Struggling to get output frequency for hourly data

[HazelMooney]

What version of the code are you using?
CLM4.5, CESM2.2

Describe your problem or question:
I am trying to generate hourly data for BVOC emissions from the CLM. I had my settings in user_nl_clm as: hist_mfilt = 720 hist_nhtfrq = -1
I did this following several examples in CESM documentation using -1 for hourly data. However, this has not generated a data point per hour, i still only have 1 data point per month (720 hours).
I am seeing mixed messages online about whether using nhfrq <0 causes averaging or not.
The only examples i can find for hourly output use -1, but i do not want averaging, i want a data point for every hour.
I can see that the model timestep is 1800 seconds = 30 minutes.
Please can you advise?

Do i need to instead use hist_nhtfrq =2 for output per 2 timesteps ?

 

[HazelMooney]

I would like to add that for previous runs i did the following
for daily data i used hist_mfilt = 1 hist_nhtfrq = -24 for 1 file per day
for monthly data i used hist_mfilt = 1 hist_nhtfrq = 0
STOP_OPTION has been set to nyears

 

[slevis]

@HazelMooney
Nothing stands out as wrong to me in what you've done, so I would try a few things until I got something that worked if I were you. For example...

Since you wrote that
hist_mfilt = 1 hist_nhtfrq = -24 for 1 file per day worked, then I would try this next:
hist_mfilt = 1 hist_nhtfrq = -1 for 1 file per hour
If this doesn't work, then I'm at a loss. Whereas if this works, then you can attempt other mfilt values until you get one that you like.

 

[slevis]

Oh, but your other idea should also work
hist_mfilt = 1 hist_nhtfrq = 2 for 1 file per hour
and again, assuming that this works, then change mfilt to something else that works and that you prefer.

 

[HazelMooney]

Thank you for your reply @slevis
I will give that a try.

Another example that might help, i used hist_mfilt = 24 hist_nhtfrq = -1 i got 1 file per day. For 1 year this is equivalent to an array of 744 (time) by 384 and 576 coords. With chunk size showing as 24. Giving me 1 value every hour.

However for the output using hist_mfilt = 720 hist_nhtfrq = -1 i got 1 file per month as expected. But the array is showing as 2 by 384 by 576 for data variables. Chunk size showing at 2. I’m not sure if this is helpful in shedding any light?

 

[HazelMooney]

Oh, but your other idea should also work
hist_mfilt = 1 hist_nhtfrq = 2 for 1 file per hour
and again, assuming that this works, then change mfilt to something else that works and that you prefer.

Hi,

I have not been able to fix this.
I used hist_mfilt =720 and hist-nhtfrq =2 and still got only 1 data point per month
i tried going back to something that had worked, hist_mfilt = 24 hist_nhtfrq = -1 where i got 1 file per day, but this time it also only gave me 1 data point per month and generated monthly files.

I am really puzzled why my settings are being over-ridden in some way.

I have several experiments which have previously used a 30yr run of monthly data. I simply want to restart these and get 1 year of hourly output. I have now tried 5 different configurations and am simply not getting hourly data output.

Please can you advise anything else i can try?

Is there soomething i need to change regarding a flag for averaging vs instantaneous, as well as requesting this output?

My successful 30 year runs used monthly averaging, making 1 file per month.

I welcome your support.

 

[oleson]

I'm not sure what's going on. I tried a global simulation with release-cesm2.2.2 with this in the user_nl_clm:

hist_mfilt = 720
hist_nhtfrq = -1

and I got a file:

cesm222_I2000Clm50BgcCrop.clm2.h0.2000-01-01-00000.nc

that has 720 time samples.