The Development of Optical Classification Models for Ambient Aerosols Using Machine Learning

点击翻译按钮获取中文摘要

This project will develop a tool using machine learning to classify ambient aerosols based on measured optical properties. Clustering methods will be used to identify samples with similar optical properties without having to make any a priori assumptions about their identities. The clusters will be used to create classification models that can differentiate aerosol types. The accurate optical classification of aerosols can help to improve climate models by providing constraints on how aerosols are treated in climate models.The objectives of this project are to: (1) identify and define clusters of ambient aerosols with distinct optical properties; (2) build optical classification models based on observations and clusters from Objective 1; (3) translate optical classification models for use with low-cost sensors. Outreach activities include the use of low-cost sensors to help students at the middle school, undergraduate and graduate levels learn firsthand about aerosols in the atmosphere.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.