High climate sensitivity in CMIP6 model not supported by paleoclimate
High climate sensitivity in CMIP6 model not supported by paleoclimate
复制标题
CMIP6 模型中的高气候敏感性不受古气候支持
DOI：
10.1038/s41558-020-0764-6
复制
发表时间：
2020
影响因子：
30.7
通讯作者：
Otto-Bliesner, Bette L.
中科院分区：
文献类型：
--
作者：
Zhu, Jiang;Poulsen, Christopher J.;Otto-Bliesner, Bette L.
平衡气候敏感性（ECS）是全球平均地表温度（GMST）对大气CO2浓度加倍的长期响应。它受到1.5-4.5 C的“可能”范围的限制，自40年前的Charney报告以来几乎没有变化。在参加耦合模式相互比较项目第6阶段（CMIP 6）的现有气候模式中，二十七个中有十个的ECS高于这一范围的上限，而在二十八个CMIP 5模式中只有两个高于这一范围的上限2，3。例如，社区地球系统模型第2版（CESM 2）4（CMIP 6模型）中的ECS为5.3 C（参考文献5）。确定这种高ECS是否现实对于估计未来气候和制定有效的政策和适应计划至关重要。如果没有历史基准来测试CMIP 6模型的ECS，它们可以根据过去的温暖时期进行评估，例如早始新世气候最佳时期（EECO），一段持续高GMST~ 53-50 mya的时期6，7。在这里，我们报告EECO模拟使用CESM 2，并发现其高ECS是不支持的地质证据。我们的模拟结合了EECO边界条件的最新重建，包括古地理，植被覆盖和地表特性6。在冰芯记录之前的过去时期，大气CO2的重建依赖于地球化学和古植物学代理，具有很大的不确定性; EECO值估计为≥ 1，000 ppm（95%置信水平），最佳估计为1，625 ±760 ppm（95%置信区间）6，8，3-9×工业化前CO2（piCO 2）值285 ppm。我们使用1×、2×和3× piCO 2水平进行EECO模拟，并将这些运行中模拟的GMST和赤道海表温度梯度（MTG;占工业化前值的百分比）与最新的代理估计值（分别为29±3 C和69± 13%; 95%置信区间）进行比较。在3× piCO 2的情况下，在代理CO2范围的低端，模拟的GMST为37.5 C，比代理温度估计的上限高5.5 C（图1a）。此外，模拟的热带陆地温度超过55摄氏度，这远远高于植物光合作用的温度耐受性，并且与始新世新热带雨林的化石证据不一致。CESM 2模拟的EECO GMST为29.9 C，MTG为86%，2× piCO 2，远低于代理范围，MTG太陡（图1b）。CESM 2的GMSTs大大高于使用其前身CESM 1（参考文献12）和社区气候系统模型第4版（CCSM 4）13的结果。在6× piCO 2的CESM 1 EECO模拟中，GMST为29.8 C，MTG为76%，与代理证据9吻合良好。在CCSM 4中，16× piCO 2的CO2水平是获得EECO GMST所必需的，该值远高于替代估计值。在CESM 2、CESM 1和CCSM 4中，对非CO2 EECO气候强迫（古地理、植被以及人为气溶胶和陆地冰盖的去除）的敏感性估计分别为9.4 C、5.1 C和2.9 C，显示出模式敏感性与其工业化前ECS（5.3 C、4.2 C和3.2 C）之间单调但非线性的相关性（参考文献10）。5，14）。非线性关系的结果，从ECS的增加与GMST 9和潜在的非CO2的气候强迫的有效性模式版本之间的增加，其潜在的机制值得进一步研究。ECS对模式版本和GMST的依赖性归因于云的反馈，即，通过云对地表变暖的放大。
To the Editor—Equilibrium climate sensitivity (ECS) is the long-term response of global mean surface temperature (GMST) to a doubling of atmospheric CO2 concentrations. It is poorly constrained with a ‘likely’range of 1.5–4.5 C, which has remained nearly unchanged since the Charney report 40 years ago 1. Ten in twenty-seven of the available climate models participating in the Coupled Model Intercomparison Project phase 6 (CMIP6) have an ECS higher than the upper end of this range, in contrast to two in twenty-eight CMIP5 models 2, 3. For example, the ECS in the Community Earth System Model version 2 (CESM2) 4—a CMIP6 model—is 5.3 C (ref. 5). Determining whether this high ECS is realistic is paramount for estimating future climate and crafting effective policies and adaptation plans. Without a historical benchmark to test the ECS of CMIP6 models against, they can be evaluated against past warm periods, such as the Early Eocene Climate Optimum (EECO), a period of sustained high GMST~ 53–50 mya 6, 7. Here, we report EECO simulations using CESM2 and find that its high ECS is not supported by geological evidence. Our simulations incorporate the latest reconstructions of EECO boundary conditions, including paleogeography, vegetation cover and land surface properties 6. Reconstructions of atmospheric CO2 for past times that predate ice-core records rely on geochemical and paleobotanical proxies and have large uncertainties; EECO values are estimated to have been≥ 1,000 ppm (95% confidence level), with a best estimate of 1,625±760 ppm (95% confidence interval) 6, 8,~ 3–9× the pre-industrial CO2 (piCO2) value of 285 ppm. We conduct EECO simulations with 1×, 2× and 3× piCO2 levels and compare the modelled GMST and meridional sea surface temperature gradient (MTG; in per cent of the pre-industrial value) in these runs with the latest proxy estimates (29±3 C and 69±13%, respectively; 95% confidence interval) 7, 9. With 3× piCO2, at the low end of the proxy CO2 range, modelled GMST is 37.5 C, 5.5 C greater than the upper end of proxy temperature estimates (Fig. 1a). Moreover, modelled tropical land temperature exceeds 55 C, which is much higher than the temperature tolerance of plant photosynthesis 10 and is inconsistent with fossil evidence of an Eocene Neotropical rainforest 11. CESM2 simulates an EECO GMST of 29.9 C and a MTG of 86% with 2× piCO2, a level well below the proxy range and a MTG that is too steep (Fig. 1b). CESM2 GMSTs are substantially higher than results using its predecessors, CESM1 (ref. 12) and the Community Climate System Model version 4 (CCSM4) 13. In a CESM1 EECO simulation with 6× piCO2, GMST is 29.8 C and the MTG is 76%, agreeing well with proxy evidence 9. In CCSM4, CO2 levels of 16× piCO2 are necessary to attain an EECO GMST, values that are much higher than proxy estimates. Sensitivity to the non-CO2 EECO climate forcings—paleogeography, vegetation, and the removal of anthropogenic aerosols and land ice sheets—are estimated to be 9.4 C, 5.1 C and 2.9 C in CESM2, CESM1 andCCSM4, respectively, showing a monotonic but nonlinear dependence between model sensitivity and its pre-industrial ECSs, which are 5.3 C, 4.2 C and 3.2 C (refs. 5, 14), respectively. The nonlinear relationship results from the increase of ECS with GMST 9 and potential increases in the effectiveness of non-CO2 climate forcings between model versions, for which the underlying mechanisms merit further research. The dependence of ECS on model versions and GMST has been attributed to the cloud feedback—that is, the amplification of surface warming through …
DeepMIP 对 PMIP4 的贡献：最新古新世和早始新世气候代理数据的选择、编译和分析方法，纳入 DeepMIP 数据库 0.1 版
DOI：
10.5194/gmd-2018-309
发表时间：
2019
期刊：
--
影响因子：
--
作者：
Hollis C
通讯作者：
Hollis C