Complex hybrid models combining deterministic and machine learning components as a new synergetic paradigm in numerical climate modeling and weather prediction

A new type of numerical models, complex hybrid environmental models (CHEMs) based on a combination of deterministic and machine learning model components, is introduced and developed. Conceptual and practical possibilities of developing CHEM, as an optimal synergetic combination of the traditional deterministic/first principles modeling and machine learning components (like accurate and fast neural network emulations of model physics or chemistry processes), are discussed. An example of developed CHEM (a hybrid climate model) illustrates the feasibility and efficiency of the new approach for modeling extremely complex multidimensional systems.