An algorithm for identifying pollution sources with nonlinear measurement operators
A.V. Penenko1,2, E.V. Rusin1,2, M.K. Emelyanov1,2
1Institute of Computational Mathematics and Mathematical Geophysics of the Siberian Branch of the RAS, Hanty-Mansijsk, Russia
2Federal State Educational Institution of Higher Professional Education "Ugra State University"
Keywords: remote sensing, nonlinear measurement operator, source identification, advection-diffusion, sensitivity operator, adjoint equations, decomposition, atmospheric aerosol, RTTOV, MODIS
Abstract
The problem of atmospheric emission source identification using remote sensing data is considered. An algorithm is proposed for a three-dimensional model of atmospheric pollutant transport and a nonlinear measurement model represented as a “differentiable black box". The algorithm is based on sensitivity operators and ensembles of adjoint equations solutions. It was tested on a realistic scenario for identifying soot sources for the Baikal region with synthetic satellite measurements of the Terra/MODIS platform, which showed its applicability. Additionally, the measurement data decomposition modification of the algorithm is proposed, which made it possible to reduce the relative error of retrieving the source function by 12% compared to the version without decomposition. The results can be used in the development of remote sensing data processing systems.
|
