Reducing Uncertainty Surrounding Climate Change Using Emergent Constraints

全球气候模式（GCMs）是了解气候如何响应温室气体增加和其他强迫因子的强大工具，它们被广泛用于为关注未来气候变化的决策者和利益相关者提供指导。这些模型捕捉了气候变化的基本物理特性，但它们的效用受到重要反馈效应表示的不确定性的限制。例如，在雪反照率反馈（SAF）中，变暖的条件导致积雪减少，以吸收更多阳光的较深的地下覆盖物取代高反射的雪面，从而增强初始变暖（在较冷的气候中则相反）。在本项目中，pi通过将当前气候观测与气候变化模拟相联系，力求改善gcm中反馈的表示。为了做到这一点，pi使用了“紧急约束”方法，在这种情况下，它指的是当前和未来气候变化模拟之间的关系，这些模拟在gcm集合中是稳健的。pi先前的工作表明，从未来气候模拟中计算出的SAF强度与从相同模式的当前模拟中从季节周期中得出的SAF密切相关。从当前季节周期得到的SAF可以与实际观测值进行比较，从而提供了一种方法，利用现有观测值来确定哪些gcm最可能正确地代表SAF对气候变化的贡献。从某种意义上说，约束是从对一组gcm集合的强健集体行为的物理动机检查中产生的，而不是从第一原理推导中产生的。pi先前的工作（见AGS-0135136）确立了紧急约束作为验证气候变化模拟反馈强度的一种方法的有效性，而目前的工作试图超越验证，并使用此类约束作为改进模型中反馈表示的手段。与模型验证相比，模型开发是紧急约束的更具挑战性的应用，因为约束必须与模型中的特定物理参数化具体相关。特别是，pi试图确定GCM陆地表面分量模式的各个方面，这些方面是SAF强度的关键决定因素，并确定如何调整它们以减少SAF强度的偏差。可能的候选包括地面和植被冠层之间的雪的分配。工作的范围可以扩大到考虑海冰反照率反馈，其操作类似于上述的SAF，并包括海冰上的雪在确定反馈强度方面的作用。进一步的研究将考虑决定全球水文循环对温室气体增加引起的全球变暖的响应程度的紧急限制。在气候变暖的情况下，全球水文循环预计将加速，全球蒸发量和降水量将以每变暖一度几个百分点的速度增加。预期加速的原因是很容易理解的，但是在加速的幅度上有相当大的不确定性，正如气候变化模拟中模式间的大差异所代表的那样。这里的工作试图通过检查伴随降水和蒸发变化的大气能量收支的变化，得出水文循环的紧急约束。突发约束将气候变化模拟中与降水有关的大气能量收支变化与当前模拟中的大气能量学方面联系起来，然后可以将其与基于卫星的观测进行比较。例如，初步工作表明，气候变化模拟中晴空短波吸收与控制模拟中晴空短波吸收对总可降水量（TPW）的敏感性高度相关。这一限制是值得关注的，因为对TPW的敏感性可以受到卫星观测的限制，而且它也与gcm中短波辐射传输的参数化密切相关。减少未来气候变化模式预估中的不确定性具有社会和科学意义，因为不确定性的减少将增加面对气候变化可能影响的利益攸关方和决策者所使用的气候变化预估的价值。pi将在模型开发工作中直接与建模中心合作，在年度科学会议上召开特别会议，分享项目结果，建模社区的主要成员将提供指导。这些会议的结果将在研究出版物中加以综合，以促进改进模型参数化的广泛努力。此外，还计划开展一项外联活动，向公众解释气候模式及其不确定性。尽管明确需要关于gcm的准确、最新和易于理解的信息，它们的不确定性的来源，以及为改进它们所做的工作，但目前这些信息并不容易获得。为了满足这一需求，该项目将开发和维护一个网站，作为气候建模的教育资源。该网站将包含GCM入门，不确定性来源的讨论，以及GCM优势和局限性的评估。进一步的推广将通过在当地科学博物馆“探索立方”举行的公开讲座和示范来进行。这一推广活动将包括制作视频，在一个球体上的科学（SoS）显示器上展示，这些视频将供世界各地的其他SoS显示器使用。最后，该项目提供了研究生的教育和培训，从而在这一科学领域发展未来的劳动力。

Global climate models (GCMs) are powerful tools for understanding how the climate responds to forcing from greenhouse gas increases and other forcing agents, and they are widely used to provide guidance to policy makers and stakeholders concerned about future climate change. These models capture the basic physics of climate change, but their utility is limited by uncertainties in the representation of important feedback effects. For example in the snow albedo feedback (SAF) warmer conditions lead to reductions in snow cover, replacing a highly reflective snow surface with darker underlying ground cover which absorbs more sunlight, thereby enhancing the initial warming (the opposite holds in a cooling climate). In this project the PIs seek to improve the representation of feedbacks in GCMs by relating observations of present-day climate to simulations of climate change. To do this the PIs use the method of "emergent constraints", which in this context refers to relationships between present-day and future climate change simulations which are robust across an ensemble of GCMs. Previous work by the PIs has shown that the strength of the SAF calculated from future climate simulations is strongly correlated with the SAF derived from the seasonal cycle in present-day simulations from the same models. The SAF derived from the present-day seasonal cycle can be compared with real-world observations, thus providing a means to use available observations to determine which GCMs are most likely to correctly represent the SAF contribution to climate change. The constraint is emergent in the sense that it emerges from a physically motivated examination of the robust collective behavior of an ensemble of GCMs, rather than from a first-principles derivation.The PIs' previous work (see AGS-0135136) established the usefulness of emergent constraints as a way to validate feedback strength in climate change simulations, and the present work seeks to go beyond validation and use such constraints as a means to improve the representation of feedbacks in models. Model development is a more challenging application of emergent constraints than model validation, as the constraints must be specifically related to particular physical parameterizations in the model. In particular, the PIs seek to identify aspects of the land surface component model of the GCM which are key determinants of SAF strength, and determine how they can be adjusted to reduce biases in SAF strength. Possible candidates include the apportionment of snow between the ground and the vegetation canopy. The scope of the work may be expanded to consider the sea ice albedo feedback, which operates similarly to SAF as described above and includes a role for snow on sea ice in determining feedback strength.Further research will consider emergent constraints that determine the extent to which the global hydrological cycle responds to global warming induced by greenhouse gas increases. The global hydrological cycle is expected to accelerate in a warming climate, with increases in global evaporation and precipitation at a rate of a few percent per degree of warming. The reasons for expecting acceleration are well understood, but there is considerable uncertainty in the magnitude of the acceleration, as represented by the large model-to-model spread in climate change simulations. Work here attempts to derive emergent constraints for the hydrological cycle by examining the changes in the atmospheric energy budget that accompany changes in precipitation and evaporation. The emergent constraints connect precipitation-related changes in atmospheric energy budgets from climate change simulations to aspects of atmospheric energetics in present-day simulations, which can then be compared to satellite-based observations. Preliminary work shows, for example, that the clear sky shortwave absorption in climate change simulations is highly correlated with the sensitivity of clear sky shortwave absorption to total precipitable water (TPW) in control simulations. This constraint is of interest because sensitivity to TPW can be constrained by satellite observations, and it is also closely related to the parameterization of shortwave radiative transfer in GCMs.The reduction of uncertainty in model projections of future climate change is of societal as well as scientific interest, as reductions in uncertainty would increase the value of climate change projections used by stakeholders and policy makers confronting the possible impacts of climate change. The PIs will work directly with modeling centers in their model development work, convening special sessions at annual scientific meetings in which results of the project will be shared and key members of the modeling community will offer guidance. Results of these sessions will be synthesized in research publications, in an effort to catalyze a broad effort to improve model parameterizations.In addition, an outreach activity is planned to explain climate models and their uncertainties to the general public. Despite the clear need for accurate, up-to-date, and easily comprehensible information about GCMs, their sources of uncertainty, and the work being done to improve them, such information is not easily available at present. To address this need the project will develop and maintain a website to serve as an educational resource on climate modeling. The website will contain a GCM primer, discussion of sources of uncertainty, and an assessment of the strengths and limitations of GCMs. Further outreach will be conducted through public lectures and demonstrations held at the Discovery Cube, a local science museum. This outreach will include the development of videos to be presented on a Science on a Sphere (SoS) display, and these will be available for use on other SoS displays worldwide. Finally, the project provides for the education and training of a graduate student, thereby developing the future workforce in this scientific area.