10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Mutual Information Method for Understanding Key Variables in Atmospheric New-Particle Formation
MARTHA ARBAYANI ZAIDAN, Ville Haapasilta, Rishi Relan, Pauli Paasonen, Veli-Matti Kerminen, Heikki Junninen, Markku Kulmala, Adam S. Foster, Helsinki University
Abstract Number: 269 Working Group: Aerosol Modeling
Abstract New-particle formation (NPF) is an important phenomenon occurring in the atmosphere. This process is very non-linear and complex, involving atmospheric chemistry of precursors and clustering physics as well as subsequent growth before NPF can be observed. Thanks to ongoing efforts, tremendous amounts of atmospheric data is now available, generated from simulation models and laboratory experiments as well as continuous measurements directly from the atmosphere. This fact makes data analysis by human brains as well as via traditional statistical methods more challenging, but on the other hand enables to usage of modern data science techniques. In this work, we investigate the use of mutual information method to understand the relationship between observed NPF events (measured at Hyytiälä, Finland) and a wide variety of simultaneously monitored ambient variables: trace gas, aerosol particle concentrations, meteorology, radiation and a few derived quantities. The investigation aims to identify key factors contributing to the NPF. The applied mutual information method finds that the formation events correlate with water content, sulfuric acid concentration, ultraviolet radiation, condensation sink and temperature. Previously, these quantities have been known to be key factors in the phenomenon via dedicated field studies and laboratory as well as theoretical research. The novelty of this work is in demonstrating that the same results can be obtained by the proposed data analysis method which operates without supervision and physical insight. This suggests that the method is appropriate to be implemented widely in the field of atmospheric sciences to discover other interesting phenomena and its relevant variables.