Spectral discrimination of coarse and fine mode optical depth

TLDR

The recognition that the aerosol particle size distribution is effectively bimodal permits the extraction of fine and coarse mode optical depths from the spectral shape of the total aerosol optical depth. Partial optical validation is provided by demonstrating the physical coherence of the simple model, showing that coarse‑mode variation aligns with thin‑cloud imagery and fine‑mode variation aligns with clear‑sky and haze imagery, and confirming that the retrieved fine and coarse optical depths are well‑correlated, if weakly biased, relative to formal inversions of combined solar extinction and sky radiance data. The purely optical technique avoids intermediate PSD computations, yields direct optical outputs commensurate with the spectral information content of the total optical depth, provides robust separation of fine and coarse modes yielding more intrinsic aerosol optical statistics, enables closer scrutiny of standard cloud‑screening algorithms, and shows that perturbations from operational cloud screening are generally small to moderate, though diurnal events can cause significant commission or omission errors in the fine and coarse optical depths.

Abstract

The recognition that the aerosol particle size distribution (PSD) is effectively bimodal permits the extraction of the fine and coarse mode optical depths (τ f and τ c ) from the spectral shape of the total aerosol optical depth (τ a = τ f + τ c ). This purely optical technique avoids intermediate computations of the PSD and yields a direct optical output that is commensurate in complexity with the spectral information content of τ a . The separation into τ f and τ c is a robust process and yields aerosol optical statistics, which are more intrinsic than those, obtained from a generic analysis of τ a . Partial (optical) validation is provided by (1) demonstrating the physical coherence of the simple model employed, (2) demonstrating that τ c variation is coherent with photographic evidence of thin cloud events and that τ f variation is coherent with photographic evidence of clear sky and haze events, and (3) showing that the retrieved values of τ f and τ c are well‐correlated, if weakly biased, relative to formal inversions of combined solar extinction and sky radiance data. The spectral inversion technique permitted a closer scrutiny of a standard (temporally based) cloud‐screening algorithm. Perturbations of monthly or longer‐term statistics associated with passive or active shortcomings of operational cloud screening were inferred to be small to occasionally moderate over a sampling of cases. Diurnal illustrations were given where it was clear that such shortcomings can have a significant impact on the interpretation of specific events; (1) commission errors in τ f due to the exclusion of excessively high‐frequency fine mode events and (2) omission errors in τ c due to the inclusion of insufficiently high‐frequency thin homogeneous cloud events.