[ncl-talk] Probability density function issue with missing values

[Giorgio Graffino]

Hello NCL experts,
 
I'm using the NCL function pdfx 
(https://www.ncl.ucar.edu/Document/Functions/Contributed/pdfx.shtml) to 
compute probability density functions for a number of datasets. I think 
there is a problem with how the function deals with missing values.
 
I'm attaching two plots: Volcano_tempmovingtrend_timeseries_MMMpdf.png 
is obtained from ensemble-mean datasets (no missing values), while 
Volcano_tempmovingtrend_timeseries_MEMpdf.png is obtained from 
individual ensemble members of the same datasets (loads of missing 
values). The red lines in Volcano_tempmovingtrend_timeseries_MEMpdf.png 
are really messed up, probably because of the great amount of missing 
values in that dataset.
 
Is pdfx skipping missing values in the computation? What workaround can 
I implement to avoid the issue?
 
Many thanks,
Giorgio
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20240715/f66e7efc/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Volcano_tempmovingtrend_timeseries_MEMpdf.png
Type: image/png
Size: 106546 bytes
Desc: not available
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20240715/f66e7efc/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Volcano_tempmovingtrend_timeseries_MMMpdf.png
Type: image/png
Size: 111530 bytes
Desc: not available
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20240715/f66e7efc/attachment-0003.png>