Collaborative Research: A Computational Framework for Assessing the Observation Impact in Air Quality Forecasting

该奖项是根据2009年美国复苏和再投资法案(公法111-5)资助的。这项研究开发了对空气质量模拟中的观测影响进行明智评估所需的算法和计算框架。模型约束优化框架中的新算法将允许考虑数据在时空域中的位置、观测类型、仪器类型以及在存在多个观测系统的情况下的数据交互。具体而言，研究的重点是：开发与四维变分数据同化方案一致的基于高阶伴随的观测影响技术；评估预报对观测误差方差的敏感性，并估计输入误差统计规格中的不确定性的预报影响；开发有效的计算方法，以便能够实际实施观测影响算法；通过观测系统实验和评估新观测系统的潜在影响来验证新技术。准确描述与各种人为活动有关的大气污染物分布的能力对于化学天气预报以保护人口、回答与我们星球的未来有关的科学问题以及制定合理的环境政策至关重要。要准确地表示大气的化学成分，需要通过数据同化将模式和观测紧密地结合在一起。数据同化是模式预测利用测量来产生大气状态的最佳表示的过程。随着越来越多的观测资料可用，并正在规划新的测量网络，至关重要的是发展最佳利用数据的能力，更好地管理遥感资源，并设计更有效的实地实验和网络，以支持大气化学和空气质量研究。这项研究开发了必要的计算工具，以优化现有观测系统提供的信息，并为今后改进观测网络和仪器设计提供指导。新的发展还将有助于新的实地实验和新的化学品监测网络的设计进程。

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).This research develops the algorithmic and computational framework needed for a judicious assessment of the observation impact in air quality modeling. Novel algorithms in the framework of model-constrained optimization will allow to account for the data location in the time-space domain, observation type, instrument type, and data interaction in the presence of multiple observing systems. Specifically, the research is focused on: development of high-order adjoint-based observation impact techniques consistent to four-dimensional variational data assimilation schemes; assessment of forecast sensitivity to observation error variances and estimation of the forecast impact of uncertainties in the specification of the input error statistics; development of efficient computational approaches to allow for practical implementations of the observation impact algorithms; validation of the novel techniques through observing system experiments and assessment of the potential impact of new observing systems.The ability to accurately represent the distribution of atmospheric pollutants in relation to various anthropogenic activities is essential for chemical weather forecasting to protect the population, for answering science questions related to the future of our planet, and for designing sound environmental policies. An accurate representation of the chemical composition of the atmosphere requires a close integration of models and observations through data assimilation.Data assimilation is the process by which model predictions utilize measurements to produce an optimal representation of the state of the atmosphere. As more observations are becoming available and new measurement networks are being planned, it is of critical importance to develop the capabilities to best utilize the data, to better manage the sensing resources, and to design more effective field experiments and networks to support atmospheric chemistry and air quality studies. This research develops the computational tools required to optimize the information provided by the existing observing systems and to provide guidance for future improvements to the observational network and instruments design. The new developments will also help the design process of new field experiments and of new chemical monitoring networks.