Collaborative Research: RUI--Applying Measurements, Models, and Machine Learning to Improve Parameterization of Aerosol Water Uptake and Cloud Condensation Nuclei

大气气溶胶是大气中无处不在的颗粒，由灰尘、煤烟、污染物甚至树木的自然排放物组成。气溶胶对天气和气候至关重要，因为它们散射阳光，并作为形成云滴的基础。该奖项将为来自阿巴拉契亚州立大学和佐治亚理工学院的一组研究人员提供资金，以研究随着湿度的增加颗粒的生长，以及作为云滴基础的颗粒大小范围。政府间气候变化专门委员会（IPCC）的报告强调，气溶胶对气候的影响是气候预测的一个关键不确定性因素。该项目具有显著的教育和培训效益，计划有8-12名本科生和硕士生参与该项目。阿巴拉契亚州立大学是一所以本科为主的大学，将受益于与研究密集型机构的合作。该合同的首要科学目标是训练、评估和应用经过测量训练的模型，用于计算气溶胶液态水含量（ALWC）和云凝结核（CCN）光谱，这些模型位于北卡罗来纳州布恩的阿巴拉契亚州立大学的气溶胶网络站点。ALWC不能直接测量，但可以通过更常用的气溶胶光学特性来估计。2024年冬季和夏季的密集野外活动将为开发、训练和评估机器学习模型提供必要的数据，这些模型将用于计算ALWC和CCN光谱。然后，这些模型将被回顾性地应用于阿巴拉契亚圣测量的历史数据库，以检查气溶胶吸湿性、ALWC和CCN光谱是如何以及为什么变化的。更具体地说，研究人员将测试以下假设：机器学习模型，如Random Forest，当使用具有区域代表性的颗粒数大小分布和气溶胶光学特性进行训练时，能够预测Appalachian St.站点的ALWC和CCN光谱；近年来，美国东南部气溶胶成分的变化导致阿巴拉契亚圣测量到的吸湿性气溶胶减少。吸湿性颗粒的减少反过来导致alwc的降低。在过去十年中，气溶胶组成、吸湿性和细模态粒径的变化正在降低阿巴拉契亚圣站点不同过饱和值下的CCN浓度。该项目由地球科学理事会共同资助，旨在支持人工智能/机器学习在地球科学领域的发展。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。

Atmospheric aerosols are ubiquitous particles in the atmosphere that are made up of dust, soot, pollution, or even natural emissions from trees. Aerosols are crucially important for weather and climate because they scatter sunlight and act as the base for developing cloud droplets. This award will provide funding for a team of researchers from Appalachian State University and Georgia Tech to study the growth of particles with increasing humidity, and the range of particle sizes that serve as the base for cloud droplets. Aerosol impacts on climate have been highlighted in the Intergovernmental Panel on Climate Change (IPCC) reports as a key uncertainty for climate projections. The project has significant educational and training benefits, with plans for 8-12 undergraduate and Master’s level students to be involved in the project. Appalachian State is a primarily undergraduate university and will benefit from collaboration with a research-intensive institution. The overarching scientific objective of this award is to train, evaluate, and apply measurement-trained models for calculating aerosol liquid water content (ALWC) and cloud condensation nuclei (CCN) spectra at an aerosol network site at Appalachian State University, in Boone, North Carolina. ALWC cannot be directly measured, but it can be estimated from more commonly-measured aerosol optical properties. Intensive field campaigns during the winter and summer of 2024 would provide the necessary data to develop, train, and evaluate machine learning models that would be used to calculate ALWC and CCN spectra. Those models would then be retrospectively applied to the historical database of measurements at Appalachian St. to examine how and why aerosol hygroscopicity, ALWC and CCN spectra are changing. More specifically, the researchers will test the following hypotheses:1. Machine learning models such as Random Forest, when trained using regionally-representative particle number size distributions and aerosol optical properties, are capable of predicting ALWC and CCN spectra at the Appalachian St. site;2. Changing aerosol composition in the Southeastern US is leading to less hygroscopic aerosols measured at Appalachian St. over recent years. Less hygroscopic particles in turn are leading to lower ALWC.3. Changing aerosol composition, hygroscopicity, and fine-mode particle size over the last decade are reducing the CCN concentrations at the Appalachian St. site at different supersaturation values.This project is co-funded by the Directorate for Geosciences to support AI/ML advancement 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.