SST update and real.exe interpolation

Any issues with the actual running of the WRF.

SST update and real.exe interpolation

Postby ivicaj » Fri Oct 26, 2012 8:40 am

Dear,
I am running 341 version and have faceted strange problem regarding SST update during the simulation.
Using ECMWF files which have SST inside, changing in time, I want to run WRF for a long period and need sst_update. There is no external SST data, just the one from ECMWF grib file (!) which is better than frozen state at the start.
On the other hand, metgrid.exe creates met* files that look reasonable, SST field inside has values in the range which I would expect say round 20C +/- 3(of course they are in K), landsea mask is fine etc...
Then we arrive to the problem because real.exe uses these files to create lower boundary wrflowinp files which look wrong close to the coast, like it never used mask during interp. In my example I have values for SST oscillating for 10K reaching values like 32C, obviously leaking from the ground (Land can have temps about 35C but sea is never above 22C in the period of time), and there are no values like that in the met_em files as well.
I am suspecting on the way real.exe is doing interpolation from the met_em files.

Possible hack that I could do is to use externally available SST data (i.e. ocean model) and just interpolate them (carefully, using only wet points from ocean model onto wet points at wrflowinp file).
In the case I do not have ocean data I still could do better interpolation from met_em files taking into account land/sea mask. Does it make sense?

Think there is possibility that other users having CFL problems with sst_update = 1 are basically having wrong values with big difference for SST fields simply because real.exe interpolation scheme.

Cheers,
Ivica
ivicaj
 
Posts: 8
Joined: Tue Sep 09, 2008 9:37 am

Re: SST update and real.exe interpolation

Postby ivicaj » Mon Oct 29, 2012 5:34 am

I got a bit confused with LAND/SEA masks;
there are 2 masks dealing with it: LANDMASK and LANDSEA (the one only in met_em files), which are not the same over the coastal region, as I would like to think of. I know that for inland lakes it is not the SEA but the LANDMASK valid option, however in a surface driver I can find that XLAND_SEA is used...
So, in that sense I can not find where and why is LANDSEA used?

Should LANDMASK and LANDSEA mask (in met_em) be the same over the sea or in strict coastal zone?

Obviously, my naive flowchart starting with met_em files holding SST only on the LANDSEA mask is not compatible with LANDMASK used by the others, like real.exe creating mismatch in a coastal zone.

If someone has will to share some light I am grateful,

cheers
Ivica
ivicaj
 
Posts: 8
Joined: Tue Sep 09, 2008 9:37 am

Re: SST update and real.exe interpolation

Postby ivicaj » Mon Oct 29, 2012 6:35 am

Think I got to the solution, if anybody has the similar problem.
The key is to change option in METGRID.TBL for SST so it is not using LANDSEA mask but LANDMASK, in my case SST entry looks like:
========================================
name=SST
missing_value=-1.E30
flag_in_output=FLAG_SST
interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
fill_missing=0.
masked=land
interp_mask=LANDMASK(1)
========================================
And now there is no land values leaking into the SST giving unreasonable values actually originating from the land.

Cheers.
Ivica
ivicaj
 
Posts: 8
Joined: Tue Sep 09, 2008 9:37 am


Return to Runtime Problems

Who is online

Users browsing this forum: No registered users and 7 guests