Collaborative Research: Towards Better Understanding of the Climate System Using a Global Storm-Resolving Model

天气现象出现在所有的空间尺度上，从云层内湍流的上下运动到跨越时区的锋面系统，再到环绕地球的急流。自然，不同的模型被用来捕捉不同尺度的现象，包括网格间距约为10米的大涡模拟（LES）模型，用于模拟单个云或小云团。LES模式通常应用于一个有限的区域，宽度可能为10公里至100公里，更大尺度的运动对云的影响表现为施加全区域范围的条件，例如整个区域的单一垂直温度和湿度剖面。这种模拟的缺点是它们不能捕捉小尺度和大尺度之间的双向相互作用，例如小云对大尺度温度和湿度剖面的影响。因此，能够捕获更大范围尺度的模型将是非常有价值的。在这方面被证明非常有用的一个模式是由首席研究员（PI）在21世纪初开发的大气模拟系统（SAM）。SAM已被用作LES模型，例如以1米分辨率模拟建筑物周围的流动，但也被用于网格间距约5公里的区域，以模拟跨越热带的水道区域的波浪运动。SAM一直是研究云行为的主要模式，包括对流云的聚集和云对温室气体引起的变暖的响应，特别是云响应在多大程度上加剧或抵消变暖。最近，PI开发了一个全球版本的SAM，称为gSAM，它将笛卡尔坐标扩展到球坐标，并进行了其他修改，以表示全球范围内的流动。该模型继承了SAM的所有特征，并增加了浸入阶跃地形，这比以前的模型更理想化，并假设了一个平坦的表面。gSAM增加真实感的另一种方式是从观测初始条件开始运行模拟的能力，允许对现实世界天气系统演变的短期“预报”，也称为“预测”。最近的一项研究使用了这一特征，以及“轻推”来重新分析数据，以模拟苏格拉底野外活动期间观察到的情况（见AGS-16628674）。这项研究的结论是，早期冰粒破碎后形成的云冰粒在决定云的宽度方面发挥了作用，从而调节了到达南大洋表面的阳光量。该奖项的目标是进一步发展gSAM，并将其作为天气和气候研究的资源提供给全球研究界。这项工作包括致力于改善两极附近的模型行为，使用机器学习技术提高辐射传输计算的准确性和效率，改善输入/输出性能，并根据卫星数据验证模拟。开发了额外的资源来促进模型的使用和采用，包括全套文档和教程，用于多种配置和分辨率的初始和边界条件数据集，以及几个六个月模拟的模型输出。该模型在GitHub上维护，用户可以使用GitHub存储库为代码开发做出贡献。PI还维护一个模型网站，跟踪使用该模型的出版物，并提供额外的信息和资源。由于gSAM是SAM的扩展，因此很容易将其配置为作为有限域LES模型运行，从而继续为SAM用户社区提供服务。由于gSAM作为开展广泛主题的基础科学研究的工具的力量，这项工作具有更广泛的影响。一个特别令人感兴趣的领域是云与气候变化之间的相互作用，因为云对气候变暖的敏感性可能会影响气候变暖的程度。gSAM还可以帮助我们了解在变暖的世界中极端降水事件的强度可能如何变化。在这两种情况下，gSAM都有助于降低在全球尺度上研究气候过程和在局部尺度上研究云特性的研究团体之间的障碍。该项目还支持一名研究生，从而建立下一代科学劳动力。该项目由地球科学理事会和先进网络基础设施办公室共同资助，以支持地球科学领域的人工智能/机器学习和开放科学活动。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Weather phenomena come in all spatial scales, from the turbulent up-and-down motions within clouds to frontal systems that span a time zone to the jet streams that circle the globe. Naturally different models are used to capture phenomena at different scales, including Large Eddy Simulation (LES) models with grid spacings of perhaps 10m used to simulate individual clouds or small cloud clusters. LES models are typically applied on a limited domain, perhaps 10km to 100km wide, and the influence of larger scales of motion on the clouds is represented by imposing domain-wide conditions, for instance a single vertical profile of temperature and moisture for the whole domain. The drawback of such simulations is that they fail to capture two-way interactions between small and large scales, for instance the effect of small clouds on the large-scale temperature and moisture profiles. Models that can capture a larger range of scales would thus be quite valuable.One model which has proved quite useful for this purpose is the System for Atmospheric Modeling (SAM), developed by the Principal Investigator (PI) in the early 2000s. SAM has been used as an LES model, for instance in simulations of flow around a building at 1m resolution, but has also been used with grid spacings around 5km to simulate wave motions in a channel domain spanning the tropics. SAM has been a workhorse model for studies of cloud behaviors including the aggregation of convective clouds and the response of clouds to greenhouse gas-induced warming, in particular the extent to which the cloud response intensifies or counteracts the warming.Recently the PI developed a global version of SAM called gSAM, which extends the Cartesian coordinates to spherical coordinates and makes other modifications to represent flow on a global domain. The model inherits all of the features of SAM and also adds an immersed step topography, an advance over previous versions which were more idealized and assumed a flat surface. Another way in which gSAM adds realism is the ability to run simulations starting from observational initial conditions, allowing short-term "forecasts", also called hindcasts, of real-world weather system evolution. A recent study used this feature, along with "nudging" to reanalysis data, to simulate conditions observed during the SOCRATES field campaign (see AGS-16628674). The study concluded that the formation of cloud ice particles from the shattering of earlier ice particles plays a role in determining the width of clouds, thus regulating the amount of sunlight that reaches the surface of the Southern Ocean.The goal of this award is to further develop gSAM and make it available to the worldwide research community as a resource for weather and climate research. The work includes tasks devoted to improving model behavior near the poles, improving the accuracy and efficiency of radiative transfer calculations using machine learning techniques, improving input/ouput performance, and validating simulations against satellite data. Additional resources are developed to facilitate use and adoption of the model, including a full suite of documentation and tutorials, initial and boundary condition datasets for multiple configurations and resolutions, and model output for several six-month simulations. The model is maintained on GitHub and users can contribute to code development using GitHub repositories. The PI also maintains a model website that tracks publications using the model and provides additional information and resources. Since gSAM is an extension of SAM it is easily configured to run as a limited-domain LES model, thereby continuing to serve the SAM user community.The work has broader impacts due to the power of gSAM as a tool for conducting basic science research on a wide range of topics. One area of particular interest is the interaction between clouds and climate change, as the sensitivies of clouds to a warming climate could affect the amount of warming that occurs. gSAM can also contribute to our understanding of how the intensity of extreme precipitation events is likely to change in a warming world. In both cases gSAM serves to lower the barriers between the research communities studying climate processes on the global scale and cloud properties on the local scale. The project also supports a graduate student, thereby building the next generation scientific workforce.This project is co-funded by a collaboration between the Directorate for Geosciences and Office of Advanced Cyberinfrastructure to support Artificial Intelligence/Machine Learning and open science activities in the geosciences.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.