Through the Mist: Satellite Remote Sensing of Volcanic Ash with Ground-truthed Ash Detection Modeling in the Aleutians

Ash produced in volcanic eruptions, depending on its composition and volume, can have widespread implications ranging from toxic effects on flora and fauna to further-afield impacts on aviation, ocean fertilization and climate forcing. Although the Aleutians are well-studied volcanoes, increased knowledge of the eruption plume chemistry and accurate detection of ash clouds in complicated meteorological conditions will improve the risk assessment of eruptions due to enhanced abilities to detect, forecast, and predict eruptions and eruption pacing. To achieve these goals using remote-sensing imagery, knowledge of the influencing factors on the spectral signature of ash is needed, including the composition of eruption clouds. To understand volcanic hazards and detect eruptions from satellites, it is essential to address this knowledge gap to improve understanding of volcanic systems. This work builds off brightness temperature difference methods of detecting ash plumes, the collected optical properties (refracted index) of volcanic ash, and novel detection methods proposed by Arias et al. (2019) and spectral mixing models. This work also assesses the impact of clean volcanic class crystals versus glass grains with precipitated salts, which has been shown to impact the spectral properties of a volcanic plume. Ash detection models incorporate the Spectrally Enhanced Cloud Objects (SECO) algorithm with lab-measured reflectance of ash samples from the 2008 eruption of Okmok as the training dataset. To test the robustness of the model, it is compared to the three-band ash detection method, as it has already been proven that the two-band method originally posited is too simplistic and comes with flaws in distinguishing dust and ash particles. Although most volcanic ash detection models incorporating the refractive index of volcanic ash generally assumes andesite as the composition of erupted products, the wavelength of ash particles varies from 1.38 to 1.66 m, dependent on silica (SiO2) content of erupted ash products. These slight variations in wavelength due to composition become problematic in detecting eruptive plumes, making the 2008 Okmok eruption a good candidate in testing as the dominate composition of eruptive products is basaltic andesite therefore containing less SiO2 than existing detection models.