[GRASS-dev] Re: sample dataset update

[Markus Metz]

On Thu, Jun 23, 2011 at 6:44 AM, Helena Mitasova <hmitaso at ncsu.edu> wrote:
> Below are some plans for data updates. It would be useful to put this on wiki
> so that others can comment, add, modify. I was thinking about
> http://grass.osgeo.org/wiki/Sample_datasets
> but this is really for downloads, so perhaps as a new section here?
> http://grass.osgeo.org/wiki/GRASS_7_ideas_collection

I would suggest to keep GRASS 7 ideas and the sample datasets separate
because 1) we do not know when GRASS 7 will be released and 2) the
additional datasets you listed below are very interesting, also for
GRASS 6.x. Therefore I would tend to add any new datasets to
http://grass.osgeo.org/wiki/Sample_datasets
even though this is currently for downloads only, but the two main
sites for sample data download are
http://grass.osgeo.org/download/data.php
and
http://www.grassbook.org/data_menu3rd.php
so it could be ok to add new ideas and TODO's to the wiki?

I like the idea of having separately downloadable mapsets for the two
existing sample locations. This keeps download size manageable and
users can pick the sample data they are interested in. Of course
additional sample locations are also an option and I think that there
are already some new sample locations in preparation, e.g. for Italy.

Another issue are the examples based on the sample data, both in the
book and in the documentation. All the examples for vector network
analysis that use cost columns are wrong. I am busy updating the
v.net.* manuals for GRASS 7 and will (hopefully) soon backport the
manuals to 6.5 and 6.4. Other examples in the manuals, and the manuals
themselves, may need a closer look too...

With regard to GRASS 7 vector format, I want to change it further and
would therefore postpone a GRASS 7 sample dataset a bit.
Forward/backward compatibility will be limited: rebuilding topology
will be required when switching from 6.x to 7 or back (currently GRASS
6.x can read GRASS 7 vectors as they are). For the existing sample
datasets spearfish and nc_spm, v.build.all takes less than a minute on
my laptop (6.x and 7), so this is not really an obstacle.

For external file formats, we could also provide more links to free
datasources, however, examples based on these data may not work when
these datasources are updated.

Markus M

>
> Feel free to post it on wiki (or suggest where it would fit and I will post it).
> After I get some feedback I will prepare the first version of the new mapsets for networks and timeseries
> to test whether they are suitable,
>
> thanks, Helena
>
> ------------------------------------------------------------------------------------------------
>
> Prepare new data sets (grass7 vector format?):
> - provided as separate MAPSETS that can be inserted into nc_spm_08 or gisdemo_ncspm (or their grass7 versions)
> - packaged all into single location (nc_spm_11) which could reach 1GB.
> - each MAPSET will have an example of application
>
> proposed new MAPSETS:
>
> Road and street networks (nconemap, ncdot)
> http://www.ncdot.org/it/gis/DataDistribution/DOTData/default.html
> LRS road network, Integrated Statewide Road network, Bike Paths
> - streets for wake http://www.wakegov.com/gis/services/data.htm`
> -add boundaries and POI for reference?
> -include emergency centers ? hurricane evacuation routes? potential emergency shelters?
> -include stream network?
>
> Time_series_coast
> coastal lidar point clouds, 1m res DEMs, orthos, 15 years in 10-15 snapshots
> extended version of this
> http://courses.ncsu.edu/mea582/common/media/01/NagsHead_series.zip
> also used for coastal analysis toolbox add-ons
>
> Landuse_img
> new 4 channel 1m res 2009 imagery NAID
> new NLCD 2006 from USGS and NLCD2001
> 0.15m res ortho?
> include any post storm, post hurricane, post flood imagery for training purposes?
> see example here:
> http://www.ncdot.org/it/gis/DataDistribution/SpecialData/NCStormDamageAndCleanup.html
>
> Time_series_census
> this may take more time to prepare,
> data are available for 1970,80,90,2000,2010
> combine with election districts?
>
> LOCATIONS:
> - cleanup and provide nc_ll, nc_utm, nc_spf for testing projections
>
> EXTERNAL data sets for testing external data formats
> - point cloud asci, las
> - shape
> - geodatabase?
> - MRsid, geotiff?
>
> Anything else from here?
> http://courses.ncsu.edu/mea582/common/GIS_anal_lecture/GIS_Anal_webdata.html
> http://www.nconemap.com/Default.aspx?tabid=286
> http://www.wakegov.com/gis/services/data.htm
> (we provide link to this so only data to be used for GRASS tutorials and testing should be in the data set)
>
>
> On Jun 20, 2011, at 11:15 AM, Markus Metz wrote:
>
>> Hi all,
>>
>> we may want to think about an update of the sample datasets. I have
>> recently updated the documentation of the v.net.* modules including
>> examples and noticed that roads digitized as multilanes have wrong
>> line directions, i.e. not matching drive directions. This makes
>> assignment of different forward/backward costs impossible. The vector
>> map roads in nc_spm_08 should also get a new layer with unique
>> category values. The updated, correct examples are anything but
>> intuitive and would be much shorter and simpler with already existing
>> unique categories in roads at PERMANENT.
>>
>> Helena also mentioned the LiDAR-based DEM in nc_spm_08 which needs to
>> be expanded. With the development of a test suite for G7, I am sure
>> that there will be more requests for updates of the sample datasets.
>>
>> Time for a TODO list somewhere?
>>
>> Markus M
>
>