18 January 2024
Currently, ARM operates at least four instruments that measure a portion of the ambient aerosol size distribution. These instruments include the scanning mobility particle sizer (SMPS) and the aerodynamic particle sizer (APS). Most users are interested in the entire size distribution or a portion of the size distribution that extends across the measurement range of multiple instruments. However, merging these distributions is not trivial because the instruments all employ different measurement principles and, in most cases, report data as a function of different representations of the aerosol diameter. This value-added product (VAP) uses machine learning (ML) models trained from manually-labeled data quality assessments to provide additional quality checks on the MERGEDSMPSAPS VAP, which merges size distributions measured by the SMPS and APS.
View More
MERGEDSMPSAPS combines the SMPS and APS data onto a common mobility diameter grid using an approach described by Beddows et al. 2010. This process is sensitive to noise in the input data and sometimes results in merged size distributions that are non-physical, yet difficult to algorithmically filter out.
The MERGEDSMPSAPSML VAP provides a simple data quality evaluation for every merged size distribution, allowing scientists to easily differentiate between good and bad data. This VAP is useful for scientists who need a representation of the aerosol size distribution from approximately 10–20,000 nm diameter. VAP data can be used for calculating aerosol scattering and mass loading, estimating the impact of aerosol on clouds, and verifying aerosol-related quantities in models.
References:
Beddows DCS, M Dall’osto, and RM Harrison. 2010. “An Enhanced Procedure for the Merging of Atmospheric Particle Size Distribution Data Measured Using Electrical Mobility and Time-of-Flight Analysers.” Aerosol Science and Technology, 44(11), https://doi.org/10.1080/02786826.2010.502159.
Pedregosa F, G Varoquaux, A Gramfort, V Michel, B Thirion, O Grisel, M Blondel, P Prettenhofer, R Weiss, V Dubourg, J Vanderplas, A Passos, D Cournapeau, M Brucher, M Perrot, and É Duchesnay. 2011. “Scikit-learn: Machine Learning in Python.” Journal of Machine Learning Research 12: 2825-2830.