GRASS GIS Sample Datasets

Description and discussion of (future) standardized GRASS GIS 7 dataset - an updated, simplified and extended nc_spm_08, to be used in manual pages, tutorials, courses, code testing, development and OSGeo Live.

User-oriented page with examples and list of existing datasets is at ​GRASS Wiki: GRASS GIS Standardized Sample Datasets.

Contributors: Helena Mitasova, Markus Neteler, Vaclav Petras, Anna Petrasova, Hamish Bowman, [contribute and add yourself]

Proposal

The complete sample data set will include:

• ncarolina_spm_complete location with PERMANENT mapset with baseline data and several additional mapsets with specialized data in NC State plane [m] (see ​North Carolina - EPSG codes)
• world_ll location in geographic coordinate system [deg]
• external data in various coordinate systems

Basic NC location with PERMANENT mapset

This is based on the original nc_spm_08 location with simplified, standardized names for map layers, some updates. To keep the data set simple, more specialized files were moved into separate mapsets:

• ncarolina_spm_base
  • PERMANENT
    • raster
    • vector
  • practice1 (empty mapset)

Common rules

• Names of maps/layers must be the same for all standardized datasets. No additions to names such as _10m or _wake_county are allowed. This also implies that names must be in English, national language is not allowed for national standardized datasets (however, if desired, we can can work on a script which would automatically ​rename multiple maps in dataset and would also find and replace names in documentation and tutorials).
• Description in tables here should be usable as title of the map. Separate details in description, which should not be part of the title, using commas or parentheses. Titles can differ between standardized datasets and can use national language (unlike names).

Rasters

• resolution and extent (in the sense of number of rows and columns) should be the same of very similar for all standardized datasets (obviously, actual geographical extent can be different)
• standard resolution: 10m (marked as std)
• standard rows x columns (cells): xxx (x) (marked as std)

name	area for NC	description for NC	resolution	rows x columns (cells)	note
basins	South-West Wake county	Watersheds derived from NED 30m	std	std
elevation	South-West Wake county	Elevation NED	std	std
elevation_shade	South-West Wake county	Shaded relief	std	std
geology	South-West Wake county	Geology derived from a vector map	std	std
lakes	South-West Wake county	Wake county lakes	std	std
landuse	South-West Wake county	Landuse in 1996	std	std
orthophoto	for CC or rural area	orthophoto (R, G, B, NIR)	1m res or better
soils	South-West Wake county	soil type			should be vector data?

Vectors

name	area for NC	description for NC	feature type	number of features	note
boundary_region	South-West Wake	region boundary	polygon map
boundary_state	State of NC	NC State map	polygon map
census	Wake County	Wake County census blocks with attributes, clipped	polygon
firestations	Wake County	fire stations	points map
geology	Wake County	North Carolina geology map	polygon map
geonames	Wake County	geonames	points map
hospitals	Wake County	North Carolina hospitals	points map
history_markers					move to archeology mapset?
parcels	for CC or rural area
points_bare_surface	CC or rural	bare ground lidar points for interpolation
points_of_interest	Wake County	points of interest (examples?)	points map
railroads		North Carolina railroads	lines map
roadsmajor		Wake County major highways and roads	lines map
schools	Wake County	schools	points map
streams	South-West Wake	streams	lines map
streets	Wake County	roads and streets	lines map
zipcodes	Wake County	zip codes	polygon map

Specialized Mapsets

To be distributed with the ncarolina_spm_base location so that they include PROJ information and are readable by GRASS. Mapsets cannot be distributed without a location because they lack PROJ_INFO.

(note that providing just mapsets did not work, it was very confusing for users. But perhaps the only problem was that that "GRASS welcome screen" cannot unpack (unzip, untar) a mapset and copy it into an existing mapset. (MN: we need a button "Download sample data" in the welcome screen!)

• elevation: several elevation models at different scales, lidar
• landsat: set of Landsat scenes with different timestamps
• networks: vector networking data + LRS
• orthoimg: set of aerial image scenes for image classification, including images from UAV
• modis: MODIS time series with temporal GRASS DB (can be easily generated with ​http://pymodis.fem-environment.eu/)
• climate: climatic time series with temporal GRASS DB
• archeology: sites? historical maps/topography

Baseline world location

Already there: demolocation/ in the source code

• world_ll_base
  • PERMANENT

• specialized mapsets to be distributed with world_ll_base
  • climate
  • landcover

Notes after HVA discussion in 2015 with Hamish

• barebone dataset for OSGeo Live which will include a script to generate the derived data needed for tutorials (?)
• metadata
• maps (layers)
  • exclude layers which can be generated
  • secref elevation? + orthophoto
  • add NC WMS service to GUI
  • soils, geology, lakes just vector
  • elevation, landuse, orthophoto
  • fields - parcel plots - with anonymized names - for secref - planimetry
  • SPOT image
  • elevation and precipitation points into baseline
  • add zipcodes
  • powerlines (extended)
  • separate layers for state and counties boundaries
  • natural earth for latlon dataset (data are on ​github)

Comments

The data sets can be distributed separately or we can have packages with several mapsets or all mapsets, depending on the size. I found that packaging and distributing mapsets without location is not practical, so I ended up distributing the specialized mapsets with nc_spm_baseline or world_baseline - is this OK?. We also need to figure out how to include the original metadata that come with the original data - link to the source in history file may be enough.

The baseline location+mapset should be simple with easy to understand names of map layers, my only issue is the loc_ncspm_baseline name as I am not able to come up with a simple name that would say that this is a location with North Carolina data in state plane meters coordinate system - maybe loc_ncarolina would be better, assuming that state plane in meters is the official coordinate system for NC? But I also keep ncspf for feet and ncutm.

Notes about data sources

There is a lot of data, main challenge now is to select a consistent, meaningful, but not too large data set. Many data sets are regularly updated and new ones are posted but tutorials and man pages require stable data to work the history file should include link to the original data source with a note that an updated version of the data map be available from there

Significant natural heritage areas and natural heritage element occurrences posted on NC one map.

TODO

List of actionable items.

• replace historical markers with historical places (done?)
• Keep in mind what tasks we want to do with the data when selecting them (e.g., table join, selections, buffering etc). This influences the choice.

Integration with GRASS GIS unit test suite

All tests in the "gunittest" environment (see also ​overview) need to be written in a way that the map names correspond. See also

​http://grass.osgeo.org/grass71/manuals/libpython/gunittest_testing.html#data

Download of draft location package

• 2014: TBD
• 2013: ​LOCATION ncarolina_spm_base0.3.1.zip (40MB)