Abstract Decadal and longer-term variability in the tropical Pacific has the potential to impact climate worldwide, yet paleoclimate observations and climate model simulations disagree on the nature of this variability. In this study we investigate whether biases from paleoclimate reconstruction methodology and the currently available proxy network cause reconstructions of ENSO to overestimate the multidecadal-century scale variability of the Pacific. We build new reconstructions using a variety of commonly employed methods and test the effects of four types of bias on analogous pseudoproxy reconstructions. These biases include (1) the number of records used in the reconstruction, (2) chronological uncertainty, (3) noise in individual proxy records, and (4) the geographical distribution of the proxy network. We find that each of these biases cause ENSO reconstructions to overestimate the importance of low-frequency variability. These biases may partially explain why multiproxy reconstructions of Pacific variability have stronger low-frequency variability than predicted by historical observations or climate models, and why these reconstructions do not have coherent signals over the multidecadal-century scale.

中文翻译：

---

热带太平洋海温变化的多代理重建中是否存在低频偏差？

摘要 热带太平洋的年代际和长期变率有可能影响全球气候，但古气候观测和气候模型模拟对这种变率的性质存在分歧。在这项研究中，我们调查了古气候重建方法和当前可用的代理网络的偏差是否会导致 ENSO 的重建高估太平洋的几十年尺度变化。我们使用各种常用方法构建新的重建，并测试四种类型的偏差对类似伪代理重建的影响。这些偏差包括 (1) 重建中使用的记录数量，(2) 时间顺序不确定性，(3) 单个代理记录中的噪声，以及 (4) 代理网络的地理分布。我们发现这些偏差中的每一个都会导致 ENSO 重建高估低频变化的重要性。这些偏差可能部分解释了为什么太平洋变率的多代理重建具有比历史观测或气候模型预测的更强的低频变率，以及为什么这些重建在几十个世纪的尺度上没有连贯的信号。