| Slide 1 |
| If we assume a perfect model, we can grossly underestimate the errors |
| We compare several methods to handle model errors |
| Slide 4 |
| Slide 5 |
| "1a." |
| 2. Dee and daSilva bias estimation scheme (1998) |
| Slide 8 |
| Slide 9 |
| Slide 10 |
| Diurnal model errors |
| Slide 12 |
| Slide 13 |
| Correct state-dependent model errors |
| Impact of model error, and different approaches to handle it |
| Simultaneous estimation of inflation and observation errors |
| Motivation |
| MK method to estimate the inflation parameter (Miyoshi 2005, Miyoshi&Kalnay 2005) |
| Diagnosis of observation
error statistics (Desroziers et al, 2006, Navascues et al, 2006) |
| Simultaneous estimation of inflation and observation errors |
| Tests within LETKF with Lorenz-96 model |
| Slide 22 |
| Slide 23 |
| Slide 24 |
| online estimated observational errors |
| Slide 26 |
| Slide 27 |
| Summary |
| A few more slides |
| Adaptive sampling with the
LETKF-based ensemble spread Junjie Liu |
| 500hPa zonal wind RMS error |
| Slide 32 |
| Analysis sensitivity study within LETKF |
| Slide 34 |
| Comparison of
ensemble-based and variational-based data assimilation schemes in a
Quasi-Geostrophic model. Shu-Chih Yang et al. |
| Analysis increment (color
shaded) vs. dynamically fast growing errors (contours) |
| Slide 37 |
| No-cost LETKF smoother |
| LETKF minimizes the errors of the day and thus provides an excellent first guess to the 3D-Var analysis |