Trace Gas Inversion For InfraRed Spectra (TGIF-IRS): An algorithm to retrieve trace gas concentrations in volcanic plumes

Abstract:

An inversion algorithm (Trace Gas Inversion For Infrared Spectra, TGIF-IRS) was developed to derive the concentration of trace gases in volcanic plumes from hyperspectral infrared atmospheric radiance measurements. The TGIF-IRS inversion algorithm is set up to predict SO₂ path length concentrations and/or location specific concentrations (assuming plume geometry). TGIF-IRS is based on a Partial Least Squared Regression (PLSR) model which was trained from data obtained using thermal IR radiative transfer calculations for different volcanic plume conditions, viewing geometries and atmospheric conditions. The IR radiative transfer calculations were based on MODTRAN5 (version 5.3.2) for computing the transmission and emission terms in a simplified radiative transfer equation. The TGIF-IRS algorithm was developed using PLSR because it is useful for applications with a large number of input variables which are not always independent. The input variables included the background sounding conditions (temperature, moisture and pressure vertical profiles), trace gases (minus SO₂), plume geometries, plume temperatures and assumed temperature differences between the plume and the environment. The TGIF-IRS inversion algorithm was found to work well for elevation angles larger than 5 degrees. At lower elevation angles, the background sky was no longer cold and plume gaseous emissions became less distinguishable. Simulated long wave thermal IR radiances were compared to measured radiances and found to be in agreement confirming that the simulated thermal radiances were accurate.