The following is the text of email from Eric Rogers (NCEP/EMC) of 2/12/2004.

I have tweaked my 8-day eta web graphics to my satisfaction. You can find it at http://wwwt.emc.ncep.noaa.gov/mmb/mmbpll/etapll8day/. Each link in the left side panel will bring up a 90-192 h loop. There are two Eta, GFS comparison loops:

DGEX, GFS: side-by-side loop of the Eta and GFS initialized at the same time DGEX, previous GFS : side-by-side loop of the Eta and the older GFS run which was used as lateral boundary conditions for the Eta extension. This comparison is the key one because the DGEX will tend to resemble the 6-h old GFS more than the on-time GFS, especially at the later times when the run-to-run variation of the GFS at days 7+ can be quite large.

A note on lateral boundary tendencies for the DGEX. Assume for this discussion that the DGEX cycle is 00Z. The GFS output frequency is 3-h to 180h, and 12-h thereafter. So until the DGEX run gets to 174 h, it has a 3-h lateral boundary condition update using the previous 18z GFS run. After DGEX forecast hour=174, the next available GFS forecast is the 192-h forecast from the previous 12z GFS (valid at hour 180 of the DGEX run). For the DGEX 180-186 h forecast segment, we get the 192-h forecast from the previous 18z GFS, and for the final 186-192 h piece, we get the 204-h forecast from the previous 12z GFS. So the boundaries are updated at 6-h intervals from 174-192 h. If you look at the DGEX vs. previous GFS" loop, you'll see on the gif file the initialization time of each run, and beyond DGEX forecast hour 174, this flip-flops for the GFS between the previous 18z and 12z GFS runs.

I will be saving the last week's worth of runs on the web page. There are old runs at these links now but they do not have the DGEX vs. previous GFS comparisons, I just got that working today.

I should point out that I am actually starting the DGEX run from the 78-h Eta-12 forecast. I decided to back up 6-h to allow adjustment on the smaller 12km domain to be more or less complete before forecast hour 90. When starting at hour 78, it turns out the DGEX 84-h forecast looks reasonable, so I can add that forecast to this web page (although it involves manually editing each html looping script) if it is deemed necessary. Starting from 78-h would be my preferred configuration unless there is not enough room in the production slot. I ran some tests comparing 78-h vs. 84-h start times and saw little difference by days 7 and 8.

This is running off of our parallel Eta-12 forecast (http://wwwt.emc.ncep.noaa.gov/mmb/mmbpll/etapllsup12.etax_opstest/),

NOT the production Eta. If my parallel forecast does not run, the DGEX won't either. We are only running the parallel Eta-12 forecast at 00z, which is why there are no 12z runs. NCO will implement the latest Eta-12 change package sometime in March, and at that time I can run some 12z forecasts.

Finally, I will entertain reasonable requests for changes to the web page (like additional variables).