N-Dimensional Auto-Clustering of Lidar Measurements to define Defacto Aerosol Types

2026 ARM Summer School Project

K-means Clustering Algorithm designed to classify aerosol types within High-Spectral Resolution Lidar (HSRL) Observations

Project Scope

• While remote sensing instruments like HSRL provide rich information about aerosol and cloud properties throughout the atmospheric column, it does not provide good interpretation despite being sophisticated and depends a lot on the eye of beholder.
• Do natural clusters emerge and if we color-code them by aerosol type, do they match what scientists already know?
• If their clusters can be reproduced with cheap uncalibrated LIDARs, research-grade data can be extended to sites that cannot afford HSRL.

Final Presentation

• Link to Final Presentation