Maarten,
I believe that what your colleague should do is add a bounds variable
for the pressure and reduce the number of elements in the hPa coordinate
variable by one. The bounds variable provides a lower and upper bound
for each layer, so it captures the value currently being stored in the
extra element of the hPa coordinate variable. The values stored in the
hPa coordinate variable can be the lower bound pressures, the upper
bound pressures, or any value in between the two (layer center
pressures, for example).
Grace and peace,
Jim
On 2/19/15 10:56 AM, Maarten Sneep wrote:
Hi,
A colleague has a question that left me without an answer. He writes:
I have a 3D ozone field with n_lon elements in longitude direction,
n_lat elements in latitude direction and n_layer elements in the
vertical direction. The corresponding pressure grid has the same number
of elements in longitude and latitude directions, but since the pressure
is given on the layer interfaces, the number of elements in the vertical
direction is one more than for my ozone field. The dimensions of both
fields are therefore:
O3 = [ n_lon, n_lat, n_layer ]
hPa = [ n_lon, n_lat, n_layer+1 ]
I have looked at the CF-conventions 1.6 document, but
* it's not clear to me if I could use a "boundary variable" (section 7).
* the pressure field is irregular, so a hybrid solution
(P = hyb_a + hyb_b*P_surf) is not an option (appendix D).
The question is how to write these fields as two datasets to a netCDF
file in a CF-compliant way, so that it is clear that the pressure field
gives the vertical extent of the layers of the ozone field. I have
thought of the following options:
* use 4 dimensions (lon, lat, layers, layers+1), but in that case, the
ozone and pressure datasets have a different third coordinate, while
they are obviously linked.
* separate the pressure field into a surface pressure field and all
higher levels, but then you have two dataset for the air pressure.
What is the preferred way of storing data with a vertical coordinate
that is specified on the interfaces rather than on the layers
themselves, ans where a coordinate variable doesn't provide the detail
we need?
Kind regards,
Maarten Sneep
--
CICS-NC <http://www.cicsnc.org/> Visit us on
Facebook <http://www.facebook.com/cicsnc> *Jim Biard*
*Research Scholar*
Cooperative Institute for Climate and Satellites NC <http://cicsnc.org/>
North Carolina State University <http://ncsu.edu/>
NOAA's National Climatic Data Center <http://ncdc.noaa.gov/>
151 Patton Ave, Asheville, NC 28801
e: [email protected]
o: +1 828 271 4900
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
