ClimateGlobalChange/tempestextremes

Spurious pixels in NodeFileCompose's snapshots

Opened this issue · 2 comments

stella-bourdin commented

Hi,

I am using NodeFileCompose to retrieve snapshots of surface wind speed along a Mediterranean cyclone track. However, there are some sort of "spurious pixels" appearing that do not seem to correspond to anything in the original data. The resolution of the original data is ~10km. I tried dx from 0.05° to 0.5°, but you can always see them.

Here is the NodeFileCompose command I used :

~/tempestextremes/bin/NodeFileCompose \
        --in_nodefile medicane.csv \
        --in_connect connect_MED.txt \
        --in_nodefile_type SN --in_fmt "(auto)" \
        --in_data "med_2d_10.nc;med_2d.nc" \
        --out_grid "XY" \
        --out_data "snaps_0p1.nc" \
        --var "sfcwind(10m),_VECMAG(uas(10m),vas(10m)),psl,hfls,hfss,pr,prw,rsut" \
        --varout "sfcwind,ws,psl,hfls,hfss,pr,prw,rsut" --snapshots \
        --dx 0.1 --resx 100

Here is the field on the original grid:
image

And here are the corresponding snapshots at different resolutions:
image
image
image
image

A similar thing happen when I use a radial grid instead.
Do you understand what is happening? Do you have an idea how to fix it?

Stella

paullric commented

That is very strange... Is there any way you could provide me with example data that reproduces this result?

stella-bourdin commented

Sure! Sorry I definitely had the intention to and forget to do so (and also had the wrong settings in GitHub so I did not see the notifications, now I should be more reactive!)
Here is the csv nodefile: https://sharebox.lsce.ipsl.fr/index.php/s/D524EffEPRbOr4j
And here is a sample of the two data files:
https://sharebox.lsce.ipsl.fr/index.php/s/xyNZVsIncROr382
https://sharebox.lsce.ipsl.fr/index.php/s/dtUP9cBbZdIw4bV