Statistics for Dynamic Models with Application in the Marine Sciences

成功的环境预测需要有效地利用模拟模型和现有的观测资料。统计建模提供了一种结合这两种信息来源以及其他科学知识的方法，以便尽可能多地了解所考虑的系统。这个过程有时被称为数据同化，是基于贝叶斯统计原理。主要应用领域是天气、气候和海洋预报。必须制定和应用特殊的估算方法，因为随着科学和计算技术的进步被纳入模拟代码，复杂的环境模型正在不断完善。观测变得越来越复杂，包括监测时间序列、卫星空间图像以及探测和自适应穿越海洋深度的复杂自主仪器。数据和模型的融合，使我们有希望进行更熟练的环境预测，沿着对这些系统有更好的科学理解。我建议的工作涉及进一步发展先进的数据同化，使用最先进的统计方法，使我们朝着这一目标前进。

Successful environmental prediction requires making effective use of simulation models and available observations. Statistical modeling provides a means of combining these two sources of information, as well as other scientific knowledge, in order to learn as much as possible about the system under consideration. This procedure is sometimes referred to as data assimilation, and is based on Bayesian statistical principles. Key application areas are in weather, climate, and marine forecasting. Special estimation methods must be developed and applied since complex environmental models are constantly being refined as scientific and computational advances are incorporated in the simulation code. Observations have become increasingly sophisticated ranging from monitoring time series, spatial imagery from satellites, and complex autonomous instruments that probe and adaptively transit the ocean depths. In the fusion of data and model lies the promise of more skillful environmental prediction, along with better scientific understanding of these systems. My proposed work deals with further development of advanced data assimilation that uses state-of-the-art statistical methods to move us towards this goal.