I'm trying to do a cycled run with two assimilation points. The steps are the following:
1. WPS initialization on 12Z GFS data
3. WRFDA 3DVAR obs assimilation at 12Z analysis time (obsproc, da_wrfvar, da_update_bc)
4. 3 hour WRF run, from 12Z to 15Z, wrfrst output file created at 15Z
5. Set namelist.input's start time and analysis time to 15Z
6. WRDDA 3DVAR obs assimilation at 15Z analysis time (using the wrfrst as first guess):
- obsproc and da_wrfvar runs normally, but da_update_bc gives "Wrong data for Boundary."
I also tried rerun real.exe after step 5 (with namelist start time set to 15Z - generating new wrfbdy file) with no luck. What step did I miss?