260 likes | 385 Views
Uncertainty visualisation in the Model Web. Lydia Gerharz, Christian Autermann , Holger Hopmann , Christoph Stasch Institute for Geoinformatics ( ifgi ), University of Münster lydia.gerharz@uni-muenster.de. Overview. Introduction Uncertainty visualisation methods
E N D
Uncertainty visualisation in the Model Web Lydia Gerharz, Christian Autermann, HolgerHopmann, ChristophStasch Institute for Geoinformatics (ifgi), University of Münster lydia.gerharz@uni-muenster.de
Overview • Introduction • Uncertaintyvisualisationmethods • UncertWeb visualisationclient • Hands-on Exercises • Vectordatavisualisation • Raster datavisualisation • Howtoprepareyourowndata • Wrap-up& Discussion
Uncertaintyvisualisation Communicateuncertainties in geospatialdatato • allowmeaningfulinterpretationofmodelresultsormeasurementsfordecisionmaking • explorespatialand temporal distributionofuncertainties
Uncertaintyvisualisationmethods Techniques • Adjacentmaps • Bi-variatemaps • Sequentialmaps Modes • Static • Dynamic • Interactive e.g. Animation ofrealisations
Methods (i) – Focus metaphors Contourcrispness Fog Fillclarity Resolution MacEachren (1992)
Methods (ii) – Adjacentmaps Value anduncertaintymapsareshownnexttoeachother Rodriguez et al. (2006) Avoids visual overload, but hard to connect two maps mentally
Methods (iii) – Probabilityofexceedance Descriptivestatistics: UseIPCC (2001) terminologytodescribeprobabilityofexceedance van de Kassteele & Velders. (2006)
Methods (iv) – Stochasticaldimension in a GIS Aguila software • Cumulativeprobabilitydistributionforeachpixelorobject Browse eitherthroughprobabilityorvalues (thresholds) • Cumulative/exceedanceprobability • Confidenceintervals Time seriesvisualisation Scenario view Pebesma et al. (2007)
Other methods Hierarchicalspatialdatamodel Whitening Hengl (2003) Kardos et al. (2003) Confidence intervals
Uncertaintyvisualisation in the Model Web Output Output Input Input Data service e.g. meteorologicalmeasurements Model service e.g. meteorologicalforecastmodel Model service e.g. airqualitymodel Final result Web-baseduncertaintyvisualisationclient
Aimwithin UncertWeb Develop a tool that • enablescommunicationofuncertainties in spatio-temporal datato different usergroups • allows easy integration into model workflows following the Model Web paradigm • visualises inputs, outputs and intermediate steps • supports different uncertaintyandgeospatialencodings
UncertWeb visualisationtool • Interactive, web-basedthinclient • Supports different encodings • Uncertainties: UncertML 2.0 • Raster data: NetCDF, GeoTIFF • Vectordata: Observations&Measurements (O&M) • Open Source, based on JavaScript libraries • OpenLayers (spatial, temporal, spatio-temporal data) • jStat (non-temporal, non-spatialuncertainties) • ExtJS (interactive web applicationcontrols) https://svn.52north.org/svn/geostatistics/main/uncertweb/
Implementation details • Vectordata • Encodedas O&M and UncertML in XML/JSON format • Directlyreadbytheclient • Raster data • NetCDFandGeoTiffcannotbedirectlyreadbytheclient • RESTfulVisualisation Service (VISS) • Create visualisations (raster) fromcomplexsources • Web Mapping Service (WMS) • Stores createdrasters • Providestile-caching • Manyclientsavailable
U-O&M encoding • Uncertainty Observation type toencode UncertML types (distribution, samples, statistics)
NetCDF-U encoding • Encodeuncertaintyasdimensionorancillary_variable • refattributeto UncertML definition
Architecture overview Web client SOS Raster map U-O&M as XML or JSON WMS reference VECTOR DATA WMS Stores createdraster VISS Createsvisualisation Add layer NetCDF-U WCS Stores sourcedata RASTER DATA
Visualisation methods Support for: • Non-spatial & spatialdata • Temporal & Spatio-temporal data • Continuous & categoricaldata • Multivariate data • Different userbackgroundsandexperiences • Different usabilityofvisualisationmethods • Adjacentmapsfornoviceusers • Multidimensional mapsforexperts
Visualisation methods – Basic plots Continuous data Categorical data
Visualisation methods – Adjacent maps Continuous data Categorical data
Visualisation methods – Multidimensional approach Continuous data Categorical data
Usingthetool Menu toolbar Mapnavigation Mapwindow Legend
Addingnewresources 1) By Add Resourcebutton 2) By URL Parameter 2a) http://giv-uw.uni-muenster.de/vis/v2/?url=http://giv-uw.uni- muenster.de/data/netcdf/biotemp.nc&mime=application/netcdf 2b) http://giv-uw.uni-muenster.de/vis/v2/?netcdf=http://giv-uw.uni- muenster.de/data/netcdf/biotemp.nc
Exercises http://giv-wikis.uni-muenster.de/agp/bin/view/Main/UncertaintyVisualisationWorkshop
Wrap-up & Questionnaire http://surveys.ifgi.de/ UncertWeb Questionnaire Part B.1: Visualization Tool Further comments/questions?!