Crowdsourced Bathymetry One Solution for Addressing Nautical Chart Data Deficiencies Maxim van Norden, the University of Southern Mississippi Paul Cooper, CARIS USA John Hersey, the SURVICE Engineering Company Abstract Crowdsourced bathymetry (CSB) data may be collected by any type of vessel, using a variety of sonar systems and for myriad reasons. Enlisting the resources of recreational boaters, pilot boats, tug boats, cruise ships, as well as fully equipped research ships in the opportunistic mode, this acquisition of bathymetric data may potentially open data streams of current observations to navigators, cartographers, scientists, engineers, and coastal zone planners. The strengths and benefits of CSB data are the temporal frequency in repetitive and constant observations in heavily trafficked areas, access to an unlimited workforce, availability of critical nautical data for the maritime community within a short timeframe, and engagement of the wider user community that will readily contribute to the mapping of our coastal zone. A team from the SURVICE Engineering Company, CARIS USA, and the University of Southern Mississippi has demonstrated the potential of CSB to process bathymetric data that provide valuable information to users. This paper discusses the use of autonomously collected time, observation, and position data from vessels that operate without regard to bathymetric survey objectives to characterize the representation of seafloor structure. Examples of comparative differences with existing chart data are discussed for ongoing Baltimore Harbor, South Carolina Atlantic Intracoastal Waterway (AICW), and Antarctic Peninsula case studies. Introduction Crowdsourcing is a distributed problem-solving and production process that involves outsourcing tasks to a network of people (referred to as the crowd). This process can occur both online and offline. The difference between crowdsourcing and ordinary outsourcing is that a task or problem is outsourced to an undefined public rather than a specific other body. According to Franzoni and Sauermann [2012]: Crowd science is attracting growing attention from the scientific community, but also policy makers, funding agencies and managers who seek to evaluate the potential benefits and challenges of crowd science. Based on the experiences of early crowd science projects, the opportunities are considerable. Among others, crowd science projects are able to draw on the effort and knowledge inputs provided by a large and diverse base of contributors, potentially expanding the range of scientific problems that can be addressed at relatively low cost, while also increasing the speed at which they can be solved. Using this crowdsourcing definition, it is a difficult leap to think of CSB as a viable alternative to rigorous and systematic hydrographic surveying. But we must think of CSB in terms of technology enhancements. Hydrographic surveys conducted prior to the emergence of global positioning system (GPS) technology relied on shore-based positioning using sextant resections and electronic ranging from geodetically located stations. Position ranges were often lost due to weather and electronic conditions causing breaks in survey calibration and 1
progress. Maintaining survey integrity required many checks and calibrations to deliver a final accurate product. Pre-GPS hydrographic surveys achieving a Total Horizontal Uncertainty (THU) of ±20 m were rare. Today, even the cheapest GPS system available to the public provides a vast improvement in positioning accuracy and reliability compared to the pre-gps positioned hydrographic data on National Oceanic and Atmospheric Administration (NOAA) charts. In addition, approximately 50% of the sounding data shown on U.S. NOAA nautical charts is pre-1940, collected by antiquated leadline soundings and wire drags. But even to survey just the 500,000 square nautical miles of the most navigationally significant Exclusive Economic Zone (EEZ) waters is estimated to require more than 100 years [HSRP 2010, pp.7 8]. Accordingly, as budgetary resources continue to decline, the guardians of purely authoritative geospatial data are beginning to take crowdsourced data seriously. Horizontal and Vertical Uncertainties of Today s Chartplotters Current chartplotter technology used by professional mariners and recreational boaters is well suited to providing crowdsourced bathymetry. Typically a GPS receiver using standalone positioning has a positioning uncertainty of ±10 m at 95% confidence. This uncertainty can be considerably reduced to ±3 to 5 m at 95% confidence using differential global positioning system (DGPS) or wide area augmentation system (WAAS) corrections [van Norden and Hersey, 2012]. Even greater accuracies (decimeters) in position can be attained if the user has the capability to use a state-operated kinematic Real Time Network (RTN) along the Intracoastal Waterway (ICW) or a globally corrected Differential Global Navigation Satellite System (DGNSS) offshore (e.g., C-NAV Corrections Service or the Fugro SeaSTAR). Regardless of the quality of the GPS position fix, GPS-to-transducer offsets >3 m must also be accounted for in the positioning of soundings, or they will significantly contribute to the THU. For the typical single-beam fish finder type of transducer used by the boating public operating in boating weather, other contributors to the THU of soundings (e.g., latency, lever arm of a mastmounted GPS receiver, and boat motion) will not be significant when compared to soundings positioned at typical nautical chart scales [van Norden and Hersey, 2012]. A reasonable assumption for the depth sounder or fish finder measurement uncertainty is ±0.1 m. Most boat operators will know the vertical offset between their fish finder transducer and the waterline within ±0.3 m. Position and depth errors resulting from roll and pitch can be ignored as long as the GPS antenna is not extremely high and motion does not significantly exceed the half-angle of the transducer beamwidth [van Norden and Hersey, 2012]. It is far more difficult, however, to determine tide and sound speed uncertainties, which are significant components of the Total Vertical Uncertainty (TVU), without evaluating a specific geographic area for both tidal and sound speed spatial and temporal variations [van Norden and Hersey, 2012]. If that variation is not extremely large, then a large number of observations in the same locale will provide both a reliable gridded sounding and a reasonable value (<3 ft) for the associated TVU. ARGUS Crowdsourced Bathymetry Crowdsourced data can significantly augment authoritative geodatabases and provide answers to critical mapping deficiencies. The challenge in the marine geospatial sector is to ensure the reliability of crowdsourced data by managing and structuring the process to ensure that it can be confidently relied upon as useable and accurate. Crowdsourcing of nautical chart data has been implemented by several organizations, including TeamSurv (http://www.teamsurv.eu/), 2
Olex (http://www.olex.no), and SURVICE (http://argus.survice.com). The remainder of this paper focuses on the patent-pending Autonomous Remote Global Underwater Surveillance (ARGUS) system, which SURVICE has developed in partnership with CARIS to provide a reliable solution to this challenge. ARGUS provides automated acquisition and processing of crowdsourced bathymetry. The system consists of a compact automated onboard unit designed to universally interface with and process the outputs of a vessel s navigation and depth systems and port this output to a central server. Hydrographic and statistical processing modules facilitate quality control, calibration and error corrections, and filtering for application to a wide range of temporal and spatial interests. SURVICE is currently working with CARIS on the application of the processing modules, which includes the use of CARIS s Bathy DataBASE (BDB) and Spatial Fusion Enterprise (SFE). The ARGUS Crowdsourced BDB (ARGUS CS-BDB) provides near-real-time feedback via web access to the processed incoming data and to the continuous solution sets generated from all participating vessels [SURVICE Engineering Company, 2013]. Originally demonstrated through a NOAA research grant, ARGUS CS-BDB has been in use since 2008 in the United States, and has processed nearly 70 million soundings from recreational and commercial vessels. Participating vessels now range from 6-m high-speed fishing boats to 300-m commercial cruise liners, all of which operate without regard for the ARGUS hardware or for the bathymetry mission. Once installed, the system autonomously uses shore-based wireless access points, cellular networks, or the vessel s existing broadband connection to automatically offload data; the system then uses the same connection to provide computer access to the CARIS SFE server for the processed data and visualizations. The SFE outputs are overlaid onto available nautical charts so the vessel crew and other data users can quickly identify areas where the charted data is not consistent with the most recent solutions of the ARGUS CS-BDB system. We envision two streams for the data to flow through: 1. The data will be collected, streamed back to a CARIS BDB site, processed, then output using CARIS SFE to subscribers on a regular quick-turnaround basis. ARGUS vessels would be given compensable subscription for their contributions. The data would be marked Not for Navigation since it is not provided by an official hydrographic authority and can be used for situational awareness only. 2. The data will stream back to the CARIS BDB site to be additionally processed and analyzed before being provided to a hydrographic authority, where it will be evaluated and entered into the official nautical chart production flow for distribution. Ongoing ARGUS CS-BDB developments include process and throughput sharing for weather and water quality environmental data, as well as onboard systems and vessel dynamics monitoring. The bathymetric correction process continues to evolve and will also take advantage of these and other external inputs. Example Area Baltimore Harbor, Maryland Figure 1 indicates the overall coverage of CSB data overlaid on NOAA Chart 12281 (ENC- US5MD11M) for Baltimore Harbor. These data, about 800,000 soundings, were tide corrected by subtracting one-half of the mean tide range (0.57 ft) and adding vessel draft for all CSB data. A sound speed of 1,500 m/s was assumed for all reports. Figure 2 is a larger-scale view of the Fort McHenry Channel. Note the overall consistency with the U.S. Army Corps of Engineers (USACE) maintained 50-ft channel depths. Figures 3 and 4 indicate areas where CSB data 3
differ from charted soundings on Chart 12281, although time constraints prevented these differences from being verified and the causes investigated. Obviously, however, the prudent mariner should be cautious if operating in the vicinity of these reports, and the value of CSB data to warn the mariner is apparent. Figures 3 and 4 are also of value to the charting and channel maintenance authorities (in this case, NOAA and USACE) to indicate that channel maintenance, a re-survey, and/or changes to the nautical chart may be in order. Figure 1. ARGUS Crowdsourced Bathymetric Data for Baltimore Harbor in 10-m Grids Cells Overlain on NOAA Chart 12281. Figure 2. Larger-Scale View of ARGUS Data Updated to 31 December 2012 for Fort McHenry Channel, Which Is a Maintained 50-ft Channel. 4
Figure 3. Areas Shown Where CSB Data Differ from Charted Soundings by More than 2 ft on Chart 12281. Figure 4. Baltimore Solution Set with Shoaling Indications in the East Channel. 5
Example Area Atlantic Intracoastal Waterway (AICW), South Carolina Figure 5 is an overlay of CSB data on NOAA Chart 11518 (US5SC34M) for the AICW in South Carolina. The data, about 1.3 million soundings, were tide corrected by subtracting one-half of the mean tide range (1.5 ft) and adding vessel draft for all CSB data. The USACE maintained depth is 12 ft. This figure indicates where CSB data differ from the maintained depth on Chart 11518. Once again, Figure 5 indicates the value of CSB data to inform boaters and tow operators using the ICW of possible areas of concern. In addition, the data shown in Figure 5 can also be used as a programmatic tool for the USACE to help determine dredging and maintenance priorities. Table 1 indicates CSB data and AICW differences by state. Figure 5. Overlay of CSB Data on NOAA Chart 11518 (US5SC34M) for the AICW in South Carolina. Note Red Dots Are Areas Shoaler than 10 ft in the 12-ft Maintained Channel. Table 1. Percentage of CSB Depths (by State) Shoaler than the Maintained 12 ft AICW State ARGUS Soundings Under 12 ft (%) Virginia 83,250 3.5 North Carolina 740,735 20.3 South Carolina 1,364,187 9.6 Georgia 680,289 9.9 Northeast Florida 945,511 26.2 Total 3,813,972 15.8 6
Example Area Wilhelm Archipelago/Antarctic Peninsula In many cases, CSB will be the only data in a remote area (such as the Wilhelm Archipelago/Antarctic Peninsula, shown in Figures 6 and 7) because charting authorities may not have the resources to conduct surveys, may not have the requirements, may not have the authority, or may not be physically able to access areas where there are not enough bathymetric or hydrographic data to produce official charts. Note the unsurveyed area label on this Admiralty chart in Figure 7. It is these areas where CSB will be an especially valuable resource to users. In addition, these users will continue to contribute to the body of data, making it more robust and increasing its value through independent verification and validation of previous transits and collection operations. Constant and persistent transits collecting data through an area will provide statistical confidence in measured depths using less-than-surveyquality sensors. International charting authorities will be able to use CSB in production of charts for navigation once the techniques mature. Figure 6. ARGUS Collected Data in the Vicinity of the Antarctic Peninsula. 7
Further Work Figure 7. An Unsurveyed Area (Upper Right Corner) in an Admiralty Chart. CSB Data Are Providing New Information for Remote Waterways Never Before Surveyed. CARIS and SURVICE are in the process of addressing data quality assessments of CSB data by quantifying the measurement uncertainty of water depths and horizontal positions and by providing an estimate of the accuracy of absolute water depth measurements with TPU data. The organizations are also in discussions with the International Hydrographic Organization (IHO) for developing data quality standards for CSB data in terms of IHO S44 Standards and for inclusion into the S-100 data registry for a future Marine Information Object (MIO) layer for the Electronic Chart Display and Information System (ECDIS). Other technical work planned includes the following: Include horizontal offsets between the GPS antenna and the transducer in the processing calculations. Encourage the use of WAAS and DGPS positioning instead of standard GPS positioning. Use available environmental and oceanographic models to correct for sound speed. Investigate the value of correcting modeled sound speed profiles with measured water temperatures from ARGUS-equipped vessels that have a thermistor at the transducer. Investigate the most cost-effective ways to correct for tides from operating NOAA or other tide stations and from historical tide prediction stations. Simultaneously obtain bathymetric data from a hydrographic survey suite and a commonly used chartplotter to analyze depth variances and to determine their significance for charted depths. Quantify the limits of object detection for the purpose of locating marine debris. Develop methodologies for change detection that can be applied to the continuous stream of ARGUS data. Develop metrics for the cumulative use of different coastal areas. 8
Conclusions CSB is providing new information for remote waterways that have not (or not recently) been surveyed, as well as providing continuous, ongoing monitoring for highly travelled areas. This is a value-added resource for the navigation safety of maritime commerce, commercial fishing, and recreational boating. At a fraction of traditional hydrographic surveying costs, CSB has the potential to contribute to improved safety and limited environmental risk. Combined with a constant supply of data from an unlimited crowdsourcing workforce, the steady improvement of sensing system capabilities, and the progression of data correction and assessment procedures, the evolving ARGUS CS-BDB solution sets will continuously approach IHO standards. References Franzoni, Chiara, and Henry Sauermann. Crowd Science: The Organization of Scientific Research in Open Collaborative Projects. Social Science Research Network, http://papers.ssrn.com/, November 2012. Hydrographic Services Review Panel. Most Wanted Hydrographic Services Improvements. Federal Advisory Committee Update Report, USDOC/NOA/NOS, 2010. SURVICE Engineering Company. http://argus.survice.com/. February 2013. Van Norden, Maxim, and John Hersey. Crowdsourcing for Hydrographic Data A Tool for Updating Nautical Charts. Hydro-International, Vol. 16, No. 7, October 2012. Biographies of the Authors Maxim van Norden is the Coordinator of the Hydrographic Science Masters Degree Program, University of Southern Mississippi. He retired from the Naval Oceanographic Office (NAVOCEANO) with 36 years of world-wide hydrographic experience. He is a Certified Offshore Hydrographer and serves on the NSPS/THSOA Hydrographer Certification Board. He has an interest in non-traditional hydrography. John Hersey is the ARGUS CS-BDB project manager and Research and Technology Team Leader for the SURVICE Engineering Company. He spent 20 years with the U.S. Army as a developer of empirical test and analysis methodologies for the evaluation of combat systems, and has spent the last 4 years leading research and development efforts for SURVICE. Paul Cooper is the Vice President of CARIS USA. He had a 35 year career in the Naval Oceanographic Office (NAVOCEANO). He retired as the Director of the NAVOCEANO International Program for worldwide cooperative hydrographic surveys. He is active in Integrated Ocean Observing System and Ocean Observing Initiative activities. Affiliation Maxim F. van Norden Coordinator and Instructor University of Southern Mississippi 1020 Balch Blvd. Stennis Space Center, MS 39529 Tel 1-228-688-7123 Fax 1-228-688-1121 Maxim.vannorden@usm.edu John A. Hersey Research and Technology Team Leader Applied Technology Operation 9
SURVICE Engineering Company 4695 Millennium Drive Belcamp, MD 21017 Tel 1-410-297-2378 Fax 1-410-297-2379 john.hersey@survice.com Paul R. Cooper Vice President CARIS USA Inc. 415 North Alfred Street Alexandria, VA 22314 Tel 1-703-299-9712 ext.11 Fax 1-703-299-9715 paul.cooper@caris.com 10