[Freegis-list] Free GIS for realtime data and 3D visualization?

[Brent Wood]

--- Peter Fischer <peter_7003 at yahoo.com> wrote:

> Thank you all for your answers, Brent, Markus, Daniel and Paul,
> 
> 
>   It looks like GRASS GIS (because of its 3D capabilities) and/or UMN
> MapServer
> are good choices for me. Some details (especially concerning WMS/WFS/WCS)
> are still not that clear to me. At best the following could be realized by a
> GRASS
> GIS configuration which conforms to a WMS/WFS/WCS architecture:
> 
> 
>   Some questions do I have to your answers:
> 
> 
> 1. I thought the WMS server puts together the data it gets from the WFS 
>     (vector/feature data) and from the WCS (raster data) into a single
>     raster image?

Sort of, yes, that is a typical basic WMS operation. However some more
intelligent WMS client applications (eg, OpenLayers) enhance this, by doing
things like downloading layers as separate transparent PNG's & overlaying them
locally in the client. Thus a "redraw" with layers turned off is pretty much
instantaneous, as the client turns off that PNG. It also downloads the images
in tiles covering an area greater than the the actual region shown, & caches
them locally, so any panning/scrolling is often managed with no WMS calls, just
local images. This does create initial overhead, as each layer/tile is a
separate WMS transaction, but provides a much better viewing experience IMHO
:-) 

Note, however, that a WMS does not generally get it's data from a WFS, but from
the same underlying vector data repositories that a WFS uses. They usually
independently access the same data. As I understand it, the overlay of WFS
vector features om WMS maps is done at the client end, to allow client
interaction with the features.


> 2. Daniel, as far as I have figured out, WFS basic is without write
> access, and WFS-T (transactional) is with write access. Do UMN MapServer or
> GRASS GIS (Markus) already support this? It would be important to be able
> not only to change styles (like with the SLDs) but to do annotations (I tried
> FIST on their site, its nice, but annotations failed when I tried – the map 
> wasn’t displayed any more) or even change the vector data already stored in 
> the vector layer database (WFS for example).

I haven't used FIST for quite a while. Looked at that, MapLab & Ka-map, but
have pretty much found OpenLayers suits our needs at the client end. (these are
all different clients/client building apps that do quite different things).
Mapserver is not a WFS server. I understand the combination of Geoserver as a
WFS-T engine and mapserver as a WMS engine is popular. 

See http://docs.codehaus.org/display/GEOSDOC/WFS

geoserver also supports WMS, so you do not actually require mapserver, but
there seems to be an overall feeling in the user community that mapserver
overall is a better WMS provider. But I guess opinions will vary :-)

<snip> (I'll leave the GRASS questions to those who know it better :)

> 5. In the same constellation (GRASS GIS with WMS/WFS/WCS), is display of
>     realtime vector data coming in from sensors like GPS signals
>     overlayed over the map possible? –  The architecture would be an 
>     additional data source (to WFS and WCS) with the GPS signals and the WMS 
>     puts them together to one raster image and sends them to the multiple  
>     views? – By the way, the views are different, not identical.

Yes. via several means. We use GPS data fed straight into a PostGIS database
which is re-rendered via a WMS server every few seconds. Not elegant but goos
enough for our at-sea needs. Things like Geo-RSS feeds of the data may be
possible. A mapserver mapscript application could certainly access an NMEA GPS
stream on a serial or UDP port & enable it as a layer, or via a mapserver WMS
layer with an SQL selecting just the last n values in a Postgis table where the
data is automagically logged to. OpenLayers could be tweaked to refresh just
this layer automagically.

Ther are many approaches by which this can be (& has been) done. Fleet/vehicle
tracking apps for exanple.



Cheers,

  Brent