Quantification of elemental and organic carbon in atmospheric particulate matter using color space sensing-hue, saturation, and value (HSV) coordinates

•PM composition data is needed to address global health and climate challenges.•HSV color space was measured from a large and diverse set of PM loaded filters.•A predictive model for EC and OC was developed using HSV color coordinates.•Both EC and OC could be estimated effectively using the model.•A cost effective EC/OC method can be applied to previously cost prohibitive research.

A fast and cost effective application of color sensing was used to quantify color coordinates of atmospheric particulate matter collected on filters to quantify elemental and organic carbon (EC/OC) loading. This is a unique and novel approach for estimating OC composition. The method used a colorimeter and digital photography to obtain XYZ color space values and mathematically transformed them to HSV cylindrical-coordinates; a quantification method was applied to estimate the NIOSH and IMPROVE (TOR) EC/OC loadings from a set of globally diverse PM samples. When applied to 315 samples collected at three US EPA Chemical Speciation Network (CSN) sampling sites, the HSV model proved to be a robust method for EC measurement with an R2 = 0.917 for predicted versus measured loading results and a CV(RMSE) = 16.1%. The OC quantified from the same sample filters had an R2 = 0.671 and a CV(RMSE) = 24.8% between the predicted and measured results. The method was applied to NIOSH EC/OC results from a set of samples from rural China, Bagdad, and the San Joaquin Valley, CA, and the EC and OC CV(RMSE) were 30.8% and 49.3%, respectively. Additionally, the method was applied to samples with color quantified by a digital photographic image (DPI) with EC results showing good agreement with a CV(RMSE) of 22.6%. OC concentrations were not captured as accurately with the DPI method, with a CV(RMSE) of 77.5%. The method's low analytical cost makes it a valuable tool for estimating EC/OC exposure in developing regions and for large scale monitoring campaigns.

Graphical abstractDownload high-res image (191KB)Download full-size image