Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

GI2012 cajthaml-quality


Published on

12. Sächsisches GIS-Forum
Dresden: 18./19.05.2012

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

GI2012 cajthaml-quality

  2. 2. Agenda1. Introduction2. State of the art of the Czech cadastre3. DUE software4. Estimation of pos. acccuracy of points5. Estimation of areas6. ConclusionsTerminology note: in this presentation the terms uncertainty and accuracy are considered as identical 19.05.2012 GI2012 2
  3. 3. Introduction Data Quality is still marginal, but important in the process of SDI building NMCAs has particular systems (Quality Management Systems) of data production including data quality INSPIRE trying to improve quality standards has to be established in the SDI because of its higher usage and improvement Quality Awareness is rising up with INSPIRE (data specifications, GCM, tec. guidelines) 19.05.2012 GI2012 3
  4. 4. Quality standards in production Internal quality External quality Users Clients PDAs Maps Computers Users Services Tablets Data Capture Production Output Selection Usage Apps Specification Specification Licencing Metadata, Software policy catalogues ISO 19158 ISO 19131 GeoRM, ISO 19115, OGC, … , ISO 19157 metadata GIS GIS, PDAs … Audits Audits Audits Access control SLAs Certification Certification Certification Certification Accreditation Accreditation AccreditationEdited accoroding to: Y. Bedard - Geospatial Data Quality + Risk Management + Legal Liability = Evolving Professional Practices 4 19.05.2012 GI2012
  5. 5. State of the art of the Czechcadastre ◦ DKM (digital cadastral map) - map with the highest positional accuracy with most points in the range of up to 14 cm. This cadastral map is created by new cadastral mapping by accurate field surveying techniques, ◦ KMD (cadastral map digitized by readjustment) - cadastral map, created by reprocessing of the available cadastral evidence. Cadastral parcels are digitized over transformed raster images (digitized points are identified from new and old survey sketches, documentation of detailed survey of changes etc.), ◦ Analogue cadastral map – scanned as raster images of old cadastral maps. As the KMD progresses slowly and is costly, analogue cadastral maps are nowadays digitized into UKM (simplified goal directed cadastral map). The COSMC complied with requests from the Ministry of Interior and Municipalities to maintain the UKM as a simple vector image without attribute values and techniques of KMD. 19.05.2012 GI2012 5
  6. 6. Quality of cadastral mapsQuality code Characteristic (standard coordinate Lineage (source of measured (previous error with description of lineage of points) – in relation to old classes of the point) positional classes and positional mapping technologyuncertainty) 3 < 0.14m Field surveying with agreement of land owners 4 Standard coordinate error < 0.26m Photogrammetry 6 Digitized points from maps at 1:1000 7 Digitized points from maps at 1:2000 8 Digitized points from old maps at Other digitalization, 1:5000 and smaller scales + high surveying with agreement of positional uncertainty points, land owners without agreement of land owners 19.05.2012 GI2012 6
  7. 7. Data Uncertainty Engine Gerard B. M. Heuvelink – professor Wageningen University and Research Centre, Netherland James D. Brown – Institute for Biodiversity and Ecosystem Dynamics, Amsterdam University, Netherland Creation – Harmonirib: DUE software for estimation of ◦ Positional accuracy (uncertainty) ◦ Temporal accuracy (uncertainty) ◦ Attribute accuracy (uncertainty) Data Attributes: ◦ Numerical variables (e.g. rainfall) ◦ Discrete numerical variables (e.g. bird counts) ◦ Categorical variables (e.g. land-cover) Supported file formats ◦ ESRI shapefiles *.shp ◦ Simplified GeoEAS *.eas ◦ ASCII raster *.asc ◦ ASCII file for simple time-series *.tsd 19.05.2012 GI2012 7
  8. 8. Sources of uncertainty Basic cycle – 5 stages = basic steps: 1. Importing (saving) data as objects with attributes Model Model 2. Describingofthe sources of uncertainty Description Params. uncertainty states 3. Defining an uncertainty model, aided by Input the description model data4. Evaluating the quality or goodness of the uncertainty model Model Model Model definition Output 5. Generating structurerealizations of uncertain output data for use in MCS (Monte Carlo Sim.) with modelsData ± U Model ± U Output ± UIn: Brown J. - Results on assessing uncertainties in data and models 19.05.2012 GI2012 8
  9. 9. Possitional accurracy of pointestimation Pos. accuracy of surveyed points Analogue cadastral map as an example Evaluation and comparison of two data sets: ◦ Digitized analogue cadastral map ◦ Universe of discourse = laser scanning data -> Probability Distribution Function creation based on comparison of identical points coordinates difreences -> 19.05.2012 GI2012 9
  10. 10. Step by step approach1. digitization of analogue cadastral map2. acquisition of samples of spatial data in the test area by mobile laser scanning (establishing the universe of discourse of data set),3. point cloud digitization - obtaining corner points of buildings identical with cadastral map content in 3D - they will be used to determine/derive probabilistic error model,4. creation of a 2D digitized design file – MicroStation Bentley SELECT series 2 version was used to digitize 3D design file (this is a simple step - convert 3D file into 2D)5. evaluation of systematic error (bias) – systematic error calculation or spatial statistics (geostatistic) or it’s variogram evaluation,6. determination of probability model parameters7. generation of realizations by the Monte Carlo method 19.05.2012 GI2012 10
  11. 11. Probability Distribution Function Histogram 16 120,00% Sample – buildings from laser scanning = universe of discourse: 14 100,00% Oxy = dx 2 + dy 2 = 2,41 m 12 Position deviation 80,00% 10 1 n σ ( X )  var( X )  D( X ) =  xi  E x  = 1,78 m 2Rate (count) 2 2 8 Variance 60,00% n i=1 Četnost Rate σ = D X  = var X  = 1,33 m 6 Kumul. % 40,00%Standard deviation 4 20,00% 2 0 0,00% Classes [m] 19.05.2012 GI2012 11
  12. 12. Area of a lot estimation Use of the same data sets Calculate area of a lots from laser scanning data -> compare it with areas digitized – to improve values of areas Calculate global or local marginal deviations to announce needs of recheck/resurvey/recalculate areas Important for purposes of: ◦ Taxation ◦ Subsidies (e.g. farmers) 19.05.2012 GI2012 12
  13. 13. Conclusions Calculating tolerances for control measurements of geographic databases – good to check new survey sketches – detect problematic areas Calculating of complicated areas with Monte Carlo simulation is easier then with other ways Improve or confirm estimation of data quality - code of points testing with samples and with realizations from DUE -> output in metadata It could be easy to present positional accuracy also for INSPIRE purposes 19.05.2012 GI2012 13
  14. 14. USE OF THE DATA UNCERTAINTY ENGINE (DUE) BY NATIONAL MAPPING AND CADASTRAL AGENCIES Thank you very much for your attention Dipl. – Ing. Tomas CajthamlMany thanks to:•GEOVAP Pardubice - for laser scanning data and trial software•Bentley Systems - for MicroStation and Descartes trial software 19.05.2012 GI2012 14