Collaborative Research: Understanding Transient Behavior of Climate Feedbacks and Its Role in Decadal Climate Variability and Prediction

气候反馈在耦合模型中的表现仍然是其对温室气体增加所驱动的未来气候变化预测的最大不确定性来源。然而，反馈的符号和大小关键取决于背景气候的动力和热力学性质。由于气候系统的内部变异性和外部强迫的波动，气候模式不可避免地激发低频变率并产生广泛的背景气候。反馈与这些背景气候的漂移，因此气候反馈的瞬时行为，给我们评估气候变化加速和突然气候变化的风险带来了重大挑战。在模式中描述反馈过程的演变性质对于理解十年气候预测中的技巧和偏差的来源特别重要。然而，由于缺乏反馈强度的局部测量，我们在实际模式预测(例如，局部温度变化)和气候反馈的瞬时行为之间建立联系的能力受到极大限制。本项目将使用一种新开发的技术--大气-地面气候反馈-响应分析方法(CFRAM)来解决这一问题。这项研究将利用现代再分析产品和来自耦合模式比较项目(CMIP5)十年预报试验的多模式集合产品，探索CFRAM在量化气候反馈的局部和瞬时行为方面的全部潜力，并了解模式模拟十年气候预测能力的技巧和偏差。将开展的具体研究任务是：(1)利用再分析数据，量化各种反馈对地面气温年周期和年际至年代际变化的相对贡献；(2)评估CMIP5模式在描述卫星年周期和年代际预测方面的保真度；(3)审查模式技能和偏差在模拟十年气候预测能力方面的归属，并找出对卫星年周期的系统误差和对大气温度的十年预测误差有重大贡献的反馈过程。

Representation of climate feedbacks in a coupled-model remains the largest source of uncertainty in its projection of future climate change driven by increasing greenhouse gases. The sign and magnitude of a feedback, however, depends critically on the dynamical and thermodynamical properties of a background climate. Due to internal variability of the climate system and fluctuations in external forcing, climate models inevitably excite low frequency variability and generate a wide range of background climates. The drift of feedbacks with these background climates, hence the transient behavior of climate feedbacks, presents a major challenge in our assessment of the risk of climate change acceleration and abrupt climate change. Delineating the evolving nature of feedback processes in a model is of particular importance for understanding the source of both the skill and bias in decadal climate prediction. However, the lack of a local measure of feedback strength has greatly limited our ability to make connections between actual model predictions (e.g., local temperature change) and transient behavior of climate feedbacks.This project will use a newly-developed technique, the coupled atmosphere-surface climate feedback-response analysis method (CFRAM) to tackle this problem. The study will take advantage of modern reanalysis products and the multi-model ensemble products from the decadal prediction experiments of the Coupled Model Intercomparison Project (CMIP5) to explore CFRAM's full potential in quantifying local and transient behavior of climate feedbacks and understanding the skill and bias of a model's ability to simulate decadal climate prediction. Specific research tasks that will be undertaken are (i) to quantify the relative contributions of various feedbacks to the annual cycle and interannual to decadal variations in surface air temperature (SAT) using reanalyses data; (ii)to evaluate the fidelity of CMIP5 models in representing the SAT annual cycle and decadal prediction of SAT; and (iii) to examine the attribution of models skills and biases in the ability to simulate decadal climate predictions and identify feedback processes that have major contributions to both systematic errors in the SAT annual cycle and forecast errors in decadal prediction of SAT.