Stefano Corradini

Satellite microwave (MW) and millimetre-wave (MMW) passive sensors can be used to detect volcanic clouds because of their sensitivity to larger volcanic particles (i.e., size bigger than 20 µm). In this work, we combine the MW-MMW observations with thermal-infrared (TIR) radiometric data from the Low Earth Orbit (LEO) spectroradiometer to have a complete characterisation of volcanic plumes. We describe new physical-statistical methods, which combine machine learning techniques, aimed at detecting and retrieving volcanic clouds of two highly explosive eruptions: the 2014 Kelud and 2015 Calbuco test cases. For the detection procedure, we compare the well-known split-window methods with a machine learning algorithm named random forest (RF). Our work highlights how the machine learning method is suitable to detect volcanic clouds using different spectral signatures without fixing a threshold. Moreover, the RF model allows images to be automatically processed with promising results (90% of the area correctly identified). For the retrieval procedure of the mass of volcanic particles, we consider two methods, one based on the maximum likelihood estimation (MLE) and one using the neural network (NN) architecture. Results show a good comparison of the mass obtained using the MLE and NN methods for all the analysed bands. Summing the MW-MMW and TIR estimates, we obtain the following masses: 1.11 ± 0.40 × 1011 kg (MLE method) and 1.32 ± 0.47 × 1011 kg (NN method) for Kelud; 4.48 ± 1.61 × 1010 kg (MLE method) and 4.32 ± 1.56 × 1010 kg (NN method) for Calbuco. This work shows how machine learning techniques can be an effective tool for volcanic cloud detection and how the synergic use of the TIR and MW-MMW observations can give more accurate estimates of the near-source volcanic clouds.

In this study, we focus on the eruption of Mount Etna on Christmas 2018, which emitted great amounts of SO2 from 24th to 30th December into the free troposphere. Simulations based on two different estimations of SO2 emission fluxes are conducted with the chemistry-transport model MOCAGE in order to study the impact of these estimations on the volcanic plume modeling. The two flux emissions used are retrieved (1) from the ground-based network FLAME, located on the flank of the volcano, and (2) from the spaceborne instrument SEVIRI onboard the geostationary satellite MSG. Multiple spaceborne observations, in the infrared and ultraviolet bands, are used to evaluate the model results. Overall, the model results match well with the plume location over the period of the eruption showing the good transport of the volcanic plume by the model, which is linked to the use of a realistic estimation of the altitude of injection of the emissions. However, there are some discrepancies in the plume concentrations of SO2 between the two simulations, which are due to the differences between the two emission flux estimations used that are large on some of the days. These differences are linked to uncertainties in the retrieval methods and observations used to derive SO2 volcanic fluxes. We find that the uncertainties in the satellite-retrieved column of SO2 used for the evaluation of the simulations, linked to the instrument sensitivity and/or the retrieval algorithm, are sometimes nearly as large as the differences between the two simulations. This shows a limitation of the use of satellite retrievals of SO2 concentrations to quantitatively validate modeled volcanic plumes. In the paper, we also discuss approaches to improve the simulation of SO2 concentrations in volcanic plumes through model improvements and also via more advanced methods to more effectively use satellite-derived products.