Samm Elliott

FYI @domcharrier - We use `!$acc declare` very frequently in our code.

The vast majority of the code only uses `!$acc routine seq`. I wouldn't prioritize implementing the routine directive for gang/worker/vector at this point.

@cenlinhe - Yes `EMISSI` is passed to `SFLX` and modified for each iteration of the loops from lines 734-1567 but the problem is that it isn't initialized for each iteration...

> I think the emissi in Line 1693 should be emiss_rural(i,j). That would make more sense.

If you all don't mind I would appreciate your advice regarding these options. Let me know if this is not the proper channel for discussing such topics. My goal internally...

@weiwangncar > I tend to agree with you on your proposed fix to initialize qsatu in the loop you identified. Is setting it to zero the correct thing to do?...

@weiwangncar - I just checked and it looks like adding `QSATu(K) = 0.0` in the loop at [phys/module_cu_mskf.F line 4365](https://github.com/wrf-model/WRF/blob/21c72141142fc6c8d203d2bf79f1990e45a0aef8/phys/module_cu_mskf.F#L4365) also produces bit-for-bit results and has very similar effects to...

@weiwangncar - That makes sense. Thanks for the explanation! Should I create a pull request?