Vom Laptop zum Großrechner: Neues in GRASS GIS 7

Embed Size (px)

Citation preview

Vom Laptop zum Grorechner:
Neues in GRASS GIS 7

Markus Neteler
Fondazione E. Mach CRI, Italy
http://gis.cri.fmach.it

In collaboration with:
Markus Metz, Luca Delucchi, FEM

FOSSGIS Konferenz19-21 Mrz 2014

Edmund Mach Stiftung in Trento, NorditalienLeiter einer Arbeitsgruppe von 12 Personen, alles basiert auf Open Source software

Introduction to GRASS GIS

Developed since 1982

Vax 11/780minicomputer

Wie alles begann

Letztes Jahr habe wir 30 Jahre GRASS GIS gefeiertDieses Jahr hoffentlich den GRASS 7 release

Community sprints, even more...

Entwicklung heute sie u.a. so aus wie auf dem Foto:

Community sprints werden regelmaessig veranstaltetSehr effizient durch hohe Entwicklerdichte und Poweruser (docu, Uebersetzungen etc)

Naechste Woche OSGeo C tribe sprint in Wien

GRASS 7: Faster raster processing

Cost surfaces: r.costFrom non-linear (G6) to linear computation time (G7)

Other speed figure:
PCA of 30 million pixels
in 6 seconds on this small
presentation laptop...

GRASS 6.4GRASS 7Was gibt es Neues in GRASS GIS 7?

1. Geschwindigkeit

Kostenoberflaechen

PCA (Principal Comp)

GRASS 7: New LIDAR LAS support

Import of LAS files as
- 3D vector points or
- raster map (using univariate statistics - binning to selected res.)

Georeferencing of 3D point clouds (archeology)

Unterstuetzung fuer die Verarbeitung massiver Datensaetze, wie sie beispielsweise bei Laserscanning entstehen

Hier schnelle DEM Erzeugung mit binning

Auch Georeferenzierung von 3D Punktwolken moeglich

GRASS 7: Faster Vector Topology

Spatial query exampleQuery of vector point mapsGUI: click on vector map, what is there?

CLI: v.what east_north=east,north

GRASS 6.4GRASS 7GRASS GIS ist ein topologisches GIS

GRASS 7 mit praktisch neu entwickeltem Spatial index etc

North Carolina: boundary_county

Vector boundaries: simplification

Original: many vertices

10m threshold: reduced vertices

20m threshold: even less vertices

v.generalize in=boundary_county \out=boundary_county_simple_10 \method=douglas threshold=10.00

Generalization: simplify the geometry
without changing much the shape
(reduce vertices)

Vorstellung neuer Funktionen im Vektorbereich:

Generalisierung, dh. Vereinfachung der Vektorgeometrien

Auch interessant, da immer mehr freie Daten zur Verfuegung stehen, die aber nicht immer 1:1 verwendet werden koennen

Increasing the number of verticesGRASS module:v.generalize

Various methods supported

Vector boundaries: smoothing

Black: original

Hermite

Chaiken + Hermite

Beim Vektor-Smoothing hingegen wird das Gegenteil durchgefuehrt, es werden mehr Stuetzpunkte hinzugefuegt

Vector network analysis in GRASS

Es gibt in GRASS schon laenger eine Toolset zur Netzwerkanalyse

Vector network analysis in GRASS wxGUI

Select points

Select method

Output

Neu ist die grafische Oberflaeche dazu

New Space-Time functionality in GRASS 7

Developer: Sren Gebbert

New Space-Time functionality in GRASS 7

Temporal GRASS (TGRASS) menus

Gebbert, S., Pebesma, E., 2014. TGRASS: A temporal GIS for field based environmental modeling. Environmental Modelling & Software 53, 1-12. (DOI)

New Space-Time functionality in GRASS 7

t.register: Registers raster, vector and raster3d maps in a space time dataset

Example: daily MODIS Land Surface Temperature time series

New Space-Time functionality in GRASS 7

t.register: Registers raster, vector and raster3d maps in a space time dataset

Screenshot: S Gebbert/A. Petras

GRASS 7: New animation tool for time series

Weekly temperature 2012 of Sicily (MODIS Land Surface Temperature maps reconstructed by
GIS & Remote Sensing Unit, FEM)

1460 aggregated maps...

Unsupervised image classification Segmentation

i.segment - Identifies segments (objects) from imagery data.

Vectorize result with r.to.vect

Simplify areas with v.clean's rmarea (topologically merge small areas according to threshold into next polygon)

Unsupervised image classification Segmentation

Different results changing threshold

GRASS 7: New bivariate Scatterplots

LANDSAT 7 2002 channels 1 and 3 of Wake county, NC

http://geo.fsv.cvut.cz/~landa/publications/2012/ogrs2012/poster/figures/

Tool for supervised classification of imagery data.Generates spectral signatures for an image by allowing the user to outline regions of interest.

GRASS 7: Supervised image classification

New Map swiping tool for multitemporal maps

Pre and post disaster images of the tsunami in Japan in 2011(MODIS images taken on February 26 and March 13, 2011)

January and June temperature of Sicily (MODIS LST reconstructed by GIS & Remote Sensing Unit)

Integration:QGIS-Processing Vector dissolving example

Dissolving geometry example:Processing (former Sextante) calls
GRASS GIS in a virtual session which
deliver the result back (here SHAPE file)

GRASS 7: Topological Vector in PostGIS 2

http://grass.osgeo.org/grass70/manuals/v.out.postgis.html

http://grasswiki.osgeo.org/wiki/PostGIS_Topology

Developer: Martin Landa

Cofunded by Municipality of Trento, Italy

New GRASS 7 Python API

http://grass.osgeo.org/wiki/GRASS_and_Python

http://dx.doi.org/10.3390/ijgi2010201

Visualization: GRASS data export
for Augmented Reality

GRASS GIS, Paraview, and ARToolkit

http://it.wikipedia.org/wiki/Imaging_a_risonanza_magnetica

Our FEM colleague -->

Personal comm.

A researcher's brain... From MRT scan
to Voxels

A researcher's brain... From MRT/CT scan
to Voxels

Convert DICOM to PNG

Import PNG into GRASS

Put 2D maps into volume

Export Voxels to Paraview

New cool stuff:
massive data processing

Since 2005 GRASS GIS is running natively on 64bit CPUs

GRASS GIS 7 also offers Large File Support on 32bit Windows

Installed on Grids and TOP500 supercomputers (AKKA Ume,
ENEA Frascati, Aurel Bratislava, )

Runs on Linux, AIX, Solaris, freeBSD, netBSD, ...

Various ways of parallelization

Since we like to think big, why to getting our tools on Supercomputers running...

Indeed, we are already there:PROJ4, GDAL and GRASS GIS are available even on TOP500 systems.

The harder part was to get it compiled on Non-Linux Supercomputers :-)

For massive data processing currently job managers are supported. Work is underway for GPU based clusters which require a major source code restructuring.

Massive data processing:
also for you (even Laptop)

https://plus.google.com/u/0/communities/111147786674687562495

Since the home computer
only has 8 GB of RAM...8.5 billion points...Since not everybody has a supercomputer reachable, a lot of efforts have been done on processing massive data on even consumer hardware.

Here an example from a power user who managed to process 8.5 billion lidar points on his home computer.

With this touching base I would like to conclude my quick ride through 30 years of Open Source GIS development and conclude with...

Where is the stuff?

GRASS GIS Software:Free download for MS Windows, MacOSX, Linux and source code:http://grass.osgeo.org/download/

Addons (user contributed extensions):http://grasswiki.osgeo.org/wiki/GRASS_AddOns

Free sample data:Rich data set of North Carolina (NC)
available as GRASS GIS location and in common GIS formatshttp://grass.osgeo.org/download/sample-data/

User Help:Mailing lists (also in different languages):http://grass.osgeo.org/support/Wiki:http://grasswiki.osgeo.org/wiki/Manuals:http://grass.osgeo.org/documentation/manuals/

Markus Neteler

Fondazione E. Mach (FEM)Centro Ricerca e InnovazioneGIS and Remote Sensing Unit38010 S. Michele all'Adige (Trento), Italyhttp://gis.cri.fmach.ithttp://www.osgeo.org

[email protected]@osgeo.orgEnjoy GRASS GIS 7!

PostGISomics

http://grass.osgeo.org

http://trac.osgeo.org/grass/wiki/Grass7/NewFeatures

Thanks for your attention!

2014, Markus Neteler Luca Delucchi, Italy CC-BY-SA license

million pointsseconds

11.122582817.1

24.4903281168.5

39.161334516.318.5

million pointsseconds

11.28777548.9641.196

25.784252265.0745.195

310.653336490.47514.224