Some limitations of statistical interpolation

Author:Pierre Gauthier
 

Table of content                Click here to start      Return to Main Page

  • Some limitations of statistical interpolation
  • 4D variational data assimilation (4D-Var)
  • Preconditioning through a change of variable
  • Operations involved in a single iteration of 4D-Var
  • Equivalence between the Kalman filter and 4D-Var
  • Comparison of analysis increments obtained with 3D-Var vs. that of 4D-Var (6-h) (single surface observation)
  • Vertical cross-section of the analysis increment
  • 4D Variational Data Assimilation
  • Definition and properties of an adjoint operator
  • Computation of the gradient using the adjoint model
  • Tangent Linear model and Adjoint Model (LeDimet and Talagrand, 1986)
  • Example: the Lorenz (1963) model
  • Adjoint of the propagator R(t0,ti) of the TLM
  • Steps involved in one iteration of 4D-Var
  • Error growth and the TLM and adjoint models
  • Other applications of the adjoint model
  • Sensitivity analyses
  • Sensitivity analyses (S. Laroche, J. Morneau and M. Tanguay)
  • 120 hr forecast and verifying analysis GZ 500
  • 24-h forecast error of the operational model (Geopotential at 500 hPa)
  • Corrections brought by the sensitivity analysis at 0 hr (Geopotential at 500 hPa)
  • Forecast error at 24-h (GZ 500 hPa)
  • Verification of the 5-day forecast, valid on 6 October 2000 at 00 UTC
  • Corrections brought by the sensitivity analysis at 0 hr (Geopotential at 500hPa)
  • Vertical cross-section of the changes brought by the sensitivity analysis at 0-hr over the centre of Northern Pacific (valid on 1 October 2000)
  • Verification of the 5-day forecast based on the sensitivity analysis
  • Summary
  • Time series of precipitation rates averaged over the Northern Hemisphere (Gauthier and Thepaut, MWR 2001)
  • Current issues...


    		•