- Automatic Generation Cleaning & Generalization of 50 Contours for NRCan's Geospatial Database
Automatic Generation Cleaning & Generalization of 50 Contours for NRCan's Geospatial Database
As presented by Johann Levesque & Etienne Bonhomme of Natural Resources Canada Abstract: Natural Resources Canada (NRCan) has developed a process to automatically generate, clean and integrated contour lines from Canadian Digital Elevation Data (CDED). Part of this workflow is an impressive collection of 25 FME workbenches that have been developed to drastically minimize the need for manual editing. Manual editing has been reduced in some cases from days to one to two hours. We would like to show you some example of those tools.
<ul><li> 1. Automatic Generation, Cleaning andGeneralization of 50k Contours forNRCans Geospatial Database (GDB)Johann LevesqueProject OfficerEtienne BonhommeProject Officer April 20, 2012</li></ul>
<p> 2. Overview Background Objectives Whats the problem! How did we solve it? Workspaces Overview Examples Did we succeed? Whats next? Questions 3. Background 14 000 map titles for NRCans NationalTopographic Series (NTS) 4000 of them still have imperial contours 800 dont have contours (never published) Contours are the essence of a Topographic Map Essential to complete the mapping of the North =National Coverage 4. Objectives Generate clean contours using one source = CanadianDigital Elevation Data (CDED) Complete the Mapping of the North by April 2012 Increase productivity by automating the process Reduce the manual editing by reversing the ratios 80% editing 20% machine 90% machine 10% editing (ideally!!!)And of course with limited resources! 5. Whats the problem! In fact, its many problems Self intersection Contours crossing Contours in water Ugly contours And many more 6. How did we solve it Using ! Opalis Validation14 workspaces = 1 hour processing Contour AutomaticEditionInspection Delivery Creation Cleaning GDAL: raw ArcGIS ArcGISSDE DB: cleancontourscontours33 workspaces = 5-6 hours processing 53 workspaces!!!6 workspaces 7. Workspaces Overview Load and delete Format conversion Report creation Fill and change attributes Simple data manipulation Loop and embedded custom transformer Complex data manipulation Embedded python script 8. Include lakes into CDED Problematic CDED and Hydrography does not fit all the time Hydrography is not always classified as lake Solution Find lakes with statistics (area, length, CDED elevation) Set lakes flat with a buffer around them Recreate the CDED Bonus Reduce to almost zero intersections with lakes 9. Amalgamate contours Problematic Small isolations = noise Bad data structure Solution Restructure the data then amalgamate thecontours (Python and Shapely done by Centre forTopographic Information Sherbrooke) Bonus Gives clean data with less structure problems 10. Neatline vertices cleaning Problematic Data specification No vertices within 5 meters of the neatline Small portion of contour Surface creation problems Solution Find vertices in error (list) Loop trough the list and remove them Bonus Clean data Minimize support (following workspaces run fine) 11. Surface creation Problematic Not able to set elevation point attributes automatically Not able to set contour attributes automatically Solution Create surfaces with contours and neatline Find type of surface with CDED Bonus Attributes are set automatically We save hours of editing on huge file 12. Create contours on summit Problematic Can miss contour on summit to include elevationpoint Use circle or same contour with negative buffer Solution Find summit Use CDED to find the best shape Bonus Better contours All attributes set automatically 13. Adjust re-entrance Problematic Linear rivers does not match the CDED Solution Set linear rivers orientation with CDED Move contours to fit rivers (Python and Shapelydone by Centre for Topographic Information Sherbrooke) Bonus Correct small re-entrance problems automatically.Re-entrance can be 25% of overall errors 14. Neatline snapping Problematic Contours does not connect to the correspondingcontours on adjacent map Solution Find matching contours on both side Snap them and check for intersection Bonus Operators does not have to tie them manually Can save moving hundreds of lines 15. Did we succeed? We think so! We are around 65% machine-35% editing. Prettygood We think we can do better We have beaten our April deadline by 3 weeks Analysis on 14 files (before and after) Mean number of errors before: 1452 Mean number of errors after: 508 Reduction of 65% 16. Did we succeed? 17. Whats next? We think we can achieve 80% machine 20%editing when we solve the hydro intersectionproblem (double line and single line rivers). To reach 90% machine 10% would require betterdata. There is so much fudging we can do 18. Thank You! Questions? For more information: Johann Levesque Johann.Levesque@NRCan-RNCan.gc.ca Natural Resources Canada WWW.GEOGRATIS.GC.CA WWW.GEOBASE.CA 19. Error report sample 20. Simple data manipulation 21. Complex data manipulation 22. Include Lake into CDED 23. Amalgamate Contours 24. Neatline vertices cleaning 25. Create Surface 26. Create Contour on Summit 27. Adjust Reentrance 28. Neatline Snapping </p>