The CCTM run successfully, but NaN or Infinity detected on processor 000 in DIFF, ADV, COUPLE in log

Hello everyone, I’m running the CCTM with gridded emissions in CMAQ 5.3.1, WRF3.9. The d01 region of my simulation covers most parts of China. The log finally shows that the run was successful, but it also shows that NaN or Infinity was detected on processor 000 in DIFF, ADV, and COUPLE. I’m concerned if this will affect my simulation results. Additionally, my emission has only one layer. I wonder if there’s something wrong with my emission. When I only add the anthropogenic emission, it looks good as shown in pic1. However, when I add the bio-emission, this problem occurs. My bio-emission is made from Megan 3.2, pic2 shows the ISOP value. The pic3 shows the simulated ozone concentration variation. However, it fails to capture the diurnal differences and the peaks and valleys accurately. I would greatly appreciate any advice or suggestions. Thank you. @ hogrefe.christian@ cgnolte @lizadams
CTM_LOG_029.v531.txt (281.9 KB)

Yes, this line

After VDIFF : Gas NaN | Aer 1.634E+09 | Non 1.137E-04

after the very first call to VDIFF where emissions are processed points to some problem with emissions. And since you already determined that this doesn’t occur if you don’t include the biogenic emissions as gridded file #6, you would want to investigate investigate all variables and time steps in that file to see which one seems off. One way to do that would be to run m3stat on that file. You could also compile CCTM in debug mode, that should then crash the model when NaNs are encountered and the crash message might provide more clues on where the problem originates.

I have the same problem. Could you please tell me how did you fix that?

My biogenic emissions made the fault. There exists some NaNs in my CT3.csv.

Hi Axueaha:
Thank you very much for your reply, I just checked mine and found NaNs in my CT3.csv too. So did you change these NaNs to 0 or do you have another solution?

I tried this way, but it didn’t work as expected. When I replaced the NaNs with 0, the megan stopped working entirely. However, when I used Megan 3.1 to generate the CT3.csv file, it worked successfully. This led me to speculate that the issue might be related to the source code of the megan3.2. But I’m not sure. You can try it.

Hi Axueaha
Thank you very much I will try Megan 3.1 to generate CT3.csv. Thanks.

Hi Axueaha
May I ask you where did you download MEGAN3.1 please. I download MEGAN3.1 from UCI BAI - MEGAN3_1 and there isn’t prepmegan4cmaq_cantype.x in the code. Thank you very much.

Hi Starry,
I forgot where I got the prepmegan4cmaq_cantype.f90. I upload the f90 file. Hopefully the cantype.x can be compiled.
prepmegan4cmaq_cantype.f90.txt (37.7 KB)

Hi Axueaha
Thank you very much.

Hi Axueaha
Do you have ‘’, ‘’, ‘’, ‘’, ‘’, ‘’ for MEGAN3.1 prepmegan4cmaq_cantype.x? I don’t have access to download these input files. I changed scripts to use MEGAN3.2 input files but failed. Thank you.

The file is huge, or you leave your mailbox.

Hi Axueaha is my email address. And thank you very very much.

Hi Axueaha
I don’t know if I left a bad email address so I didn’t receive your email. Here is my another email address hope this one can make it right.
Wish you all the best.

Sorry, I’ve been a little busy recently. The email has been sent.

Hi Axueaha
Thank you very much. You helps me a lot and I am very grateful. I still found NaNs
in CT3.csv outputing by MEGAN3.1. So maybe there is something wrong with something I don’t know. I will try to replace NaNs with 0 in the result.
Wish you all the best!

I remember something. There are some NANS always and that’s ok. The problem is the value 100 in CT3.csv. So maybe you can check your csv file again.

Hi Axueaha
You are right. There are lots of ‘100.0000’ in CT3.csv from MEGAN3.2 but not MEGAN3.1. However I replaced CT3.csv and rerun MEGAN3.2 the result still couldn’t work successfully with CMAQv5.4. So I replace NaNs in MEGAN3.2 result with 0 and finally it worked.
Cheers! Thank you for your kindly help.