Inverse problems in atmospheric science and their application

This paper reviews various kinds of inverse problems in atmospheric science and oceanography, and introduces powerful methods to treat these problems-variational data assimilation (VAR) and improved discrepancy principle, and discusses some essential difficulties in VAR. Due to the ill-posedness of these problems, the regularization method is also applied, i.e., additional terms are added to the cost functional as a stabilized functional with physical meaning. Inversions of four specific problems, such as the inversion of one-dimensional sea temperature model, the inversion of parameters in an ENSO cycle model, the inversion of wind field with single-Doppler data, and the inversion of satellite remote sensing, indicate that, adoption of the regularization method in VAR will overcome the ill-posedness, constrain calculational oscillations in iteration, and speed up convergence of solutions