Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

size of wrfout file decrease

Q

qingling

Member
Size of wrfout file decrease from 379M to 79M using WRFV4.6.1 as show in the figure blow. Moreover, model was't interrupted. It is so confused.
if you have some ideals about that, looking forward to your reply! Thanks a lot!
1763003392506.png
 
Can you look at the two files with large size difference and check whether they contain same variables? A quick look using ncview can also tell whether the results are reasonable. I am suspicious that something may already went wrong after 01-25_07:00:00, although the model continued its integration.

Please keep me updated about what you found. Thanks.
 
Thank you for your relpy! I checked two files with large size difference. They have same variables,but the value of the smaller file (wrfout_d02_2001-01-25_09_00_00) is NaN as show in the figure below.
1763017377060.png1763017388929.png

In addition, I think you are right. I checked the rsl.error.0000 and found the abnormal appearance that some NaN values are appeared as show in the figure blow.
1763017964410.png
 
This explains why large differences occur in the file size. Apparently something went wrong in this case.

Is this a global WRF run? What is your forcing data? Please upload your namelist.input for me to take a look.
 
Ming Chen said:
This explains why large differences occur in the file size. Apparently something went wrong in this case.

Is this a global WRF run? What is your forcing data? Please upload your namelist.input for me to take a look.
Click to expand...
Thank you for your reply. This is a regional WRF (in China) run using ERA5 forcing. I have been attached namelist.input
 

Attachments

  • namelist.input
    5.9 KB · Views: 5
Ming Chen said:
This explains why large differences occur in the file size. Apparently something went wrong in this case.

Is this a global WRF run? What is your forcing data? Please upload your namelist.input for me to take a look.
Click to expand...
Hi Ming Chen. I tried to set mp_physics = 8 and successfully run as show the figure below.
1763113466964.png
 
Your namelist.input looks fine to me except that you should turn off cumulus scheme for D02, i.e.,

cu_physics = 2, 0,


WSM 3-class simple ice scheme is well tested and I don't think it could be the reason that caused model crash in your previous run. I am suspicious that somehow the cumulus scheme in D02 may interact with WSM3.

I also want to note that, personally I think Thompson scheme performs better than WSM3, especially for high resolution case.
 
Last edited:
Ming Chen said:
Your namelist.input looks fine to me except that you should turn off cumulus scheme for D02, i.e.,

cu_physics = 2, 0,


WSM 3-class simple ice scheme is well tested and I don't think it could be the reason that caused model crash in your previous run. I am suspicious that somehow the cumulus scheme in D02 may interact with WSM3.

Personally I think Thompson scheme performs better than WSM3, especially for high resolution case.
Click to expand...
ok, thank you. I will try to set cu_physics = 2, 0. Thanks a lot.
 
Ming Chen said:
Your namelist.input looks fine to me except that you should turn off cumulus scheme for D02, i.e.,

cu_physics = 2, 0,


WSM 3-class simple ice scheme is well tested and I don't think it could be the reason that caused model crash in your previous run. I am suspicious that somehow the cumulus scheme in D02 may interact with WSM3.

I also want to note that, personally I think Thompson scheme performs better than WSM3, especially for high resolution case.
Click to expand...
Hi，Ming Chen. I have a question. I set cu_physics = 2, 0, But, I checked the RAINC and found the all values of RAINC are zero. How can I check the convective precipitation in my model?
 
RAINC is convective precipitation produced by cumulus scheme. If you set cu_physics = 2, 0, RAINC in D02 should be zero, but it should have values in D01.
 
Ming Chen said:
RAINC is convective precipitation produced by cumulus scheme. If you set cu_physics = 2, 0, RAINC in D02 should be zero, but it should have values in D01.
Click to expand...
Ok, thanks a lot. The D02 is 3km x 3km. So, I close the cumulus option and the model will explicitly calculate the convective precipitationin the D02 in term of the WRF user guide. I think the convective precipitation exist in the results of model output. I don't konw if my understanding is correct?
 
You must log in or register to reply here.
Top