SGER: North Atlantic Climate Variability from a Nonlinear Perspective

该奖项由探索性研究小额资助计划（SGER）赞助，将帮助支持开发一套新颖的非线性工具，并将其应用于众所周知的气候事件——北大西洋涛动（NAO）的分析。该项目的成功可以证明非线性技术在古气候研究中的价值，揭示线性技术与非线性技术的优缺点，并有助于更好地理解社会感兴趣的时间尺度上的气候变化。开展这一探索项目的关键原因是，许多古气候研究都依赖于一种强有力的均变论假说：将特定沉积特征与气候变量之间的短期相关性应用于更长的沉积记录，以重建气候变量。当三个独立的、不那么均变的古温度计显示，在年平均温度和冰中水的稳定同位素比率之间的短期校准将格陵兰岛过去的温度变化低估了三倍时，这种方法的危险在更广泛的科学界变得明显起来。研究人员和他的同事们建议，提高这种分析可信度的一种方法是遵循完全独立的均变论分析途径，并比较结果。均变说古气候分析中常用的统计工具是线性的。鲁棒非线性“神经网络”技术已经在多个领域得到了发展，研究人员组装了合适的非线性工具，并将其应用于西南极洲古气候条件的重建。策略是将这些工具应用于经典的古气候问题之一：重建北大西洋涛动（NAO）的历史。他将评估非线性技术是否在验证间隔内重建这一重要的气候特征方面做得更好，以及是否提供了新的见解。如果正如他和他的同事们所怀疑的那样，新技术在验证区间内表现得更好，那么它们产生的更长的NAO历史将用于更广泛的研究，并且它与以前重建的差异将为古气候研究的不确定性提供一种估计。这项研究的广泛意义是深远的，因为如果研究人员取得成功，他们将为提高古气候重建的信心提供一种新的工具。

This award, under the auspices of the Small Grants for Exploratory Research (SGER) program, will help support the development of a novel set of nonlinear tools and their application to the analysis of a well-known climatic event, the North Atlantic Oscillation (NAO). Success of this project could demonstrate the value of nonlinear techniques in paleoclimate studies, reveal strengths and weaknesses of linear versus nonlinear techniques, and contribute to better understanding of variability in climate at time scales of societal interest.The crux of the reason for pursuing this exploratory project is that much paleoclimatic research relies on a strong form of the uniformitarian hypothesis: a short term correlation between a particular sedimentary characteristic and a climatic variable is applied to a longer sedimentary record to reconstruct the climatic variable. The danger in such an approach became apparent the wider science community, when three independent, less-uniformitarian paleothermometers showed that a short-term calibration between mean-annual temperature and stable-isotopic ratios of water in ice underestimated past temperature changes in Greenland by threefold. The researcher and his colleagues suggest that one way to improve confidence in such analyses is to follow a completely independent uniformitarian analysis pathway and compare results. The statistical tools used commonly in uniformitarian paleoclimatic analyses are linear. Robust nonlinear "neural network" techniques have been developed in diverse fields, and the researcher has assembled appropriate nonlinear tools and applied them to reconstruction of paleoclimatic conditions in West Antarctica. The strategy is to apply these tools to one of the classic paleoclimatic problems: reconstruction of the history of the North Atlantic Oscillation (NAO). He will assess whether the nonlinear techniques do a better job in reconstructing this important climatic feature in validation intervals, and whether new insights are provided. If, as he and his colleagues suspect, the new techniques are better in the validation interval, then the longer NAO history they produce will be of use in broader studies, and differences between it and previous reconstructions will provide one estimate of the uncertainty in paleoclimatic research.The broader importance of the proposed research is far reaching because if the researchers are successful, they will provide a new tool for improving the confidence of paleoclimatic reconstructions.