Ensemble statistics (percentiles)

[TarekAMZ]

Hi,
I was wondering if it is possible to calculate the ensemble percentiles across all timesteps, e.g., similar to calculating the ensemble mean or max using ignore_time = True like the example below from the documentation.

data = nc.open_data(file_list)
data.ensemble_max(ignore_time = True)

I appreciate the help.

[robertjwilson]

Thanks for using the package @TarekAMZ.

Right now, the functionality does not exist. I will raise an issue to add this.

Depending on how your data is structured, the following could work

data = nc.open_data(file_list)
data.merge("time")
data.tpercentile(....)

Under the hood, nctoolkit will use CDO's mergetime operator to combine files in this case. If you have duplicates, it will end up duplicating times if they appear across multiple files. I think it should work in most cases. Though at times you can have an ensemble where time is defined in a totally incompatible way so the mergetime operator doesn't work. I'd try it and check for warnings/errors which ought to be self-explanatory.

Alternatively, you could try this:

data = nc.open_data(file_list)
data.cdo_command("cat", ensemble = True)
data.tpercentile(....)

In most cases, I suspect either approach will work. It's just a question of whether the file structure is such that they can be merged.