I am running quite successively some I4DVAR experiments. Also i am still working on understanding all the subtleties of 4DVAR, i am reasonably happy with my results.
I made 3 runs (same domain, same forcing ) with 3 differents datasets (surface velocities)
i run it daily with 8 inner loop and 1 outer.
- First run did 5 month (which is the objective)
- Second run did 2 month and stop during the third month with : CG_DELTA not positive at the 5th inner loop.
- Third run did 3 month and stop during the 4th month : Negative Ritz value found.
So the only differences come from the data sets but when i look at them i see nothing particularly abnormal.
Do you have any insight of why this happens while it use to run smooth up to now ?
Discussion about tangent linear and adjoint models, variational data assimilation, and other related issues.
1 post • Page 1 of 1