crowd sourced depth data

Download Crowd Sourced Depth Data

If you can't read please download the document

Upload: cleanerx

Post on 09-Aug-2015

40 views

Category:

Government & Nonprofit


0 download

TRANSCRIPT

  1. 1. 18.05.15 OpenSeaMap Developer Meeting 2015 Jens Kbler Crowd Sourcing Depths OpenSeaMap
  2. 2. OpenSeaMap Developer Meeting 2015 Jens Kbler Agenda File Formats Track File decomposition Systematic error correction Boat Gauge Tide Contour Line Generation
  3. 3. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing File Formats Detecting Formats Fast decision which format has been uploaded Does not read the whole file but a limited amount of bytes Handle Compression Auto Detect which compression has been used Various packing methods
  4. 4. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing File Formats Humminbird Supports multiple sensor in a single recording Which sensor to choose ? Raymarine FSH does not have an absolute time Impossible to do tide or gauge corrections Precision of the position is bad since it must fit a 32bit number Raymarine GPX suffers the same problems
  5. 5. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing
  6. 6. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing File Formats NMEA 0183 Has multiple data sentences for a single measurement Decision about which sentence to chose must be made NMEA 2000 Has a rapid position update without time and secondwise updates of position with absolute time
  7. 7. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing File Formats Lowrance SL2 Precision of the position is bad since it must fit a 32bit number Recorded data constantly reveals multiple depths for a single position Bad position resolution Garmin ADM Multiple tracks in a single file
  8. 8. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing Clustering Tracks - Scenarios A track file may contain a single recording sequence That would be the best Multiple tracks file may reflect a single recording (as done by the hardware logger) A track file may contain multiple recording sequences (i.e. Garmin ADM files)
  9. 9. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Preprocessing Clustering Tracks For statistical reasons it would be best to process a single recording session and break the structure of the files for postprocessing Variance of positions as per recording session Associated error is isolated
  10. 10. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Considerations Errors in data Noise Jumping positioning data Most sensors are Kalman filtered, so this problem is usually not a problem when the ships move Systematical Errors Drift in Depth Sensor measurments
  11. 11. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Real world errors Low speed error At anchor High speed error Water turbulence Currents Preceeding Ship caused water turbulence
  12. 12. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Sensor Positions Depth sensor is mounted below the waterline User has to provide that information Consider GPS offset to the depth sensor
  13. 13. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Gauge corrections Gauge values are available online User may create gauges where no official gauge is present and measured online Gauges may span undivisible sections of rivers Multiple Gauges need a common measurement base Some Rivers have a form of equalized water level
  14. 14. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Gauge corrections
  15. 15. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Ocean corrections Water Depth to LAT Use DTU 10 Tide Model for Mean Sea Level Calcalute offset values to LAT by calculating 19 year minimum Online water level correction during filtering
  16. 16. OpenSeaMap Developer Meeting 2015 Jens Kbler Track Postprocessing Ocean corrections Water Depth to LAT
  17. 17. OpenSeaMap Developer Meeting 2015 Jens Kbler Calculating Contours Considerations Worldwide calculation required Depth measurments constitute a highly irregular grid Official measurments usually are regular Crowd source data is sparse This is a multivariate interpolation
  18. 18. OpenSeaMap Developer Meeting 2015 Jens Kbler Calculating Contours Multivariate Interpolation Partionized Constrained Delaunay Triangulation Idea: OpenStreetMap inshore waters are already partitionized Use the partitions to execute a single triangulation with the borders as constraints to the triangulation. Islands form constraints as holes. OpenStreetMap Water polygons provide the same information for oceans Gebco 100 depth line declares outward boundary
  19. 19. OpenSeaMap Developer Meeting 2015 Jens Kbler Calculating Contours Partionized constrained delaunay triangulation : Example in the Adriatic Sea
  20. 20. OpenSeaMap Developer Meeting 2015 Jens Kbler Calculating Contours Partionized constrained delaunay triangulation : Example in the Adriatic Sea
  21. 21. OpenSeaMap Developer Meeting 2015 Jens Kbler Calculating Contours Partitionized Contours need to be remerged again Find bordering water polygons that have a contour line Merge the contour lines together to form a single contour line that spans the partitions
  22. 22. OpenSeaMap Developer Meeting 2015 Jens Kbler Postprocessing Tracks