What is it about?
A capability of atmospheric models is to simulate the actual chemistry of airborne particles. Satellites and more generally, remote sensing technology, measure changes in radiation to determine general "types" of particles in the atmosphere. These two things, particle chemistry and particle types, are similar but not equal. In this paper, we use data analysis and machine learning techniques to create an algorithm that models can use to create particle types like those retrieved from remote sensing.
Featured Image
Why is it important?
Our results show that models can reproduce particle types like those retrieved from airborne lidar. These results are encouraging and ultimately may help to improve understanding of ambiguous particle types in terms of their actual chemistry, as well as provide another way that models can be constrained by observations.
Perspectives
This research was really challenging, but I am very optimistic about where it may lead! There are still a ton of improvements that can be made to the algorithm presented in this study, but this work is really the first to make a clear link to the observations. I hope that other atmospheric scientists start thinking about machine learning approaches to gaining insights to the complex physics and chemistry that govern the atmosphere. Ultimately, we are trying to understand our impact as humans on climate, so all approaches should be explored. When we hit a roadblock with physics or chemistry, we should seize the opportunity to turn to machine learning for quick answers/insights, where we can unlock the chemistry and physics later.
Dr Kyle W Dawson
North Carolina State University
Read the Original
This page is a summary of: Creating Aerosol Types from CHemistry (CATCH): A New Algorithm to Extend the Link Between Remote Sensing and Models, Journal of Geophysical Research Atmospheres, November 2017, Wiley,
DOI: 10.1002/2017jd026913.
You can read the full text:
Contributors
The following have contributed to this page
