A neural network approach for monitoring of volcanic SO2 and plume height using hyperspectral measurements

In this study two neural networks were implemented in order to emulate a retrieval model and to estimate the sulphur dioxide (SO2) columnar content and cloud height from volcanic eruption. ANNs were trained using all Infrared Atmospheric Sounding Interferometer (IASI) channels in Thermal Infrared (TIR) as inputs, and the corresponding values of SO2 content and height of volcanic cloud obtained using the Oxford SO2 retrievals as target outputs. The retrieval is demonstrated for the eruption of the Eyjafjallajökull volcano (Iceland) occurred in 2010 and to three IASI images of the Grímsvötn volcanic eruption that occurred in May 2011, in order to evaluate the networks for an unknown eruption. The results of validation, both for Eyjafjallajökull independent data-sets, provided root mean square error (RMSE) values between neural network outputs and targets lower than 20 DU for SO2 total column and 200 mb for cloud height, therefore demonstrating the feasibility to estimate SO2 values using a neural network approach, and its importance in near real time monitoring activities, owing to its fast application. Concerning the validation carried out with neural networks on images from the Grímsvötn eruption, the RMSE of the outputs remained lower than the Standard Deviation (STD) of targets, and the neural network underestimated retrieval only where target outputs showed different statistics than those used during the training phase.