[CF-metadata] fixed sensors, depth, datum

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 10 Sep 2008 13:18:05 +0100

Dear Dale

> + Create a z dimension with a value of 1
> + The vertical coordinate variable 'z' (i.e. z(z) ) will be
> used to fix the location of the sensor in vertical space
> (i.e. not the distance below the water's surface) referenced
> to a datum.
> + For the datum we use WGS84.

By "WGS84" I presume you mean their geoid (I am not an expert in vertical
datums). If so, as Olivier says, you could request a new standard name of
depth_below_geoid or height_above_geoid. As an alternative to a dimension of
size 1, you can use a scalar coordinate variable (CF 5.7).

> + The depth variable (pressure sensor) gives depth in vertical
> distance below surface (which varies with the tide for fixed
> instruments).

The standard name of "depth", which you have used, means depth below the
(sea) surface, which I think is what you want. Since it is not a coordinate
variable, it doesn't need a positive attribute.

If I understand correctly, the offsets are giving the height_above_geoid of
various tidal datums. I suggest that a more CF-like way to do this would be to
provide these offsets as data variables, each with its own standard name; the
standard names would be e.g. height_above_geoid_of_mean_lower_low_water. You
could request such standard names. If they are all standard tidal datums, that
might be unproblematic! The tidal datum epoch would then naturally be given as
the bounds of the time coordinate variable for these various datums.

> The instrument goes up and down with the tide and the
> sensors are at a fixed distance below the surface. The vertical
> values are all distance below the surface. How do we handle that?
> One solution is to keep the vertical dimension of unity and have a
> vertical coordinate variable referenced to depth below the surface.

Yes, I think that is the right solution.

Cheers

Jonathan
Received on Wed Sep 10 2008 - 06:18:05 BST