators of ecosystem services and simulate the expected effects of anthropogenic climate change and management interventions on forest resilience and evaluate forests contribution to the society (forest system analysis); 2)Evaluate resilience and resilience drivers of Canadian forests and their services by quantifying the uncertainty of forecasts over time regarding forest disturbances and the impacts of climate change; and 3)Develop decision-analysis approaches to find optimal and robust solutions for the adaptation of Canadian forests to uncertain future conditions and scenarios. The project applies a widely used process-based forest system model 3PG (Physiological Processes Predicting) a version developed by the applicant in Europe to predict the dynamics of ecosystem goods (e.g. timber) and services (e.g. Carbon, Water, Biodiversity) in response to detailed management interventions (thinning, fertilization, harvesting, habitat tree retention, regeneration) and global change (IPCC (Intergovernmental Panel on Climate Change) scenarios) and natural disturbances (e.g. wind, fire, drought, insects). A Bayesian inference system is specifically adopted by the applicant to simultaneously calibrate 3PG to the specific site conditions (e.g. species, management, soil, productivity) and analyze forecasting uncertainty (i.e. deep uncertainty of multiple scenarios and model-parameters' stochasticity). These uncertainties will be propagated, solely and in combination, to the final performance of forest decisions. Finally, advanced optimization techniques (genetic algorithms) will be applied to find the optimal multi-objective robust solutions. Compromise programming approaches (e.g. a regret minimization and a maximum weighted Value-at-Risk) will be applied to deal with the expectations' uncertainty of multiple objectives and risks. This research will train 3PhD, 6 MSc, and 9 BSc students to support process-based forest system decisions of Canadian forest industry to the expected and uncertain changes in Canadian forest futures e.g. in forest growth, large scale disturbances, and adopting many-objective and robust forest management strategies.