Reducing Greenhouse Climate Proxy Uncertainty

按照目前的轨迹，到本世纪中期，大气中二氧化碳（CO2）的浓度将超过550 ppm。如此高的二氧化碳浓度最后一次出现在2500万年前的新生代早期的“温室”气候中。特别是，始新世早期（约5500万至4800万年前）的特点是过去6500万年来最温暖的气候，南极洲没有冰盖，极地地区比现在高20-40摄氏度，海平面比现在高50 - 70米。这些温暖的始新世气候可以使用用于预测未来气候变化的相同气候模型来模拟，例如政府间气候变化专门委员会（IPCC）第五次评估报告（2013-14）中使用的气候模型。在这份报告中，始新世的气候模型模拟与地质记录中的温度估计进行了比较，以测试在二氧化碳大幅增加的情况下极地地区模拟变暖的准确性。PI Dunkley Jones负责整理IPCC AR 5报告中使用和计算的始新世温度估计。这项工作现在正在一个国际合作项目中得到实质性的改进，该项目将在下一次IPCC报告中运行具有一致边界条件和始新世地理的IPCC级气候模型，作为“深时模型数据相互比较项目”（DeepMIP）的一部分。然而，用于测试这些模型的关键地质数据--始新世地表温度和大气CO2浓度--的准确性的显著提高更难建立。目前对始新世早期海洋表面温度的中等可靠估计仅限于全球七个地点，在高纬度地区，根据所使用的替代方法，可能会出现高达20摄氏度的差异。目前对始新世早期二氧化碳浓度的估计甚至更不确定，范围从约300 ppm到超过2000 ppm。只有一个可靠的始新世早期数据点是基于IPCC强调的具有特别前景的CO2代用方法--基于有孔虫硼同位素和烯酮碳同位素组成的方法。在这里，我们的目标是通过利用两个新的机会，在这些“代理”估计中进行逐步改进。第一，是偶然发现了一套保存完好、未经改变的海洋微化石，由碳酸钙制成，以及始新世海藻和细菌产生的同样保存完好的有机分子生物标志物。这种化石材料的化学性质是替代温度和/或大气CO2估计的基础。这种材料的质量是如此之高，以至于我们建议为始新世早期生成约170个基于烯酮的CO2估计值，而以前没有，以及15个基于硼同位素的估计值来测试目前可用的单个数据点。这些基质及其丰度的罕见共存也提供了使用多种独立方法在同一样本集上估计海洋温度（4种方法）和大气CO2（2种方法）的机会，因此可以直接比较同一时间和地点不同方法的估计值。第二个关键机会是PI Dunkley Jones与在数据分析，统计建模和信号处理方面具有先进专业知识的天体物理学家之间的新合作。随着从任何温室气候中产生有史以来最大的代理到代理比较数据集，这项新的合作将最大限度地提高我们在任何给定的代理方法中得出系统误差的可靠结论的能力。这对于重建温暖的气候状态是至关重要的，其中基于不同代理方法的温度重建之间存在持续的差异。在这里，我们将能够直接比较来自相同样品的方法，并且具有一致的优异保存性。

On current trajectories, the concentration of atmospheric carbon dioxide (CO2) will exceed 550 ppm by the middle of this century. Such high carbon dioxide concentrations last occurred over 25 million years ago during the "greenhouse" climates of the early Cenozoic. In particular, the early Eocene epoch (~55 to 48 million years ago) was characterized by the warmest climates of the past 65 million years, with no ice sheets on Antarctica, polar regions ~20-40 degrees C warmer and sea levels ~50 to 70m higher than present. These warm Eocene climates can be simulated using the same climate models that are used to predict future climate change, such as those used in the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (2013-14). In this report, climate model simulations of the Eocene were compared against temperature estimates from the geological record to test the accuracy of modelled warming in Polar regions at greatly increased CO2.PI Dunkley Jones was responsible for collating the Eocene temperature estimates used and figured in the IPCC AR5 report. This work is now being substantially improved ahead of the next IPCC report within a collaborative international project to run IPCC-class climate models with a consistent set of boundary conditions and Eocene geographies, as part of the "Deep-time Model-data Intercomparison Project" (DeepMIP). Significant improvements in the accuracy of the critical geological data used to test these models - Eocene surface temperatures and atmospheric CO2 concentrations - are, however, more difficult to establish. Current moderately reliable estimates of ocean surface temperatures for the early Eocene are limited to only seven locations globally, and, at high latitudes, can diverge by up to 20 degrees C depending on the proxy method used. Current estimates of early Eocene CO2 concentrations are even more uncertain, ranging from ~300 ppm to in excess of 2000 ppm. There is only one sound early Eocene data point based on the CO2 proxy methods highlighted by the IPCC as having particular promise - those based on foraminiferal boron isotopes and alkenone carbon isotope compositions. Here we aim to make a step-change improvement in these "proxy" estimates by taking advantage of two new opportunities. The first, is the serendipitous discovery of a remarkable suite of very well preserved, unaltered marine microfossils, made of calcium carbonate, alongside similarly well-preserved organic molecular biomarkers produced by Eocene marine algae and bacteria. The chemistry of this fossil material is the basis for proxy temperature and/or atmospheric CO2 estimation. The quality of this material is so high that we propose to generate ~170 alkenone-based CO2 estimates for the early Eocene, where previously there were none, and 15 boron-isotope based estimates to test the single data point currently available. The rare co-occurrence of these substrates and their abundance also provides the opportunity to use multiple independent methods to estimate both ocean temperatures (4 methods) and atmospheric CO2 (2 methods) on the same sample set, and so directly compare estimates from different methodologies at the same time and place. The second key opportunity is a new collaboration between the PI Dunkley Jones and astrophysicists with advanced expertise in data analysis, statistical modelling and signal processing. With the generation of the largest ever dataset of proxy-to-proxy comparisons from any Greenhouse climate, this new collaboration will maximise our ability to draw robust conclusions about systematic errors within any given proxy method. This is vital for the reconstruction of warm climate states where there are persistent discrepancies between temperature reconstructions based on different proxy methods. Here, we will be able to directly compare methods from the same samples and with uniformly excellent preservation.