A NEURAL NETWORK APPROACH FOR VOLCANIC MONITORING OF SULPHUR DIOXIDE USING HYPERSPECTRAL REMOTE SENSED DATA

摘要

This paper describes an application of ANN for thernsimultaneous estimation of the columnar content andrnheight of the SO2 plume from volcanic eruptions usingrnhyperspectral remotely sensing data. ANN have beenrntrained using all IASI channels between 1000–1200 andrn1300–1410 cm?1, as inputs, and the correspondingrnvalues of SO2 amount and plume's height obtained usingrnthe Oxford retrieval scheme as outputs. As a case studyrnwe have chosen the Eyjafjallaj?kull volcano (Iceland),rnin particular the eruption took place during the monthsrnof April and May 2010, which had an enormous impactrnon the world economy. ANNs have been validated onrnsome independent data sets belonging to the samerneruption and also on IASI images of Grímsv?tnrneruption, occurred on May 2011. The results havernprovided values of RMSE between ANN outputs andrntargets always less than 20 DU for SO2 and 200 mb forrnheight, so demonstrating the good performance inrnretrieval achieved by the ANN technique.