Leila,

Thanks. Below is summary what I did over weekend. I cc colleagues, just to inform them and get eventual feedback before telecon on Friday.

(1) Your "wsa.gong.fits" (in "swpc....0000/wsa....inputs") has carrlong=304 while SWPC log file suggest that the file with carrlong=293 was used.

(2) I found "wsa_vel_21.5rs_2017072020_gong.fits" in my repository, it has carrlong=293 (and the same Julian date at center of map) which I believe is the file used by SWPC.

(3) It is strange that SWPC uses observations from future, i.e., after 2017-07-20T00, to make predictions...

(4) Attached slides show results computed by v2.9 (solid line) compared with those extracted from the SWPC "wsa_enlil.latest.suball.nc" file (dashed line).

(5) Temporal profiles at Earth:
- structures have basically the same arrival; there is no shift in time as in Peter's plots;
- amplitudes visibly differ.

(6) Differences in temporal profiles can be tracked to differences in the boundary values. It is strange (if we use the same WSA files). I will need to compare relevant source codes.

(7) I adjusted the code to read the original bnd.nc from SWPC (normally, v2.9 cannot read old bnd.nc because of refdate/rundate, rotation, etc. keywords). Results are presented on the right side in the attached slides.

(8) Now, temporal profiles almost match :-). There might be, among others, very slight differences in interpolation of values onto the Earth position. Thus it would be better to compare the radial profiles.

(9) However, the SWPC "....suball.nc" file does not allow to extract reasonable values for that.

Dusan