Bugs in Hm0 calculation for spectrum dfsu in some cases

[jsmariegaard]

Two problems have been identified for spectral dfsu files in the Hm0 calculation:

• deletevalues in spectrum will give nan Hm0 - it should be safe to treat them as 0
• if the file uses radians instead of degrees (unit: meter pow 2 sec per rad), the directional integration is wrong and Hm0 results will be sqrt(180/pi) too large.

[ecomodeller]

No action have been taken on this issue, is this a hint that

mikeio/mikeio/dfsu/_spectral.py

Line 324 in 4929b6f

def calc_Hm0_from_spectrum(

and related functionality is actually out of scope for MIKE IO?

[ecomodeller]

Stale

[jsmariegaard]

I think it is good to clean up - but I don't know if is a good idea to close issues because they are old and does not get attention. If they have not been fixed, it is still valuable information as least as a "known issue". Then for this issue it is still a question whether this is out of scope, but like we have other simple statistics in both generic and DataArray/Dataset I would say it's in-scope.

[ecomodeller]

If you provide a minimal reproducible example that can illustrate the two bugs, I'll fix it.

[jsmariegaard]

@stkistner or @daniel-caichac-DHI - could you please help me here?

[daniel-caichac-DHI]

What we did in wax was

• In spectra replace NaN with 0s before Moment calculations (its safe since it should not exist negative energy or energy density anyhow)
• Wasnt the radians issue fixed already in ... Faulty conversion of direction on read() in _spectral.py #724 ?

[jsmariegaard]

Sounds good but can you help me with 1-2 tiny files that illustrates the problem and could be used for tests?

[daniel-caichac-DHI]

Energy_Action.zip

I have this, it is a model output so it starts from nothing, with mike viewer you can see 0's but you can probably just delete the data and save and see what happens.
I mean, maybe mike viewer does not show the NaN's but at least this is a direct model output.
It is wave action so you might encounter Another bug.

As for radians, I know that Bjarne has data in such format