HBCU-Excellence in Research: Understanding Atmospheric Moist Convection and Organization Using Automatic Feature Identification and Tracking

对流上升流核及其生命周期是大气湿对流的基本过程。积云参数化是全球气候预报的重要组成部分。这项研究的最终目标是使用基于物理的、时变的对流上升气流核心概念来改进积云参数化。PI和Co-PI之间的密切合作将增加新的研究领域(大气科学)，并加强摩根州立大学目前的研究和教学(计算机科学)，这是一所HBCU(历史上由黑人组成的学院和大学)，最近被指定为马里兰州卓越的公立城市研究型大学。在这项研究中加入了教育部分，研究生和本科生将在摩根州立大学的STEM(科学、技术、工程和数学)领域得到支持和培训。凝聚的上升气流核心是对流热量和水分传输的引擎。然而，以往从拉格朗日角度对对流上升气流核心进行系统研究的报道尚不多见。该项目将利用现有的计算机图像识别和机器学习算法，使其适用于高分辨率的云分辨率模型模拟，以便自动识别上升气流的核心特征，并在其整个生命周期中对其进行跟踪。将进行三套逐渐复杂的模拟：触发单一对流的集合模拟、准平衡状态模拟和利用观测到的大尺度强迫的个例研究。对流上升气流核心特征，如其大小、深度、寿命和空间分布将从这些模拟中得出。扰动环境条件(如稳定性、水汽、风切变)的模型敏感性试验将揭示环境对上升气流核心特征的控制。对流组织也将在上升气流核心的范围内通过其聚集、合并和分裂来研究。虽然在目前的项目框架下不能立即实现，但最终目标是使用本研究中建立的时变对流核心概念来改进全球气候预测中的积云参数化。通过这个项目，将编制一个包含数千个对流上升气流核心及其生命周期的数据库，并将其公之于众，以供进一步研究。这是一项跨学科研究，利用信息科学中现有的数据挖掘和机器学习算法，并将其应用于嵌入对流的三维风场。识别和跟踪上升气流核心是一种新的方法，它将为湿对流的动力学和物理学及其组织提供新的见解。与时间相关的上升气流核心概念可能被用来取代当前GCM积云参数化中的定常状态羽流模型，允许使用物理基础概念进行尺度无关的积云参数化。该奖项反映了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。

Convective updraft core and its lifecycle are fundamental processes in atmospheric moist convection. Cumulus parameterization is a critical component in global climate prediction. The ultimate goal of the study is to improve cumulus parameterization using a physically based, time-variant convective updraft core concept. The close collaborations between the PI and Co-PI will add new research field (atmospheric sciences) and strengthen the current research and teaching (computer sciences) at Morgan State University, an HBCU (Historically Black Colleges and Universities) that has been recently designated as Maryland's preeminent public urban research university. An education component is built into the study, where a graduate student and an undergraduate student will be supported and trained in STEM (Science, Technology, Engineering, and Mathematics) field at Morgan State University.Coherent updraft cores are the engine for convective heat and moisture transport. However, systematic study of convective updraft cores from a Lagrangian point of view has not been performed before. This project will take advantages of existing computer image recognition and machine learning algorithms, adapt them to high resolution cloud-resolving model simulations, in order to automatically identify updraft core features and track them throughout their full life cycle. Three sets of progressively more sophisticated simulations will be carried out: ensemble simulations of triggered single convection, quasi-equilibrium state simulation, and case studies using observed large-scale forcing. Convective updraft core characteristics such as their sizes, depths, lifespan and spatial distributions will be derived from these simulations. Model sensitivity tests that perturb environmental conditions, e.g., stability, water vapor, wind shear, will reveal environmental controls on updraft core characteristics. Convection organization will also be studied in the context of updraft cores through their congregation, merging and splitting. Although not immediately achievable in the framework of current project, the ultimate goal is to use the time-variant convective core concept established in this study to improve cumulus parameterization in global climate predictions. A database including thousands of convective updraft cores and their lifecycles will be compiled and made public through this project for further study.This is an interdisciplinary study that takes advantage of existing data mining and machine learning algorithms in information sciences and applies them to 3D wind fields where convection is embedded. Identifying and tracking updraft cores is a novel approach that will provide new insights to dynamics and physics of moist convection and its organization. The time-dependent updraft core concept can potentially be used to replace the steady state plume model in current GCM cumulus parameterizations, allowing for a scale-independent cumulus parameterization with a physical underlying concept.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.