Hi Tim,
A new week and the rain has started again... time for :
Some paragraphs on itinerary modelling
With itineraries, we want to retrace the actual expedition route as accurately as possible. For this, we want to make use of all possible data sources, the first of which are of course the georeferenced specimen records (such as come from GBIF et al.).
Besides that, we are also looking for other information sources, for example field notes, old maps and the like. When properly encoded in some system of "boundaries", these extra information sources create a set of restrictions that can be used for quality control on the data. For example : if a point by its lat/lon values looks like an outlier of the general set, we could find clues in the other information sources about what is going on. Either some indication that, yes, the point is correct (e.g. its field number fits perfectly in the field number series) or some proof that, indeed, the point is incorrect (e.g. some notes about the camping sites, proving that the expedition was not near that place at the time).
Both the data set (points) and the "info set" are fed into the modeller tool, and two outputs are generated : 1/ a judgement about the points in the data set (+ visualisation, indication of faulty points) and 2/ a new "info set", if the points yielded some new information to be used in the future.
So, it would be a system that grows in knowledge as more data sets are processed, and more extra information is encoded.
In a first stage though, we'll be mostly looking at the internal coherence in the data set points - and that would make the process more "mainstream" and "OM-ready". Later on, the process will add more "semantics" to the geospatial aspect, and I don't see very clear yet how that would fit into OM.
The temporal aspect will, however, play a very important part from the beginning - an it's great news that OM will incorporate this. Have you seen the Mapbuilder timeseries demo :
http://geoservices.cgdi.ca/mapbuilder/demo/timeSeries/index.html
Basically, they perform as many WMS-requests as there are "time slices" on the WMS-server, then store the result and display one after the other in an animation.
Best regards,
Bart.
On Sat, 2007-05-05 at 22:43 -0300, Tim Sutton wrote:
Hi
Yes a couple of paragraphs of your ideas for itinerary modelling would probably make your requirements a bit clearer. Currently the openModeller library has no in built support for temporal analysis. In a future release of openModeller Desktop, I will add support for projection of niche models into multiple 'time slices' and producing an animation of the result. Typically we would use this when predicting changes in distribution in future climate scenarios.
PS I dont know if I mentioned the new linux binary thats available with the 1.0.5 release of openModeller.
Regardsbiogeosdi@lists.tdwg.org
Tim
2007/5/4, Meganck Bart bart.meganck@africamuseum.be:
Hi Tim,
- could OM take in data layers other than environmental data ?
It can use pretty much any raster layer - what did you have in mind?
Well, it would be a bit more difficult than just a raster layer : in fact, we try to gather extra information from the data as we
process
it, so there would have to be some kind of (temporary) storage.
Also, the data would be more time-oriented, not only spatially (as
rasters
are). I don't know if that would be feasible for OM - and if I'm expressing myself clearly here... ;-)
Perhaps I'll make a short text with some explanation of what exactly we would try to do ?
best regards,
Bart.
On Thu, 2007-05-03 at 13:21 -0300, Tim Sutton wrote:
Hi Bart!
2007/5/3, Meganck Bart bart.meganck@africamuseum.be:
Hi Tim,
I have been thinking about how to make the "itinerary" project a use-case for our GeoAppInter workflow.
It's clear that until now I've been working a bit too enclosed -
so
the
Campinas workshop was an eye-opener.
In fact the itinerary implementation would not be that different
from
the demo we did. It's just another way of looking at the modeling
part.
For that I have some questions about OpenModeler :
Cool it would be great to include itineray support.
- how easy or how difficult is it to make your own algorithms for
OM ?
In what language do they have to be implemented ?
Its pretty easy - requires a basic knowledge of C++ though.
- could OM take in data layers other than environmental data ?
It can use pretty much any raster layer - what did you have in mind?
Regards
Tim
best regards,
Bart.
-- Bart Meganck,
ICT & GIS engineer, African Zoology Department.
Royal Museum for Central Africa, Leuvensesteenweg 13, B-3080 Tervuren, Belgium.
Tel : +32(0)2 769 56 92 Fax : +32(0)2 767 02 42 Email : bart.meganck@africamuseum.be Website : www.africamuseum.be
###########################################
This message has been scanned by ICT - Africa Museum
3/5/2007 - Filtered through antispam by ICT
--
-- Bart Meganck,
ICT & GIS engineer, African Zoology Department.
Royal Museum for Central Africa, Leuvensesteenweg 13, B-3080 Tervuren, Belgium.
Tel : +32(0)2 769 56 92 Fax : +32(0)2 767 02 42 Email : bart.meganck@africamuseum.be Website : www.africamuseum.be
###########################################
This message has been scanned by ICT - Africa Museum
4/5/2007 - Filtered through antispam by ICT
--