CMG: Improve the Computational Performance of Global Atmospheric Chemistry Models through Spatial Mechanism Reduction

模拟低层大气（对流层）中氧化剂浓度（臭氧和OH）的能力对广泛的环境问题至关重要。它在解决空气质量、气溶胶和酸的形成以及全球温室气体预算问题方面发挥着至关重要的作用。我们对控制对流层氧化剂的化学因素的基本理解是相当完善的，但是大气模拟的计算挑战是巨大的。化学机制包括数百种耦合的化学物质在时间尺度上的反应，从毫秒到多年不等。在全局模式中求解耦合微分方程的刚性系统的代价是妨碍足够的空间分辨率或时间范围的模拟。这种计算上的困难阻碍了我们在模拟大气化学和解决相关环境问题方面的总体进展。随着卫星观测提供了大量的大气成分数据，这个问题将在未来十年加剧。利用这些数据需要快速的模型。应对这一挑战需要在计算资源和数值算法方面取得实质性进展。本项目将通过应用数学家（Brenner）和大气化学建模师（Jacob）的合作，改进全球模型中描述氧化剂化学的数值算法。我们的中心思想是，大部分化学复杂性都局限于一个相对较小的大气域（化学边界层），靠近大陆表面，在那里发生排放。因此，有可能使用化学机制的目标还原，其中在远程大气中使用一组还原的反应物，而在化学边界层中使用完整的化学机制。由于化学边界层的动态性，需要对不同的化学物质有不同的化学边界层定义，并且需要一个匹配程序来准确地连接两个机制，因此实施起来很复杂。我们期望这个项目的结果将大大提高全球大气化学模型的能力，从而提高我们解决一系列重要环境问题的能力。

The ability to model oxidant concentrations (ozone and OH) in the lower atmosphere (troposphere) is central to a wide range of environmental issues. It plays an essential role in addressing issues of air quality, aerosol and acid formation, and global budgets of greenhouse gases. Our fundamental understanding of the chemical factors controlling tropospheric oxidants is fairly well established, but the computational challenge of atmospheric modeling is enormous. Chemical mechanisms include hundreds of coupled chemical species reacting on timescales ranging from milliseconds to many years. The cost of solving the resulting stiff system of coupled differential equations in a global model is such as to prevent simulations of adequate spatial resolution or temporal extent. This computational difficulty hinders general progress in our ability to model atmospheric chemistry and to address related environmental issues. The problem will be exacerbated over the next decade as satellite observations provide vast amounts of data on atmospheric composition. Exploitation of these data will require fast models. Meeting this challenge requires substantial advances in both computational resources and numerical algorithms.The present project will improve numerical algorithms for describing oxidant chemistry in global models through the collaboration of an applied mathematician (Brenner) and an atmospheric chemistry modeler (Jacob). Our central idea is that most of the chemical complexity is confined to a relatively small atmospheric domain (chemical boundary layer) near the continental surface where emissions take place. There is thus the possibility of using a targeted reduction of the chemical mechanism in which a reduced set of reactants is used in the remote atmosphere and the full chemical mechanism is used in the chemical boundary layer. Implementation is complicated by the dynamic nature of the chemical boundary layer, the need to have different definitions of the chemical boundary layer for different chemical species, and the need for a matching procedure to accurately connect the two regimes. We expect that results from this project will significantly enhance the capabilities of global atmospheric chemistry models and in this manner will improve our ability to address a range of important environmental issues.