[postgis-users] Indices for ST_distance_sphere

[Paul Ramsey]

What about 3112 (# GDA94 / Geoscience Australia Lambert) or 3577
(Australian Albers), those should cover things fine, no?

P.

On Tue, Oct 14, 2008 at 3:46 PM, Stephen Baillie
<steve at alliancesoftware.com.au> wrote:
> G'day Regina,
>
>> It really depends your area of interest, how far out you span out when
>> looking for objects and how accurate you want your distance search to be
>> what the best solution is.
>>
>> If your area of interest is small like a state or a UTM zone and even US,
>> its best to create a functional index  on ST_Transform(the_geom,
>> someplanarsrid) or to just keep your data in that spatial ref and used the
>> transformed coordinates for doing distance searches.  If your area is
>> wider
>> and your buffer area is also wide, that gets a bit more complicated.
>>
>> Any rate if you tell us the area you are dealing with - is it all of the
>> world, US, Europe etc. and how far you look e.g. do you need to look out
>> 10
>> meters from a point, 10 miles or 100 miles or more, then I'm sure we'd be
>> more able to give you some direction.
>>
>
> My area of interest is Australia (-10 to -50S, 110 to 160E), so I've been
> told that the only spatial ref that it makes sense to use is lat/long.  A
> reasonable ceiling on search ranges is 400km.  Since I'm already storing
> location information in lat/long, I'm not going through ST_Transform anyway.
>
> Does that shed more light on my problem?
>
> Thanks,
>
> Steve.
>
> --
> Stephen Baillie
> Developer
> Alliance Software
>
> 1/234 Whitehorse Road
> Nunawading, VIC 3131
> Australia
>
> Ph:  03 9877 9921
> Fax: 03 9894 2106
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users
>