Comparison of the light absorption coefficient and carbon measures for remote aerosols: An independent analysis of data from the IMPROVE network—I

Abstract

Using the IMPROVE network aerosol data from rural or remote sites across the United States, the ratio of the optically measured light absorption coefficient ( σ a) to the elemental carbon measured by Thermal/Optical Reflectance (TOR) analysis consistently indicates an absorption efficiency that is twice the accepted value of 10m 2g −1. Correlations between σ a and the TOR carbon strongly suggest that the discrepancy is due to an underevaluation of light-absorbing carbon rather than to an overestimation of σ a or a real, higher value of the absorption efficiency. In particular, past doubts about the accuracy and precision of the IMPROVE σ a measurement are here shown to be unsupported by the IMPROVE data. The large empirical correction that is applied to this σ a measurement, for multiple scattering effects due to filter mass loading, is given a new explanation as the effect of an increasing forward scattering fraction as sample thickness increases. The old explanation of shadowing by overlying particles in the sample is rejected as having just the opposite effect to that needed to explain the correction. The use of a diffuse source rather than a laser beam is indicated as a way to avoid the large empirical correction of σ a. Modelling of the light absorption by TOR carbon measurements, at twelve remote sites over a wide portion of the western United States, suggests the following errors in the current interpretation of TOR analysis for these sites: (1) The pyrolysis correction, based upon optical reflectance monitoring, appears to be largely wrong; and (2) The carbon evolving between 450 and 550°C in a pure helium atmosphere, currently interpreted as organic and therefore non-light-absorbing, appears to be as strongly light-absorbing as elemental carbon. However, the present analysis indicates that for a large majority (∼90%) of samples the light-absorbing carbons, as reinterpreted herein, are not only measured accurately by TOR, they are also reasonably well separated from the non-absorbing carbons evolving at and below 450°C. σ a and the TOR carbons, newly interpreted, are seen to be consistent both with one another and with the widely accepted absorption efficiency of elemental carbon. These results are statistical observations, independent of considerations about the form of the aerosol, particularly whether or not it is internally mixed. Multivariate regression of σ a vs the TOR carbons, properly physically constrained, is indicated to be a particularly useful and important analytical tool for distinguishing light-absorbing and non-absorbing carbons measured by thermal methods.