> I would go with storing all your like data in a single table with > observation_date similar to what Bborie has described. > FWIW: The approach that Bborie has below is pretty much what > > ST_Union(rast,'MEAN') does except ST_Union can do it on more than two > rasters at a time.
I'll have to try that out! When I wrote my half-month query, ST_Union wasn't really ready for testing. > > Bborie, > > I am curious if you get the same answers using ST_Union as what you are > doing. Seems like a good test to stress test ST_Union MEAN. > I'll do some testing after next week's upgrade of my production databases to PostGIS trunk. > Is your data freely downloadable from somewhere. I was going to pull some > from weather.gov to experiment with as I suspect moving averages will be a > common use case and I suspect no one has tried raster in combination with > window agg constructions. In theory it should work because all aggregates > are supposed to work in a window agg construct without programming them to > do so specifically. At least as I recall all the sql aggs I've custom built > have worked fine in window constructs. The data used to be freely available through NASA AMES until recently. I plan on making the data available again some time in 2012. If there is anything you'd like though, let me know. -bborie -- Bborie Park Programmer Center for Vectorborne Diseases UC Davis 530-752-8380 [email protected] _______________________________________________ postgis-users mailing list [email protected] http://postgis.refractions.net/mailman/listinfo/postgis-users
