Methods of Uncertainty Analysis of Complex Non-Linear Components in Integrated Systems Models

PIS建议测试一种对具有不连续状态的复杂、高度非线性模型进行不确定性分析的方法，并应用这些技术来更好地表征温室气体强迫可能在大规模地球系统中引起高度非线性响应的可能性。PIS将使用两种方法：(I)蒙特卡罗方法，其中需要对给定的复杂模型进行数千次运行；以及(Ii)确定性等效建模方法(DEMM)。DEMM将应用于海-气耦合模式，即马歇尔(MIT)的海洋环流模式(OGCM)与大气的二维表示(集成全球系统模式或IGSM)相耦合。OGCM是一个复杂的模型，它表现出非线性行为，包括温盐环流(THC)的关闭。该模式包括：排放预测和政策分析模型，预测经济活动、能源使用以及因应特定气候变化而产生的人为温室气体排放；大气组成与气候预测耦合模型，其中包括以戈达德空间研究所开发的大气全球气候模型为基础的二维陆地-海洋模型；以及陆地生态系统模型，模拟生态系统对气候和二氧化碳变化的反应，即碳和氮通量及水库规模。这项工作很重要，因为它将为气候模型不确定性的量化提供有用的见解。这一结果将引起政策制定者以及参与气候建模的科学家的兴趣。该奖项是由NSF的综合评估方法和模型(MMIA)倡议资助的。

The PIs propose to test a methodology for conducting uncertainty analysis of complex, highly non-linear models with discontinuous states and apply the techniques to better characterize the likelihood that greenhouse gas forcing could cause a highly non-linear response in large-scale earth systems. The PIs will utilize two approaches: (i) a Monte Carlo approach in which thousands of runs of a given complex model are required, and (ii) the Deterministic Equivalent Modeling Method (DEMM). DEMM will be applied to a coupled ocean-atmosphere model, viz the ocean general circulation model (OGCM) of Marshall (MIT) coupled to a two-dimensional representation of the atmosphere (the Integrated Global System Model or IGSM). The OGCM is a complex model that exhibits nonlinear behavior, including the shutdown of the thermohaline circulation (THC). The IGSM includes the Emissions Prediction and Policy Analysis (ESSA) model that predicts economic activity, energy use and resulting anthropogenic greenhouse gas emissions in response to given climatic changes; a coupled model of atmospheric composition and climate prediction that includes a two-dimensional land-ocean model based on the atmospheric GCM developed at Goddard Institute for Space Studies; and the Terrestrial Ecosystem Model that simulates the ecosystem response, i.e., carbon and nitrogen fluxes and reservoir sizes in response to changes of climate and CO2. The work is important because it will yield useful insights into the quantification of climate model uncertainties. The results will be of interest to policy makers as well as scientists involved in climate modeling. The award is funded under NSF's Methods and Models for Integrated Assessment (MMIA) initiative.