CMG Collaborative Research: Structures of Uncertainty in Coupled Phys-Bio-Biological EstimStructures of Uncertainty in Coupled Physical-Biological Estimation in the Coastal Ocean

摘要这项研究针对沿海海洋中的现象，提出了一种直接的因果关系和推理结构。它将集中在生物和物理现象之间的界面，其中因果结构是单向的，物理通过平流、温度、盐、光和湍流影响生物。物理不确定性向生物现象的转移是首要关注的问题。将对这种转移及其对状态和参数估计的影响进行严格的表述。这些例子将来自当代观测方案和模拟的经验。由于其巨大的复杂性，这些特定于地点的耦合问题倾向于仔细地制定与并行海洋测量计划相兼容的问题。欧拉模型和拉格朗日模型(基于个体)都将被使用，在空间明确的环境中运行。算法设计中的集合(蒙特卡罗)方法和最大似然估计方法将适应基本的非线性，而更简单的最小二乘方法是不合理的。环境和自然资源管理的一般问题现在正在全球范围内得到重视，特别是在沿海地区。在那里，交通、石油和天然气、海洋渔业、娱乐和废物处理日益占据主导地位；美国为这些资源和活动制定了许多公共管理要求。这项工作解决了当代美国沿海海洋问题，这些问题本质上是多学科的。拟议的合作结合了概率统计、海洋科学和计算科学方面的专业知识。所选择的问题是这些学科之间的接口的最前沿。正如海洋中所呈现的那样，这些问题既大又复杂。需要在它们的界面上取得进展，其中可行的计算算法、合理的数学公式和有用的科学简化以智力上持久的方式合作。PI在他们的学科和界面上都有自己的地位。与正在进行的主要沿海管理方案相关的进展将在各协作学科之间取得平衡，并侧重于跨学科的沿海海洋。

AbstractThis investigation addresses phenomena in the Coastal Ocean that present a directed structure of causation and inference. It will focus on the interface between biological and physical phenomena, where the causal structure is one-way, physics affecting biology via advection, temperature, salt, light, and turbulence. The transfer of physical uncertainty toward biological phenomena is of primary concern. Rigorous representations of that transfer, and its implications for state and parameter estimation, will be developed. Examples will be drawn from contemporary experience with observational programs and simulations. Due to their huge complexity, these site-specific, coupled problems favor careful problem formulation which is compatible with concurrent oceanographic measurement programs. Both Eulerian and Lagrangian (Individual-Based) models will be used, operating in spatially-explicit environments. Essential nonlinearities will be accommodated with ensemble (Monte Carlo) methods in algorithm design; and with maximum likelihood estimation where simpler least-squares approaches are not justified. General problems of Environmental and Natural Resource Management are now being appreciated globally, and particularly in the coastal ocean. Access to transportation, oil and gas, marine fisheries, recreation, and waste disposal are increasingly dominant there; and the USA has created numerous public management requirements for those resources and activities. This work addresses contemporary US coastal ocean problems that are multi-disciplinary in nature. The proposed collaboration unites expertise in Probability and Statistics, Ocean Science, and Computational Science. The problems selected are at the forefront of the interfaces among these disciplines. As presented in the ocean, these problems are big and complex. Progress at their interface is needed in which workable computational algorithms, sound mathematical formulations, and useful scientific reduction collaborate in intellectually lasting ways. The PI's have standing in their disciplines and at their interfaces. Progress of relevance to major ongoing coastal management programs will be balanced across the collaborating disciplines and focused on the interdisciplinary Coastal Ocean.