Optimal forecasting of natural processes with uncertainty assessment

The problem of optimal forecasting of environmental changes induced by various factors is discussed. The proposed technique is based on variational principles and methods of the sensitivity theory with allowance for uncertainties in mathematical models and input data. Optimal forecasting is understood as forecasting where the estimates of cost functionals are independent of variations of the sought state functions. In addition to state functions, the forecasted characteristics include risk and vulnerability functions for receptor areas and quantification of uncertainties.