Clarification on Filtering Tides within ESPC-D product

[Luana F. Bueno]

Hi Alan,

To continue using ESPC-D forecast results as BC for regional models with tides, it’s crucial to filter tides from all variables. Initially, I thought a daily hourly mean of all variables would be sufficient to eliminate the tidal effects. However, you are providing the 3Z variables at three-hour intervals. The Surf_el variable is available hourly, which allows me to create a daily mean, but I can use the Steric SSH that already filters tides.

Could you please advise on the best method to filter the tides from the new product to keep using it as BC for regional model with tides? Would a daily three-hour mean of 3Z variables be sufficient to filter the tides?

I do appreciate your help. 

Rgerads, 
Luana F Bueno

[Alan Wallcraft]

As you say, Steric SSH is the best way to get a tide free SSH but if you wanted to use a "daily" mean then a 24 hour trapezoidal rule (25 fields) extended to 2xM2 frequency (24.84 hrs) is best.  Instead of the 1st and last field being scaled by 0.5 to get 24 hours they are scaled by 0.92.

For 3 hrly fields the best I can come up with is the composite Simpson's rule over 9 fields with a trapezoidal extension to 24.84 hrs.  This can be done with one NCO ncra command, e.g.:

setenv F1 US058GCOM-OPSnce.espc-d-031-hycom_fcst_glby008
foreach ymd ( 20240810 )
# does not allow for month end
  setenv ymd1 `echo $ymd | awk '{printf("%8d\n",$1+1)}'`
  /bin/rm -f ${F1}_${ymd}12_3hrMN_ssh.nc
  ncra -w 1.4494,3.9706,2.0,4.0,2.0,4.0,2.0,3.9706,1.4494 ${F1}_${ymd}12_t0000_ssh.nc ${F1}_${ymd}12_t0003_ssh.nc ${F1}_${ymd}12_t0006_ssh.nc ${F1}_${ymd}12_t0009_ssh.nc ${F1}_${ymd}12_t0012_ssh.nc ${F1}_${ymd}12_t0015_ssh.nc ${F1}_${ymd}12_t0018_ssh.nc ${F1}_${ymd}12_t0018_ssh.nc ${F1}_${ymd1}12_t0000_ssh.nc ${F1}_${ymd}12_3hrMN_ssh.nc
end

Let me know if you find a better quadrature to use.

The above is from 12Z to 12Z from the "archive" best dataset.  If I was doing this for the forecast I would probably skip the first 12 hours and do 00Z to 00Z means.

Alan.

[Alan Wallcraft]

There is a bug in the above ncra command (t0018 appears twice), which should be:

ncra -w 1.4494,3.9706,2.0,4.0,2.0,4.0,2.0,3.9706,1.4494 ${F1}_${ymd}12_t0000_ssh.nc ${F1}_${ymd}12_t0003_ssh.nc ${F1}_${ymd}12_t0006_ssh.nc ${F1}_${ymd}12_t0009_ssh.nc ${F1}_${ymd}12_t0012_ssh.nc ${F1}_${ymd}12_t0015_ssh.nc ${F1}_${ymd}12_t0018_ssh.nc ${F1}_${ymd}12_t0021_ssh.nc ${F1}_${ymd1}12_t0000_ssh.nc ${F1}_${ymd}12_3hrMN_ssh.nc

Alan.

[Alan Wallcraft]

I have added espcd_mean9 to automate forming the daily mean.

Alan.

[Luana F. Bueno]

That's brilliant, Alan. 
I do appreciate it. 

Warm regards, 
Luana F Bueno