SWE L3 metadata updates

[jtniehof]

This is a list of ISTP metadata that needs to be changed for SWE.

Bold items need SWE team input; others are informational or Menlo can handle directly.

• There are several out of range values for the day I checked (imap_swe_l3_sci_20260428_v003.cdf):
  • PSD has a VALIDMAX of 10^-20. There are some values around 10^-19 which are maybe physical and we should change VALIDMAX? Then there are some around 10^-10, 10^15, and inf. Obviously we should replace inf with fill, but should we do that for anything over VALIDMAX, or just leave it as is? Most programs will not plot those invalid values. This affects all the PSD variables to some degree
    • Answer: replace inf with fill
    • Answer: update VALIDMAX on all PSD variables to 1e-15, do not replace large values with fill
    • Answer: add a flag UNPHYSICAL_PSD, 2 ** 11, and set for PSD greater than 1e-19
  • halo_density_integrated is negative in one point; core_temperature_phi_rtn_integrated is negative in a lot. If an integration throws a negative value, should we replace it with fill and also any related variables?
    • Answer: replace negative values with fill and make sure the NEGATIVE_MOMENT flag is set
• I'm checking on the possibility of using time_series for vector plots in CAVA; we may need to change this
  • Answer: if we leave this as time_series it should be good, no change needed.
• core_heat_flux_magnitude_integrated, halo_heat_flux_magnitude_integrated, total_heat_flux_magnitude_integrated are using Unicode mu. ASCII is preferred; IMAP convention is to use "u" for micro, so just switch the mu to u.
• raw_psd_by_theta_rebinned and raw_psd_by_phi_rebinned are missing LABL_PTR. LABL_PTR_1 should be energy_label for both; then we need to create inst_az_label and inst_el_label for the appropriate LABL_PTR_2.
• inst_el FORMAT should be E9.4 to correspond with inst_az (or at least it needs to be float, not integer).
• UNITS updates relating to BUG - CoDICE Lo L3a ISTP validation issues - Card 3160 #87 will be picked up with next deployment, as will empty VARIABLE_PURPOSE

[pleasant-menlo]

Hey @jtniehof,

We are working on updating setting the NEGATIVE_MOMENT flag when the result of an integration is negative. We had a clarifying question about where we want to do this fill value masking.

A couple of questions:

• When the result of one dimension of tensor integration result is negative, should all the dimensions of that tensor be FILLVAL?
• Should we be masking the negatives with FILLVAL before or after the integrations results go through scaling?
• If a negative integration result is masked with FILLVAL, is it okay that these are propagated as NaNs to downstream processing (e.g. if the temperature tensor integration result is negative, there would be implications on the mag, theta, phi quantities)

~ Logan and Matt

[jtniehof]

Yes, if one dim is negative, let's make all tensor elements fill.

Negative or not shouldn't be affected by scaling. Of course, what matters is that the output is fill, so don't do -1e31 and then scale that. Practically it probably makes sense to do nan and replace with fill at CDF write time.

And yes, okay to propagate NaNs, and replace with fill at CDF write time.