AAAR 34th Annual Conference
October 12 - October 16, 2015
Hyatt Regency
Minneapolis, Minnesota, USA
Abstract View
Infrared Spectroscopy Calibration Models for Prediction of Thermal Optical Reflectance (TOR) OC and EC in IMPROVE Monitoring Network Sites: Interpretation and Extended Evaluation
Matteo Reggente, SATOSHI TAKAHAMA, Ann Dillner, Ecole Polytechnique Federale de Lausanne, Switzerland
Abstract Number: 213 Working Group: Instrumentation and Methods
Abstract Dillner and Takahama (2015a, 2015b) developed and evaluated partial least squares (PLS) calibration models with infrared spectra for prediction of TOR-equivalent OC and EC in 2011 US IMPROVE ambient samples collected on Teflon (PTFE) filters. This method uses ambient samples (analyzed inexpensively and without damaging the sample) as calibration standards to allow prediction in other ambient samples, and its capability for prediction of OC and EC concentrations to within measurement precision was demonstrated when sample loading and composition of evaluation samples were well represented in the calibration sample set.
In this presentation, we introduce several sparse calibration methods that allow us to isolate the most relevant absorption bands in the calibration model; thereby aiding interpretation with regards to the vibrational modes of molecules present in aerosol mixtures that allow us to quantify TOR OC and EC from infrared spectra. Furthermore, we evaluate the applicability of our calibration models developed from the 2011 sample set to new samples analyzed from 17 sites in 2013. These new sites include one in South Korea, sites that experienced significant smoke impact, and several additional rural sites and urban sites. The rural and smoke-impacted sites are predicted with acceptable accuracy, but predictions for the South Korean site (with high sample loadings) and one new US urban site show degraded performance. We discuss means for anticipating prediction errors a priori based on the examination of similarity of new infrared spectra to those in the calibration model, and strategies for constructing better models with improved prediction accuracy.
Dillner, A. M., Takahama, S., Predicting Ambient Aerosol Thermal Optical Reflectance (TOR) Measurements from Infrared Spectra: Organic Carbon, Atmospheric Measurement Techniques, 8, 1097-1109, 2015a. doi: 10.5194/amt-8-1097-2015.
Dillner, A. M., Takahama, S., Predicting Ambient Aerosol Thermal Optical Reflectance (TOR) Measurements from Infrared Spectra: Elemental Carbon,
Atmospheric Measurement Techniques, submitted, 2015b.