A new version of a set of mathematical models for investigating transport and transformation of pollutants from natural and anthropogenic sources in the atmosphere is presented. The structure of the models, methods of discretization, and algorithms realizing the various models are constructed for application to direct and inverse modeling in problems of monitoring, forecasting, and ecological planning. Toward this end, a variational principle in combination with decomposition and splitting techniques is used to construct the numerical models. Algorithms for calculating the direct and adjoint trajectories in the source–detector system are described. Examples are given of solutions of the direct and inverse problems for estimating the scales of interactions of the type (source) –(conservation zone) (Lake Baikal).