Inverse modelling of volcanic SO2 emissions using the 4D-var method

Abstract

This study compares two approaches for data assimilation in forecasting volcanic plumes of sulphuric dioxide released in the eruption of Grimsvötn in 2011. The first data assimilation approach is source term inversion, where the data are used for estimating an effective source term. In the second approach, the assimilation proceeds in cycles, and the observations are used for updating the initial condition. In both cases, the SO2 retrievals by the OMI instrument are assimilated. The results indicate that the source term inversion is more effective at constraining the three-dimensional structure of the plume. However, as the transport distance increases, updating the initial condition provides a better fit to the column observations.