Predicting ambient aerosol thermal–optical reflectance measurements from infrared spectra: elemental carbon

[摘要] Elemental carbon (EC) is an important constituent of atmospheric particulatematter because it absorbs solar radiation influencing climate and visibilityand it adversely affects human health. The EC measured by thermal methodssuch as thermal–optical reflectance (TOR) is operationally defined as thecarbon that volatilizes from quartz filter samples at elevated temperaturesin the presence of oxygen. Here, methods are presented to accurately predictTOR EC using Fourier transform infrared (FT-IR) absorbance spectra fromatmospheric particulate matter collected on polytetrafluoroethylene (PTFE orTeflon) filters. This method is similar to the proceduredeveloped for OC in prior work (Dillner and Takahama, 2015). TransmittanceFT-IR analysis is rapid, inexpensive and nondestructive to the PTFE filtersamples which are routinely collected for mass and elemental analysis inmonitoring networks. FT-IR absorbance spectra are obtained from 794 filtersamples from seven Interagency Monitoring of PROtected Visual Environment(IMPROVE) sites collected during 2011. Partial least squares regression isused to calibrate sample FT-IR absorbance spectra to collocated TOR ECmeasurements. The FT-IR spectra are divided into calibration and test sets.Two calibrations are developed: one developed from uniformdistribution of samples across the EC mass range (Uniform EC) and onedeveloped from a uniform distribution of Low EC mass samples (EC < 2.4 μg,Low Uniform EC). A hybrid approach which applies the Low ECcalibration to Low EC samples and the Uniform EC calibration to all othersamples is used to produce predictions for Low EC samples that have meanerror on par with parallel TOR EC samples in the same mass range and anestimate of the minimum detection limit (MDL) that is on par with TOR ECMDL. For all samples, this hybrid approach leads to precise and accurate TOREC predictions by FT-IR as indicated by high coefficient of determination(R²; 0.96), no bias (0.00 μg m⁻³, a concentration value based onthe nominal IMPROVE sample volume of 32.8 m³), low error(0.03 μg m⁻³) and reasonable normalized error (21 %). These performancemetrics can be achieved with various degrees of spectral pretreatment(e.g., including or excluding substrate contributions to the absorbances) and arecomparable in precision and accuracy to collocated TOR measurements. Onlythe normalized error is higher for the FT-IR EC measurements than forcollocated TOR. FT-IR spectra are also divided into calibration and testsets by the ratios OC/EC and ammonium/EC to determine the impact of OC andammonium on EC prediction. We conclude that FT-IR analysis with partialleast squares regression is a robust method for accurately predicting TOR ECin IMPROVE network samples, providing complementary information to TOR OCpredictions (Dillner and Takahama, 2015) and the organic functional groupcomposition and organic matter estimated previously from the same setof sample spectra (Ruthenburg et al., 2014).

[学科分类] 几何与拓扑