Convert more waves data to CF and harvest in portal?

[emontgomery-usgs]

@rsignell-usgs , @kwilcox
We intentionally exclude some files that include wave height and period because other variables in the files are dimensioned oddly (by time and frequency). Is it time to revisit this?

In this example https://stellwagen.er.usgs.gov/thredds/dodsC/TSdata/MVCO_15/10571whVp-cal.nc.html, variables named *spec have dimensions that don't fit any CF models (as far as I know), but wh_4061 and wp_4060 are just dimensioned by (time, lat and lon), so should go to CF fine. Can the other variables be passed as-is to a CF version of the file?

If we want to try this, adding "p-cal", and "Wvs-cal" to the already complicated regex should pull in more wave data.

[emontgomery-usgs]

@kwilcox , any chance of implementing this soonish?
@rsignell-usgs, please check to see if you agree with the mappings I've picked for our variable names found in waves files.

our name CF name
wh_4061 sea_surface_wave_significant_height
wp_4060 sea_surface_wave_mean_period
mwh_4064 sea_surface_wave_maximum_height
wp_peak sea_surface_wave_significant_period
wvdir sea_surface_wave_from_direction

New filenames that would need to be included in the regex:
wvs_ (NOT wvsb-cal), p-cal

Thanks!

[rsignell-usgs]

@dnowacki-usgs can you review this please?

[dnowacki-usgs]

I suggest the following changes:

wp_4060: At least in EPR and GrandBay, we have been computing wp_4060 as sqrt(m0/m2) which would correspond to sea_surface_wave_mean_period_from_variance_spectral_density_second_frequency_moment and not sea_surface_wave_mean_period.

wp_peak: Peak period as we define it is the period associated with the greatest energy in the spectra. So it should be sea_surface_wave_period_at_variance_spectral_density_maximum and not sea_surface_wave_significant_period.

In reading the CF standard names I'm honestly not sure which of sea_surface_wave_from_direction_at_variance_spectral_density_maximum and sea_surface_wave_from_direction should correspond to wvdir and dwvdir.

[rsignell-usgs]

For wvdir and dwvdir you not sure of what we have or not sure of what CF specifies?

[dnowacki-usgs]

I find the CF standard names somewhat imprecise.

First, sea_surface_wave_from_direction doesn't specify how the wave direction is derived; the definition text deals only with semantics of how direction is defined (from, clockwise from due north).

The definition sea_surface_wave_from_direction_at_variance_spectral_density_maximum is:

The quantity with standard name sea_surface_wave_from_direction_at_variance_spectral_density_maximum is the direction from which the most energetic waves are coming. The spectral peak is the most energetic wave in the total wave spectrum.

We define dwvdir as "the compass direction from which waves are propagating as defined by the direction band with greatest total energy summed over all frequencies"

We define wvdir as "the compass direction from which waves are propagating as defined by the direction with the greatest energy at the peak period"

I guess sea_surface_wave_from_direction_at_variance_spectral_density_maximum is closer to dwvdir than wvdir but I'm not 100%.

[emontgomery-usgs]

@rsignell-usgs - can we go with Dan's suggestions and move forward with this?

[emontgomery-usgs]

@kwilcox , @rsignell-usgs
please:

1. modify the file regex to allow files with wvs_ and p-cal
2. modify the following mappings in https://github.com/axiom-data-science/epic2cf/blob/master/epic2cf/data.py#L747-L750. THese EPIC codes should be
   4060 -> sea_surface_wave_mean_period_from_variance_spectral_density_second_frequency_moment
   4061 -> sea_surface_wave_significant_height
   4064 -> sea_surface_wave_maximum_height
3. find a way to map these with no EPIC code
   wp_peak -> sea_surface_wave_period_at_variance_spectral_density_maximum
   wvdir -> sea_surface_wave_from_direction
   dwvdir -> sea_surface_wave_from_direction_at_variance_spectral_density_maximum

Thanks!

[emontgomery-usgs]

@rsignell-usgs, why did you unassign kyle on this ticket? I think we are ready to go on this.

[rsignell-usgs]

Looks like @kwilcox unassigned himself and assigned me.

[kwilcox]

Just want to confirm that all existing wave data in the portal should be transitioned from sea_surface_wave_mean_period to sea_surface_wave_mean_period_from_variance_spectral_density_second_frequency_moment

[emontgomery-usgs]

I think that would be OK. Perhaps more specific than is strictly valid for some. I don't foresee a huge adverse impact. I think you won't find many wp_4060 now, because we've excluded most wave files by the regex.

…

On Mon, Jul 30, 2018 at 2:00 PM, Kyle Wilcox ***@***.***> wrote: Just want to confirm that all existing wave data in the portal should be transitioned from sea_surface_wave_mean_period to sea_surface_wave_mean_period_from_variance_spectral_ density_second_frequency_moment — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#279 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABsHnpwAxRTBAysH7G0ibaS8Kb2BH4uXks5uL0m5gaJpZM4S7vWX> .

-- Ellyn Montgomery, Oceanographer and Data Manager U.S. Geological Survey Woods Hole Coastal and Marine Science Center 384 Woods Hole Rd., Woods Hole, MA, 02543-1598 (508) 548-8700 x2356 https://stellwagen.er.usgs.gov; https://woodshole.er.usgs.gov/operations/stg/

[emontgomery-usgs]

@kwilcox, @rsignell-usgs
FYI, some waves data was converted to CF- the _z1 has the displayable content (wave height, period)-see
http://geoport.whoi.edu/thredds/dodsC/silt/usgs/Projects/stellwagen/CF-1.6/CHINCOTEAGUE/10192Cdws-a.nc.html
http://geoport.whoi.edu/thredds/dodsC/silt/usgs/Projects/stellwagen/CF-1.6/CHINCOTEAGUE/10192Adws-a_z1.nc.html

From the same experiment, this style waves data did not get converted to CF:
http://geoport.whoi.edu/thredds/dodsC/silt/usgs/Projects/stellwagen/Data/CHINCOTEAGUE/10191Aaqdwvs_diwasp-cal.nc.html

[emontgomery-usgs]

Here's another new standard_name we should probably assign: sea_surface_wave_variance_spectral_density.

Should it be applied to pspec, vspec, sspec or all?

[dnowacki-usgs]

I think it could be applied to all, as they are just different ways at getting at the same quantity, via pressure fluctuations, surface detection fluctuations, and velocity fluctuations. Note that the units of pspec, vspec, sspec from an RDI are the square-root of the units specified by the standard name table. The RDI values on stellwagen are not really a variance but more a standard deviation, with units m/sqrt(Hz) = m sqrt(s). This might be allowable, based on my reading of the FAQ, as they are convertible (by squaring) to the canonical units, but we might want to present them in the form specified by the table, i.e. as m2 s.

[kwilcox]

@emontgomery-usgs I've got everything hooked up to process waves files, just tested with http://geoport.whoi.edu/thredds/dodsC/silt/usgs/Projects/stellwagen/Data/CHINCOTEAGUE/10191Aaqdwvs_diwasp-cal.nc.html and it was converted now. Is there anything you want to do with the variables based on frequency before I push these changes?