I've used the following two approaches, but in either case I am a) not sure the initialisation data is used at all; b) whenever it looks that it might have been used (that is, wildfire smoke related species are distributed in places outside the fire locations for the new run), the distribution at early hours in the new output differs noticeably from the distribution at the end of the previous run; c) in any event, the first frame of the new run does not contain any initialization data (though levels of the species I'm interested in are not 0 -- just very low levels): data only appears in the second frame (hour 1).
- Approach 1: I linked the last wrfout file of the previous run (containing a lone last hour) to wrf_chem_input_d01 and set the following in namelist.input:
auxinput12_inname = 'wrf_chem_input_d<domain>',
auxinput12_interval_m = 86400,
io_form_auxinput12 = 2,
frames_per_auxinput12 = 1
- Approach 2: I linked all previous wrfout files to wrf_chem_input_d01_yyyy-mm-dd_HH:MM:SS and set the following in namelist.input:
auxinput12_inname = 'wrf_chem_input_d<domain>_<date>',
auxinput12_interval_m = 86400,
io_form_auxinput12 = 2,
Any input on how to correctly configure this initialization would be welcome, in particular how to start the initialized background at the first frame of the new output file (or what to do otherwise).