ALATNET Seminar on Data Assimilation

Lecture 5

TL/AD Techniques

• 5.1 Adjoint technique - general description
  - definitions and properties of TL/AD operators
  - TL/AD models and their validations
  - use of TL/AD models and operators in a variational analysis

• 5.2: Linearized models (diabatic vs. adiabatic)
  - description of the physical parametrizations for TL/AD models
  - accuracy of the linearization --> evaluation of TL approximation
                                  --> importance of regularization
  - impact of including the physical parametrizations in the linearized model

• 5.3: Practical use of adjoint models
  - physical parametrizations and data assimilation (4D-Var, 1D-Var)
  - sensitivity studies
  - tuning of the physical parametrizations
  - singular vectors

Practical Advanced Tutorial 5

• PAT 5.1
  - some recipes for tangent linear and adjoint coding
  - validation methods
  - practical exercise of coding and validation

• PAT 5.2
  - coding simple example of using adjoint technique for sensitivity
  - using adjoint for tuning of the physical parametrizations

• PAT 5.3
  - solving a simple analysis problem

• PAT 5.4:
  - 1D-Var assimilation of surface parameters or precipitation or clouds