John Caron wrote:
On 6/14/2010 1:52 PM, Kevin L. Manross wrote:
In order to get a time accumulation of my grids (to create an areal
"swath" like "Storm Total Precip", but with other fields) I see where
I should use the aggregation "along" the time dimension for my
grids. However, if I want to take the maximum value of a grid point
for all the times available, how would I go about doing this via
THREDDS (or can I)?
Im not sure if this is the same as "are there simple algorithm-like
functions in NcML" question, but I think the answer is that TDS doesnt
do derived quantities. You might look at the Live Access Server (LAS,
which uses TDS underneath) or the Grads Data Server (GDS, which doesnt).
Thanks for the info John.
I will take a closer look at LAS and GDS, but before I dive into that
endeavor, I want to make sure I understand the limitations of
THREDDS/TDS as it pertains to what we're trying to accomplish.
1) TDS cannot perform any type of algorithmic functions. Is there any
capability built into TDS that can call algorithms that are packaged in
a separate Java jar package on the same server running TDS (i.e. from
within the tomcat server)?
2) TDS cannot do on-the-fly aggregations. I have to pre-specify a time
aggregation in a catalog, for example. I couldn't have a user come in
through a web browser and specify "Give me a netCDF subset for bounds
L,M,N,O and for all data available from 0017-0934 UTC" and then have TDS
compile those data on demand, correct?
I have gotten very excited about using TDS to serve our data (especially
since TDS is quite simple to set up and configure), but I want to make
sure I'm not expecting it to do more than what it is meant to be/do.