Jon,
You're welcome 
. I'm CCing geoserver-devel so that they can correct my
rampant ignorance.
We are currently serving 4D gridded data from the Weather Research and
Forecasting + Chemistry (WRF-Chem) model. Geoserver currently reads NetCDF
files directly and can index vertical and temporal dimensions. Selection
is made with the TIME= and ELEVATION= keywords in the URL request. My
understanding is that they adhered as closely as possible to the WMS 1.3.0
/ISO 19128 spec for these two keywords. THis is available in the Geoserver
1.5 release and I think cedric is working on enhancements right now (e.g.
using actual times instead of indexes.)
I did not look at how this was accomplished, but I suspect that the
non-horizontal dimensions are just used to slice a particular horizontal 2D
image out of the 4D grid.
The main functional shortcoming on the backend, I think, is the scalar
(rank 0) nature of all GIS data. In order to do things like plot wind
vectors, Geoserver will need some concept of vector (rank 1) data. The
rank 1 data type (as well as its bidirectional translation to and from
various data stores) may be the meat of your contribution. As I recall,
your experimental system had some means of assembling vector components
into a vector prior to rendering. This is going to be needed because I
don't believe that there is a generic persistence mechanism for rank 1
data. Each coverage is going to need to be individually configured so that
Geoserver knows where the vector components are coming from.
The gridded data model is, if I recall correctly, a modified version of the
OGC GridCoverage. There is a flag ("geophysical") to indicate whether the
grid represents a rendered image or a grid of "measured values" (i.e.
temperatures, radiance, u/v/w velocity, etc.). At times in the past I have
started an implementation of an ISO 19123 coverage type. I keep running
out of time. My problem is that I have to wrap my mind around the whole
problem even though I'm only writing code for a very special case. Ergo I
feel I must solve the relationship of coverage to feature, write a 70+ page
expose on the coverage concept, etc. However, even leaving these problems
aside, I would not suggest monkeying with the coverage data model unless
you're in for the long haul. I suspect that this would definitely drag
your team into the rendering pipeline.
Increasing the efficiency of extraction is also a worthy pursuit!
Bryce
PS: It's been a couple of years since I was active on geotools, so my
impressions may be a little dated.
PPS: When you say "making sure that the data model supports 4d data
properly," do you have a particular target/reference in mind? ISO 19111
(nudge)?
"Jon Blower"
<jdb@anonymised.com
sc.ac.uk> To
Sent by: "Bryce L Nordgren"
jon.blower@anonymised.com <bnordgren@anonymised.com>
com cc
Subject
12/18/2007 04:18 Re: NERC knowledge transfer bid -
AM success!
Hi Bryce,
Thanks for your mail and thanks once again for your support! Our main
work with Geoserver will be more at the "back-end", i.e. making sure
that the data model supports 4-D data properly and that the extraction
of data for the generation of maps is efficient. Most of this work
will be done in our group, but TOPP have kindly promised 100
person-hours to support us and provide advice.
I think it's unlikely that we would be able to do much work in the
image-rendering pipeline and so I had no plans to implement the
rendering of wind barbs or contours within this project. However,
this would clearly be extremely useful to a wide community (to us as
well!) so if it turns out that you are able to contribute to the
project to ensure that this happens then this would be very welcome
indeed.
It's interesting that you have served as a Geoserver test site. If
you have any 4-D data (particularly gridded data) would you be
interested in being a test site in this project? I know that we're
not providing you with any resources so if you don't have time, that's
fine.
Best wishes,
Jon
--
--------------------------------------------------------------
Dr Jon Blower Tel: +44 118 378 5213 (direct line)
Technical Director Tel: +44 118 378 8741 (ESSC)
Reading e-Science Centre Fax: +44 118 378 6413
ESSC Email: jdb@anonymised.com
University of Reading
3 Earley Gate
Reading RG6 6AL, UK
--------------------------------------------------------------