Collaborative Research: EaSM 2: Stochastic Simulation and Decadal Prediction of Large-Scale Climate

在这个项目下，将在观测和模拟最先进的地球气候的大气环流模式(GCM)中研究大尺度的低频模式(LFMS)。该项目将建立在以前由美国国家科学基金会和美国能源部资助的LFMS工作的基础上，这些工作源于大西洋和太平洋区域的风驱动环流和颠覆环流，以及它们与大气和热带气候变异性的相互作用。该项目将侧重于：(A)增进目前对LFMS及其相互作用的了解；(B)开发、修订和测试基于这些模式的概率十年预测的统计方法；(C)结合观测和GCM模拟使用这些十年预测，以深入了解气候变化和气候变异性在十年和更长时间尺度上的动力原因。该项目将继续利用先进的数据自适应光谱方法，通过观测和最近两个国际气候对比模拟项目的模拟，确定大西洋和太平洋的年代际模式。这些方法将包括多通道奇异谱分析(MSSA)和调和库普曼分析(HKA)，以及基于经验模型降阶(EMR)的随机建模。研究人员将进一步应用新的方法研究加州大学洛杉矶分校和威斯康星大学开发的同步混沌振荡器，以进一步了解太平洋内部以及大西洋和太平洋之间的十年尺度遥相关及其与厄尔尼诺-南方涛动(ENSO)的关系。使用项目第一部分研究的LFMS，研究人员将检验它们的十年可预测性，并评估我们基于EMR、MSSA和HKA的新经验预测模型所作的追溯十年预测的技巧。这些经验预测将与作为IPCC第五次评估报告一部分的初步气候预测进行比较。这种比较将有助于量化由于固有的年代际模式及其与ENSO以及与气候变化的相互作用而产生的潜在技能。这些活动的结合将产生先进的随机模拟工具，用于根据观测数据和最稳健、最可预测的近期气候变化和气候可变性因素的组合，建立未来气候的基准情景，直至2050年左右。该项目的智力优势是为气候预测和动态气候模型的验证开发通用的统计工具和方法。该项目还将促进我们对显著的大尺度低频模式之间的全球耦合的理解。该项目的一个预期成果是，通过开发和测试与全球多个LFM之间的耦合和同步相关的新指标来验证气候模型预测，改善对气候预测可靠性的估计。该项目的更广泛影响在于解决一个具有最大社会重要性的问题，即理解气候变化和变化。该项目的研究人员致力于让他们的随机模拟更容易地提供给气候社区。这项工作将促进加州大学洛杉矶分校和威斯康星大学之间的科学伙伴关系，并进一步发展这两个机构的气候动力学研究生课程。主要调查员将通过专门的网站、参考出版物、研讨会以及在国家和国际会议上的发言等方式广泛传播调查结果。

Under this project, large-scale, low-frequency modes (LFMs) will be studied in observations and simulations of state-of-the-art general circulation models (GCMs) of Earth's climate. This project will build on previous NSF- and DOE-funded work on LFMs arising from the ocean's wind-driven and overturning circulations in the Atlantic and Pacific sectors, as well as from their interactions with the atmosphere and tropical climate variability. This project will focus on: (a) improving current understanding of LFMs and interactions between them; (b) developing, revising and testing statistical methods for probabilistic decadal prediction based on these modes; and (c) using these decadal predictions in conjunction with observations and GCM simulations to gain insight into the dynamical causes of climate change and climate variability at decadal and longer time scales. This project will continue the identification of decadal modes in the Atlantic and Pacific, from both observations and the simulations from the two most recent international climate intercomparison modeling projects, using advanced, data-adaptive spectral methods. These methods will include multi-channel singular spectrum analysis (MSSA), as well as harmonic Koopman analysis (HKA), along with stochastic modeling based on empirical model reduction (EMR). The investigators will further apply novel methods for the study of synchronized chaotic oscillators developed at UCLA and UWM, to gain additional understanding of decadal-scale teleconnections within the Pacific and between the Atlantic and the Pacific, as well as their relationships with the El Nino-Southern Oscillation (ENSO). Using the LFMs studied in the first part of the project, the investigators will examine their decadal predictability and assess the skill of retrospective decadal forecasts made with our new empirical forecast models based on EMR, MSSA and HKA. These empirical forecasts will be compared against the initialized climate predictions being made as part of the IPCC's 5th Assessment Report. This comparison will help quantify potential skill due to intrinsic decadal modes and their interactions with ENSO, as well as with climate change. The combination of these activities will lead to advanced stochastic simulation tools for creating benchmark scenarios of future climate, based on a combination of observed data and the most robust and predictable elements of near-term climate change and climate variability, out to about 2050.The intellectual merit of this project is in developing versatile statistical tools and methodologies for climate prediction and the validation of dynamical climate models. The project will also advance our understanding of global coupling between prominent large-scale low-frequency modes. An expected outcome of the project is improved estimates of reliability of climate projections, by developing and testing novel metrics - associated with coupling and synchronization between multiple LFMs across the globe - for validating climate model predictions.The project's broader impacts lie in addressing a problem of utmost societal importance, i.e., understanding climate variability and change. The project's investigators are committed to making their stochastic simulations easily available to the climate community. This work will foster scientific partnerships between UCLA and UWM, and further the development of graduate curricula in climate dynamics at both institutions. The principal investigators will broadly disseminate the results by means of a dedicated web site, refereed publications, seminars and presentations at national and international meetings.