Elevator interpolating pressure to pressure levels is not constant

[olafurrognvaldsson]

When testing the latest version of WrfUtils I noticed that the pressure variable is neither constant in time nor space when data is being interpolated to constant pressure levels.

The plot above shows an example of this, top shows a time-series for the geopotential height at 900hPa and the bottom shows a time-series (from the same location) of pressure. The bottom plot should show a flat line (more or less) and the value should be 90000 Pa, and not 900 Pa.

[logi]

The easiest part of this is that the unit for pressure should be hPa and not Pa.

[logi]

@olafurrognvaldsson how does this look for a pressure lower than 900hPa? Could this be the terrain blowing through the pressure level?

[olafurrognvaldsson]

@logi what should I be looking at :)

[logi]

Click the bottom_top value in ncview and it'll show a different vertical level.

Meanwhile I've realised that I was testing this with a tiny file and might not have any visible variability in it so I'm re-running with a 1.4GB africa-50 file.

[logi]

This is what I see at the 500hPa level, so it looks pretty constant except that it's constant at 514.3 - 514.4 hPa instead of 500 hPa.

[olafurrognvaldsson]

Does the geoponential_height have the same shape in this particular point (like it did for my test case)?

[logi]

Geopotential height in 500hPa

Geopotential height in 750hPa. You can see the highest mountains as a few individual pixels (disabling interpolation in image)

Geopotential height in 900hPa. A lot of the continent sticks out

[olafurrognvaldsson]

No I meant how does the time-series for geopotential_height for point (3, 145, 43) look like?

[logi]

Ranges from 5620m to 5645m

[olafurrognvaldsson]

Can you re-run the elevator script and add variables P and PB to the output file and make it available somewhere so that I can open it?

[logi]

sleggjan:~logi/wrfout-africa-pressure.nc4 should be ready in 10 seconds

[olafurrognvaldsson]

Stuck in a skype-meeting. Will take a look in 20-30 min

[logi]

The interpolation of geopotential_height to height above sea-level is very precisely constant and I would expect the pressure variable to be about as constant when interpolated to pressure levels. I'm going to read through some code to see if it makes any sense.

[logi]

OK, I've found a bug. The interpolation constants for vertically staggered or unstaggered variables are created for pressure levels from (P+PB)/100 like for heighht fro m(PH+PHB)/g but the latter is vertically staggered while the former is not, so the interpolation is slightly wrong which would explain the 514hPa issue. The 14hPa is presumably half a sigma level and varies slightly.

[logi]

With that fix (not committed)

[logi]

@olafurrognvaldsson I believe this task is done and have moved it to the Done column. If you try "your thing" again and it looks correct, please close the issue.

[olafurrognvaldsson]

@logi should I update my WrfUtils repo? I.e. has the fix been committed ?

[logi]

Yup. You can also see the commit just above here as

[olafurrognvaldsson]

Looking good, case closed !!