The evaluation of the atmospheric pollutants emission on the basis of modelling and measurements

The problem of the evaluation of the emission to the atmosphere from several pollution sources is considered. Atmospheric monitoring data are taken as a primary source of information. Prescribed estimates of the emission of every source are accepted as a-priory given data, which should be corrected. The problem is assumed to be linear, i.e. the pollutant is considered as inert substance. But even in this case the direct solution via inversion of the influence matrix can not be used because of high sensitivity of the obtained solution to the uncertainties of the input data and matrix itself. Another previously suggested statistical methodology was based on the maximum probability criteria for the emission estimates. This method involves an arbitrarily specified parameter, which characterizes subjective trust to the a-priory given information about the emission. Modification of this parameter can drastically affect the results. The paper is dedicated to further development of the methodology and suggests the way of calculation of the optimal value of the parameter. Several extreme cases are considered.