CAREER: A Unified Multiscale Modeling Approach for Processes in the Atmospheric Boundary Layer

天气现象，包括风的特征，是大气过程在多个尺度上发生的结果，从跨越数千公里的行星和区域系统到每隔几米变化的精细湍流过程。接近地面的大气特性在不到一小时的时间内以低于千米的分辨率变化，因此使用现有的天气模型来描述它们是具有挑战性的。这项研究将开发一个统一的多尺度模拟框架，以改进对大气特性的精细分辨率预测，这些特性对于解决具有高度环境和社会相关性的问题至关重要，例如评估可再生能源、空气污染健康风险、野火风险和城市化对当地和区域气候的影响。该项目还将通过促进教学、培训和外联活动广泛影响社会。具体地说，它将创建与大气科学相关的教案工具包，以支持当地中小学的教育。该项目还将为环境工程、应用数学和统计学的本科生和研究生教育开发新的课程材料，并将培训一名博士后研究员、一名研究生和一名本科生。现有的天气模型及其物理参数是针对特定比例和假设而制定的。执行旨在捕捉不同尺度的模拟(即，耦合中尺度到微尺度的模拟)是具有挑战性的，并且需要大量的计算时间，因此识别细尺度大气过程的关键驱动因素是优化模拟设计的关键。这项研究旨在通过开发新的基于物理和数据驱动的方法来提高对大气边界层过程的理解，以增强在广泛的时空尺度上的预测和建模能力。具体地说，它将解决以下目标：1)开发一个统一的多尺度框架，以克服当前实际案例研究的理论和建模挑战；2)探索天气模式输出对物理方案和模型设置的敏感性，以确定最佳建模实践并提高对微尺度流动驱动因素的理解；以及3)开发一种将机器学习模型与基于物理的模型相结合的混合建模方法，以改进对大气流动的模拟。这项建议还将利用NSF资助的实地实验数据，旨在探索地理、地形和天气条件之间复杂的动态相互作用。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Weather phenomena, including wind characteristics, are the result of atmospheric processes occurring at multiple scales, spanning from planetary and regional systems that vary across thousands of kilometers to fine turbulent processes that vary every few meters. The properties of the atmosphere close to the surface vary at sub-kilometer resolution in less than one hour, thus they are challenging to characterize using existing weather models. This research will develop a unified multi-scale modeling framework to improve predictions of atmospheric properties at fine resolution that are critical to address issues of high environmental and societal relevance, such as assessment of renewable energy resources, air pollution health risks, wildfire risks and urbanization impacts on the local and regional climate. The project will also broadly impact society by promoting teaching, training and outreach activities. Specifically, it will create atmospheric science-related lesson plan kits to support education in local elementary and middle schools. The project will also develop new course material for undergraduate and graduate education in Environmental Engineering, Applied Mathematics and Statistics, and will train one postdoctoral researcher, one graduate and one undergraduate student.Existing weather models and their physical parameterizations are formulated for specific scales and assumptions. Performing simulations designed to capture very different scales (i.e., coupling meso- to micro-scale simulations) is challenging and requires substantial computing time, thus identification of the key drivers of fine scale atmospheric processes is critical to optimize the simulation design. This research aims to improve understanding of atmospheric boundary layer processes by developing new physics-based and data-driven approaches to enhance predictive and modeling capabilities across a wide range of spatio-temporal scales. Specifically, it will address the following objectives: 1) to develop a unified multi-scale framework to overcome current theoretical and modeling challenges for real case studies; 2) to explore the sensitivity of weather model output to physics schemes and model setup to identify optimal modeling practices and improve understanding of drivers of microscale flows; and 3) to develop a hybrid modeling approach integrating a machine learning model with a physics-based model to improve simulations of atmospheric flows. This proposal will also leverage data from NSF-funded field experiments designed to explore the complex dynamical interactions between geographical, terrain and synoptic conditions.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.