1. Video-derived Navigational and Recreational CSIs at Teignmouth Mark Davidson – University of Plymouth, UK Ismael Marino-Tapia - University of Plymouth, UK The CoastView Project

2. Contents 1. Navigation Frame of reference. Algorithm development Location of channel marker buoys Location of navigation channel & hazardous sandbars Associated CSI 2. Recreation (Bather safety) Frame of reference Location of bathing hazards Hydrodynamics (2HD model-aided waves + currents) Hazardous sand bar location Associated CSI 3. Information delivery (Web Page) Week to a page Last six Springs Met-Ocean data

3. The frame of reference Strategic objective: 3. Intervention procedure 4. Evaluation procedure CSIs Monitoring System knowledge Measurement Modelling Operational objective (s) Reference state Current state 1. Quantitative state concept 2. Benchmark Long-term management vision & policy Describes how part, or all of the strategic objective will be achieved in a four- stage process

4. “Improve maritime safety and avoid human, economical and ecological disasters” Video derived buoy positions 1.CSI: Sandbar-Buoy Inter-distance (SBID) 4. Evaluation procedure SBID < X m / Invade buffer region Current SBID and channel orientation 3. Move buoys2. Benchmarking exceeded Frame of reference: Navigation at Teignmouth “To ensure that the buoys accurately mark the channel perimeter to minimize the possibility of ships going aground” Video derived channel location (from intertidal contours and breaking patterns)

5. “Improve maritime safety and avoid human, economical and ecological disasters” Vessels grounding on sandbanks and reefs Inappropriate position of channel markers and poor signalling of hazards  Buoys drag and move from original position (specially during storms)  Bottom accretes and channel position changes

6. “Improve maritime safety and avoid human, economical and ecological disasters” Vessels grounding on sandbanks and reefs Consequences Karachi, PakistanJuly 2003Stranded on channel perimeter at entrance of port Ship broke, spilling 12,000 – 15,000 tons of crude oil. Biggest spill on Pakistan history WhereWhenHowConsequences Santa Fe, Galapagos I., Ecuador Jan 2001Ran aground while steering into harbour 3 million litres crude oil spilled. Back

7. “To ensure that the buoys accurately mark the channel perimeter to minimize the possibility of ships going aground” Navigational problems at Teignmouth Sanbanks and channel are very dynamic features  Difficult to position buoys adequately relative to channel  Hence to maintain an effective dredging strategy In the past there have been a few cases of ship grounding at the site. Our aim is to help the manager avoid a potentially catastrophic situation Back

8. Video recognition of navigation channel markers Topics to cover Algorithm for extraction of buoy position. Algorithm for extraction of buoy position.  Concept Concept  The algorithm at work The algorithm at work Examples of data retrieved and data quality considerations Buoy 2 Buoy 4

9. Video recognition of navigation channel markers Algorithm for extraction of buoy position: The concept Visual characteristics of buoy vary greatly depending on: Ambient light, tidal stage, poor visibility (rain, fog), obstructions, etc.

10. Video recognition of navigation channel markers Algorithm for extraction of buoy position: The concept 1.Reduce the search area 2.Isolate red band 3.Detection of buoy i=find{max[(I μ(x) – I min(x) ) y ]} 4. Transform oblique coordinates to planview (XYZ tide to UV)

11. Video recognition of navigation channel markers Extraction of buoy position: Algorithm at work CAMERA 4: Inner buoys CAMERA 5: Outer buoys

12. Video recognition of navigation channel markers Extraction of buoy position: Algorithm at work Method is not infallible But is simple and robust

13. Examples of data retrieved: Buoy 2

14. Data quality considerations: Buoy 4 Highly non-stationary time series (variance changes and level shifts). VERY ‘gappy’ structure (only day time, depends on image quality) Traditional methods for outlier removal don’t work (differentiation, FFT filtering, moving averages, etc.)

15. Data quality considerations: Buoy 4 Wavelet analysis Specially useful for treating non-stationary time series. Analysis in the time-frequency domain (identifies time of variance change at a given frequency.

16. Data quality considerations: Buoy 4 Routine for outlier removal Outliers located at energy peaks but use derivative to aid in the identification process. Before modification value needs to be compared with a local average to avoid “data erosion”.

17. A Navigation CSI An interactive tool that allows: Calculation of buoy – sandbank interdistace Geographical location (‘in useful coordinates’) of channel entrance Could function as a guide for dredging activities 30 January 2004 23 January 2004 Low -1.48 m ODN

18. Concluding remarks: Navigation CSI l Finish outlier removal technique (wavelet, derivatives and moving averages) l Make algorithm for buoy detection operational l Create data base for life of the Argus station at Teignmouth l Obtain (IBM?) fixed intertidal contour l Program the interactive tool for CSI calculation/ assessment

19. Risk maps 4. Evaluation procedure Frame of reference: Recreation at Teignmouth “Improve bather safety” “Maintain awareness of potential dangers for beach users, such as regions of strong currents or hazardous sandbanks” Numerical model-aided current patterns Video derived location of sanbanks 3. Update map2. Benchmarking exceeded Risk map no longer valid (changing morphology) Current conditions

20. Bathing Hazards Flow patterns from numerical model Siegle, E. 2003 used hydrodynamic model MIKE 21 and video- extracted bathymetries to model flow patterns under different morphological set-ups. Location of hazardous sandbanks Using rectified images, hazardous sandbaks where beach users might be cut off as tide rises can be identified.

21. Bathing Hazards Risk maps Use numerical model realizations and rectified images of system at low tide to generate simplified risk maps given a morphological configuration.

22. Concluding remarks: Recreation CSI l Simple concept easily accomplished – (preliminary?) l Manager has interest of implementing it on summer 2004 l Possibility of following the whole frame of reference in practice.