Inversion techniques

Inversion consists in estimating the gridded (model) state \(X\) assuming a vector of observations, noted \(Y\) defined as:

\[Y = HX+\epsilon\]

where \(H\) is a linear observation operator between the reconstruction grid space and the observation space and \(\epsilon\) is an independent observation error.

Different inversion techniques (static and dynamic) are implemented in MASSH and the codes are gathered in the mapping/src/ana.py script.

In the configuration file, you can select a specific analysis by setting the value of name_analysis. If None value is set, then a forward propagation of the selected model is performed.

Hereafter we provide some details on the main inversion techniques used in MASSH:

• Optimal Interpolation
  • Method
  • Configuration parameters
• Back and Forth Nudging
  • Method
  • Time windows
  • Nudging strategy
  • Configuration parameters
• Four-Dimensional Variational