Single Particle Measurements and Machine Learning Approaches to Study Vertical Distribution of Biological Particles

The vertical distribution of aerosol in the boundary layer and their chemical composition is important for accurately estimating aerosol climatic effects. Biological particles produced by land-biosphere-atmosphere interactions challenge understanding of the sources and formation mechanisms of biological aerosol. Primary biological particles such as spores and pollen can rupture near the surface in dry and windy conditions or when airborne under high humidity conditions, resulting in the atmospheric release of fragments that vary widely in diameter ranging from several nanometers to submicron. In this study we use single particle analysis along with various machine learning models to quantify the abundance of biological particles in atmospheric samples. Atmospheric particles are collected at different altitudes using a tethered balloon system. Single particle analysis is performed using micro-spectroscopy techniques to determine the size, morphology and elemental composition of particles. Machine learning models are trained and validated with known composition, size and morphological features of biological particles. Then we apply the machine learning models to determine the contribution of biological particles in the particle populations. This study improves our current understanding of the vertical distribution of biological particles and their impact on climate system.