Data assimilation algorithm based on sensitivity operator for a three-dimensional atmospheric transport and transformation model
A.V. Penenko1,2, A.V. Gochakov3, P.N. Antokhin2,4
1Institute of Computational Mathematics and Mathematical Geophysics of the Siberian Branch of the RAS, Novosibirsk, Russia
2Russian State Hydrometeorological University, St. Petersburg, Russia
3Siberian Regional Hydrometeorological Research Institute, Novosibirsk, Russia
4V.E. Zuev Institute of Atmospheric Optics of Siberian Branch of the Russian Academy of Science, Tomsk, Russia
Keywords: data assimilation, source identification, advection-diffusion-reaction, sensitivity operator, adjoint equations
Abstract
Three-dimensional transport and transformation models make it possible to take into account the vertical heterogeneity of atmospheric processes. However their use requires setting a large number of parameters and significant computing resources, especially when solving inverse and data assimilation problems. A new data assimilation algorithm for a three-dimensional transport and transformation model with unknown emission sources is presented, which uses an approach based on sensitivity operators and ensembles of solutions of adjoint equations implemented in the IMDAF inverse modeling system for distributed memory computers. When tested in a realistic Baikal region scenario, the algorithm enabled, based on the data of integrated vertical measurements simulating remote sensing data, reducing the error in the concentration field by 15%. With the given vertical level of the source location, the errors in the concentration field and in the source were reduced by 93% and 85%, respectively.
|
