Journals / Atmospheric Pollution Research / 2016 / Cilt: 7 - Sayı: 1
Recommendations for calibration factors for a photo-reference method for aerosol black carbon concentrations
- Pages
- 75–81
- DOI
- —
Özet
Large measurement networks of Black Carbon (BC) aerosol are important for understanding its impactson climate and health. PM2.5 filter samples were collected at three urban US locations and one Indiaurban location and were analyzed for Elemental Carbon (EC) and Organic Carbon (OC) concentrationsusing thermo-optical analysis (TO) following the IMPROVE protocol for US samples and NIOSH protocolfor India samples. Site and season-specific calibrations of an inexpensive photo-reference (PR) methodwere created with TO EC measurements of the US filter samples whereas method-specific calibration wasprepared using India filter samples. Piece-wise calibration based on filter loading was also explored.Calibrations were applied across different sites, seasons and methods to determine Root Mean SquareError (RMSE) and average absolute error in each calibration by comparing with reference EC measurements. This paper investigates various calibrations of PR method to improve the agreement between PRmethod and TO EC measurements. Difference in BC estimated error remained within ±10% among threeurban US site-specific calibrations, which suggests that site-specific calibrations are not necessary.Season based calibrations were found to perform best (least RMSE/Mean EC), when applied to sameseason test samples but resulted in large errors of up to 60% RMSE/Mean EC when applied to differentseasons, thus warranting the use of season-specific calibrations of the PR method. RMSE relative to meanEC was 50% when calibration prepared from US samples (IMPROVE protocol) was used to test Indiasamples (NIOSH protocol). However, method-specific calibration prepared from India samples reducedthe error to 24%, showing the large dependency of PR method on reference BC measurement method.Calibration based on filter loading reduced the RMSE slightly for both US urban and India samples andindicated that filters with loadings higher than 20 mg cm2 are not suitable for estimating BC by PRmethod.