WRF Error - FATAL CALLED FROM FILE: grid%tsk unreasonable

[shahbaz757]

Hi Laetis,

I hope you're doing well.

I am trying to run the WRF model with the Free IFS archive (ECMWF Data Store (ECPDS)) at 0.25-degree resolution.

I think I have found most of the necessary variables and pressure levels to initialise the WRF model.

I followed the instructions to pre-possess the grib2 files before running WPS from this threat - ECMWF open data 0p25 grib2 Vtable?

I first changed the change the compression format of grib2 files and changed the change soil variables with -set_var and -set_level so it can be read the ungrib (because in ecmwf files soil depths appear in meters instead of cm!)

I was able to successfully run the WPS (georid.exe, ungrib.exe, and metgrid.exe)

However, When I try to run the real.exe in WRF I am getting the following error.

d01 2024-06-24_00:00:00 No average surface temperature for use with inland lakes
Assume Noah LSM input
d01 2024-06-24_00:00:00 forcing artificial silty clay loam at 1315 points, out of 100440
error in the grid%tsk
i,j= 1 1
grid%landmask= 0.00000000
grid%tsk, grid%sst, grid%tmn= 0.00000000 0.00000000 0.00000000
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2987
grid%tsk unreasonable
-------------------------------------------


I have checked the TSK variable in the met_em nc file and it looks fine to me.




Can you please help me resolve this issue? I am attaching all error files and namelists I used to run the model.

Thank you
Shahbaz Khan

 

[Ming Chen]

Shahbaz,
First, would you please clarify whether the IFS data you dowwnloaded is global or regional data? If it is regional , does it cover the entire WRF domain?
Second, please tell what variables are included in your IFS data? The error message seems to imply that you don't have a correct land mask.

 

[shahbaz757]

Hi Ming,

The IFS data has global coverage (
Grid coordinates :
1 : lonlat : points=1038240 (1440x721)
lon : -180 to 179.75 by 0.25 degrees_east circular
lat : 90 to -90 by -0.25 degrees_north )


Here is the list of variables available in the grib2 file.

Variables and Levels in the GRIB2 File​

Variable (shortName)	Level (hPa)
tp	0
2d	2
u	150, 1000, 400, 600, 50, 100, 250, 500, 200, 925, 850, 300
v	150, 1000, 400, 600, 50, 100, 250, 500, 200, 925, 850, 300
t	1000, 850, 925, 500, 700, 250, 200, 150, 100, 50, 400, 300
r	850, 1000, 925, 500, 200, 700, 400, 300, 150, 600, 50, 100, 250
cape	0
gh	200, 500, 925, 300, 700, 1000, 850, 400, 250, 100, 150, 50, 600
skt	0
ssrd	0
sp	0
msl	0
2t	2
10v	10
10u	10
swvl1	0
swvl2	7
swvl3	28
swvl4	100
stl2	7
stl3	28
stl4	100
w	700, 500, 1000, 850, 300, 250, 400, 100, 150, 600, 50, 200
q	850, 500, 700, 250, 1000, 925, 200, 300, 100, 400, 150, 50, 600
lsm	0
strd	0
str	0
d	925, 700, 850, 300, 1000, 200, 500, 400, 250, 150, 600, 100, 50
vo	500, 700, 925, 850, 300, 1000, 200, 400, 250, 100, 600, 150, 50
ro	0
ttr	0

 

[shahbaz757]

grib_ls output -
20240623000000-102h-oper-fc.grib2
shortName level
cape 0
sp 0
10v 10
gh 200
gh 300
t 400
u 400
v 400
v 50
u 50
swvl1 0
t 850
t 700
gh 1000
r 700
100v 100
gh 850
ssrd 0
r 300
r 850
100u 100
gh 400
r 200
t 100
r 400
gh 700
st 0
skt 0
tp 0
r 500
u 700
v 700
r 925
u 600
v 600
gh 250
t 500
t 925
r 1000
r 250
t 250
w 700
gh 100
ssr 0
u 100
v 100
msl 0
u 150
v 150
t 300
t 1000
gh 500
u 250
v 250
t 200
gh 925
v 300
u 300
2d 2
u 925
v 925
u 200
v 200
gh 150
w 850
swvl2 7
r 150
q 700
gh 50
w 925
q 925
w 200
w 500
gh 600
q 1000
u 850
r 600
r 100
v 850
w 300
t 50
t 150
swvl3 28
t 600
10u 10
2t 2
q 500
u 1000
v 1000
w 250
stl3 28
w 400
swvl4 100
tcwv 0
u 500
w 1000
stl2 7
v 500
q 850
stl4 100
strd 0
w 100
q 250
w 150
r 50
q 300
q 200
q 400
d 925
q 150
q 100
w 600
str 0
lsm 0
q 600
vo 500
vo 700
d 1000
w 50
d 850
vo 925
q 50
d 700
d 200
d 500
vo 1000
vo 200
d 400
d 250
vo 300
ro 0
d 100
d 300
vo 400
d 600
vo 250
ttr 0
d 150
vo 850
vo 100
vo 150
d 50
unknown 0
vo 50
vo 600
145 of 145 messages in 20240623000000-102h-oper-fc.grib2

 

[shahbaz757]

I suspect that the Sea Surface Variable (SST) is missing in the IFS grib files. Is that might be the reason I am getting this error?
If so, can I use those variables from GFS data to run the model? How can I achieve that?

 

[Ming Chen]

One critical variable, the landsea mask, is missing in your IFS data. Note that landmask is important because it determines whether the land information can be correctly interpolated to WRF grids.

I suppose skt is 'skintemp'? if so, then you don't need to worry that SST is missing. This is because WRF can use skt to represent SST.

 

[shahbaz757]

Hi Ming,

I can see the LANDMASK and LANSEA variables in the met_em nc intermediate file.








However, I can't see the SKINTEMP in the met_em nc file. it is available in the the input IFS grib2 files, maybe ungrib not able to decode it. How can I find the right GRIB2 code for this variable?

Also, how can I upload the met_em nc intermediate files here so you can have a look at them? when I try it says the file is too big for upload.

 

[Ming Chen]

Can you find SST in your met_em file? Does it have valid values over land?

I believe that the issue you have is caused by landsea mask in the IFS data. We are aware that some ECMWF data have issues that can cause trouble for WPS. Please see the document:

Section 2.1.3.1 Land-Sea Mask - Forecast User Guide - ECMWF Confluence Wiki

confluence.ecmwf.int

In this case, I would suggest that you use SKINTEMP. To figure out the GRIB2 parameters for this variable, please take a look at:

Chapter 3: WRF Standard Initialization