Instantaneous vs. Averaged Output

[scapps@uci_edu]

I wanted to make sure I understand the difference between instantaneous and averaged output. Basically, instantaneous is not averaged over the timestep and averaged is...correct? So, instantaneous is the value at that particular moment in time.

I noticed that there are two ways to output an instantaneous value:

Using an "I" in the addfld call:
history.F90..call addfld ('WINDSPD ','m/s ',1,'I','Grid Cell Wind Speed ',phys_decomp)

And, using an "I" in the fincl line within the Namelist file:
fincl4 = 'WINDSPD:I'
...and:
nhtfrq(4) = 1 #Ouput at each timestep

Does the addfld specification override whatever is placed in the namelist file?

Thank you.

 

[bundy]

I'm pretty sure the namelist setting overrides the addfld setting.

You should be able to check in the resulting history file; there is an attribute called cell_method that shows what time function occured.
CLDLOW:cell_method = "time: mean" ;

Another way to test is to use one of the radiation fields like FSNT; since these only are calculated for daylight regions, the instantaneous field only has values over about half of the globe. A daily average, on the other hand, has values all over the globe. I only suggest this if you want to run a test to make sure it is behaving as you expect.

Finally, I want to check what you meant by
"instantaneous is not averaged over the timestep and averaged is".

I would say instead that instantaneous output represents the instantaneous value at the end of the timestep. Averaged output is the average of the instantaneous values at the end of each timestep since the last history file write.

For instance:

- If you write out every timestep, the instantaneous output is the same as the averaged output.

- If you write a dataset every day at midnight: the instantaneous output is the model state at midnight; the average output is the average of all the timesteps that were computed that day.

- If you write a monthly average file, the average is over all the timesteps that occured in the month.

By the way, if you are thinking of the multiple time-levels in the eularian model, I'm not considering them because, as far as I know, the output does not use information from the intermediate time-levels. If this is what you were thinking of as 'averaging over timesteps', I believe that is not correct; if you weren't thinking about it then ignore this comment- it really isn't important to this topic.

 

[scapps@uci_edu]

Thank you for your response.

Scott