On Thu, 20 Sep 2007 12:47:31 -0400, Frank Warmerdam <[EMAIL PROTECTED]> wrote:
>Neil Best wrote: >> Is it possible to use a PostGIS table as a tileindex? The docs seem to >> indicate so, but I could not find an example of how to connect to >> PostGIS through CONNECTIONTYPE OGR. > >Are you wanting to use postgis as a tileindex for a raster or a vector >layer? I only have experience doing this for raster layers, but it might >also work for vector layers. > Yes, Frank, it is for raster layers. Gregor was quite right to wonder what the advantage of tileindexing vector layers if PostGIS is being used properly. >Generally if you want to use postgis for a tileindex, you need to create >a seperate tileindex layer and then reference that layer by name in the >TILEINDEX keyword of the tiled layer. It worked like a charm! (Except now something is wrong with my legends . . . gremlins!) I never would have thought of referring to a layer from another layer and I don't recall mention of it in my scouring of the docs. Are there other applications/examples of this technique? I have some ideas about how to use this table as a unified raster catalog. As I understand it a TILEINDEX must be in the same projection as the rasters to which it refers. I am imagining that I can still use this table for a heterogeneous collection (e.g. NAIPs are in UTM zones) because I can set up views that provide the tile geometry in the proper SRS while still maintaining a geometry column in our default SRS*. Furthermore this will allow me to define a cross-section of the collection using SQL which will be useful in accessing rasters along the time dimension and any other way based on the path/file or any additional attributes. This table will serve an additional purpose in keeping track of automated, near-real-time data ingests. Frank, maybe your experiences have taken you in similar directions. I presume that you will be in Victoria next week -- maybe I could corner you to discuss this further. Thanks for the tip; this will be very useful. Neil *Which happens to be an Albers Equal Area for the continental US that has no EPSG code. I find this strange because I believe that it comes from the USGS and is widely used, but what do I know? I'll attach the WKT for inspection -- maybe it could be adopted if there was enough interest.
aeaConus.wkt
Description: Binary data
