Date:

2014-10-20 14:00

Room:

Lecturer:

Affiliation of External Lecturer:

University of Vienna, Department of Meteorology and Geophysics

Department:

Inverse modeling is a formal approach for estimation of uncertain parameters of a system under consideration given some relevant observations. In atmospheric transport modelling it is successfully applied to estimation of a source term (to provide a source term hypothesis) given concentration measurements. General foundation for the inverse modeling is provided by Bayes’ theorem, however, the problem leads to an optimization problem under some assumptions. A source term hypothesis can be inferred via minimizing a cost function measuring a mismatch between measured data and model simulations. These simulations enter the problem in the form of so-called source-receptor sensitivities (SRS) describing possible contribution of source(s) to measurements. SRSs are usually aggregated in a SRS matrix. Given the matrix and a source hypothesis, resulting receptor values can be obtained simply by matrix-vector multiplication which is beneficial in the case of applied variational approach. Real-word problems are often ill-conditioned so we have to employ a regularization to obtain a unique and stable solution. Different inversion scenarios will be discussed covering situations with both known and unknown source locations. Methodology will be demonstrated on analysis of radioxenon detection from the international monitoring network.