Multiscale analysis of ash- and water-rich volcanic plumes using linear spectral deconvolution modeling

The analysis of volcanic plumes from space can determine constituent properties such as mass and effective particle radius. Many methods rely on the upwelling transmission of thermal infrared (TIR) energy from the Earths surface through the plume before detection by the sensor. This restricts observations to plume regions that are optically thin. Additionally, the presence of water vapor/ice can obscure ash and gas detection. Analysis of the opaque plume region has recently been demonstrated using linear deconvolution modeling of TIR emissivity spectra to quantify ash composition and particle size. This technique was tested on smaller plumes using TIR data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). Now, high temporal/low spatial resolution TIR data from the from Advanced Baseline Imager (ABI) onboard the Geostationary Operational Environmental Satellite (GOES) is investigated for larger plumes. Data are first processed through the Temperature Emissivity Separation (TES) algorithm to extract per-pixel emissivity for each ABI TIR band, and then modeled with a modified version of the ASTER Volcanic Ash Library (AVAL) resampled to ABI wavelengths. The TIR spectrum of water ice was also included to determine its presence in the plume. The Nishinoshima (2020) and La Soufrière (2021) eruptions were analyzed with ASTER and GOES data respectively, to quantify the areal percentage of ice present, and use the high temporal cadence of GOES to observe changes in retrieved ash particle size and composition over the course of the eruption. Results thus far demonstrate that ash and ice are detectable in these plumes using this method. The AVAL spectral library end-members provided a good fit (low root-mean-square error) to the image data, with an andesite composition matching closely to Nishinoshima ash and a basaltic-andesite to La Soufrière ash. A greater percentage of the ice end-member was found close to the vent, likely due to higher water vapor concentration and a higher plume altitude in this region. Future work includes adapting this method to other current and future sensors at higher spectral and spatial resolution, and using these results to improve the optical inputs to modeling algorithms operating in the optically-thin plume region for increased retrieval accuracy.