A New Method of Chart Validation & Wreck/Contact database creation using 3D visualisation and Image Draping techniques Duncan Mallace 1 Ian Davies 2 1 NetSurvey Limited, Oathill Farm, Cropredy, Banbury, Oxon, OX17 1QA, UK duncan@netsurvey.co.uk 2 UK Hydrographic Office, Admiralty Way, Taunton, Somerset, TA1 2DN, UK ian.davies@ukho.gov.uk Abstract Two of the most important tasks during a hydrographic survey are to compare the hydrographic survey data that has just been collected with the current chart and wreck database. The sounding detail shown on the largest scale published chart of the survey area has to be critically examined and any significant differences reported. In particular, a comment is required for any charted dangers that were not discovered during the survey, or where the least depth found over a danger during the survey is deeper than charted. Any other errors, ambiguities or other defects have to be reported. All significant shoals together with suspected wrecks located during the course of the survey must be reported (with respect to position, orientation, extent and least depth). A definitive list of any newly discovered shoals is presented in the Report of Survey. By using the latest image visualisation techniques it is possible to drape the current nautical chart onto the hydrographic survey digital terrain model. This allows the possibility to show the difference between contour lines on the chart and on the DTM and in this way seabed movement can be observed. It is also possible to check to see if the shoalest sounding over a wreck or shoal is shallower than that on the chart. Wreck positions can be examined to see if they are in the correct location. Once a wreck or shoal has been discovered it is possible to open up the soundings that relate to that wreck, check to see if any soundings have been incorrectly deleted and then select the shoalest sounding. This sounding can then be checked against the depth value shown on the chart. Previously to create and collate a wreck database, information was taken from the survey package and manually entered into a software database, such as Access or SQL. It is now possible to create a contact file online. The contacts can be visualised in the display along with the bathymetric DTM. It is then possible to interactively re-assign individual wrecks positions to specific soundings and then to add metadata about the wreck. It is also possible to add new wrecks or shoals to the listing. This procedure can be undertaken during the data processing or Quality Control stage and enables a very accurate list to be created in a greatly reduced time. Both methods for chart and wreck validation offer significant improvements in both time and accuracy over traditional methods. They also embed themselves easily into the survey data processing and validation techniques now being employed by survey companies and hydrographic offices. Introduction Since the earliest days of hydrographic surveying, the key tenet of any survey undertaken has been the requirement to provide to provide the mariner with the ability to safely navigate in ports, harbours and around the shorelines of maritime states, by providing an impression of the seafloor in the form of a nautical chart. The main problem that the surveyor has always faced is not that of shallow water, which can be systematically mapped with anything from a leadline to a multibeam, but more the issue of contacts that may be a hazard to navigation. By far the most difficult task has been the localisation, identification and determination of least depth of these wrecks and obstructions. Obviously if all wrecks were like Figure 1, we would be able to position it and unequivocally know what it was!
Figure 1: MSC Napoli off South Devon Coast (MCA). In UK waters alone, the UKHO database holds over 29000 wrecks, worldwide this number rises to over 69000 losses increasing by an average of 100 a month, according to Lloyds Register of Shipping. Modern day vessels over 300GRT are obliged by IMO SOLAS Chapter 5 requirements to carry location and distress equipment which allows the mariner to quickly alert the authorities in the event of an emergency. Obviously, in the event of a vessel sinking this may assist in the more accurate estimate of where the vessel was lost. Current estimates of the positions of wrecks and obstructions are affected by several factors: a. Date of loss/reporting. b. How the position of the loss was established. c. Additional errors introduced by datum shifts and metrication on modern charts. In fact there may be several other issues that affect the final resting place of all but the heaviest objects sinking: a. The shape of hulls means that vessels rarely sink straight down, the movement through the water often creates a horizontal or corkscrewing motion which can carry a vessel away from the point of submergence. b. Air and reserves of buoyancy may mean that the vessel once on the seabed may remain relatively mobile until it beds into the seabed. Historical The introduction of the echo sounder in the 1920 s saw a step change in the way that surveys were conducted and allowed a more systematic approach to surveying the seafloor and wreck searches, provided of course that you could locate the position in the first place. However, you still could not positively identify the vessel and produce a definitive seabed topography. It was not until the introduction of electronic aids to navigation in the mid part of the last century, that the surveyor was able to perhaps more accurately position himself at sea and therefore survey the seabed and report or locate contacts more accurately, particularly out of sight of land. At best though, the accuracy was in the hundreds of metres. So how were these surveys conducted? Systematic surveys in the 50 s-70 s by the Royal Navy were scale and depth driven. Many surveys were conducted at a scale of around 1:250,000 (line spacing of up to 2500m) with interlining if needed. In some instances where the seabed was flat this spacing was opened further. Surveys conducted in this manner would provide an indication of bathymetry and slope and any shoals or contacts found would be interlined to obtain a least depth, although positional accuracies at best remained poor.
Contacts if located were close sounded and if depths were less than 31m, wire or chain sweeping was undertaken to obtain or ensure a least depth was obtained and a final accepted position. The issues of identity and the accuracy of the final accepted position still remained, The introduction of Decca and then Hifix and Hyperfix went a long way to solving the positional issues, but if the position of any wreck or shoal was only poorly known, you still had to find them! In an attempt to narrow down search areas, the Royal Navy adopted a systematic approach to conducting disproving searches for shoals, vigias and wrecks, which is still in use to this day. A radius of search for a reported danger depends upon a reliable assessment of the positional accuracy of the original report. If the position is reliable, the radius of the circle of search should be 2.5 times its accuracy. When acceptable data on the accuracy of position is not available, the following criteria are applied: a. When within sight of a well defined land area, the radius should be taken as a minimum of 1.5 nautical miles, although 2.5 miles should be used if time permits. b. When out of sight of a well defined land area the date of the report of the danger will have a significant effect on the area to be searched. A table of the likely accuracies has been devised from assessments made in the UKHO and elsewhere of the aids available for ocean navigation. Date of report of danger or shoal sounding Standard error in position (minimum search radius) in nautical miles On continental shelf Off continental shelf pre 1830 18 20 1850 18 20 1895 17 19 1905 15 17 1915 13 15 1925 11 13 1935 9 11 1945 7 9 1955 5 7 1965 3 5 1975 3 5 Table 1: Radius of search for searches (courtesy of UKHO). This helped to reduce the size of searches undertaken, but if you couldn t find or didn t know what contacts lay between your lines of soundings, it was difficult to say that an area had been fully surveyed or that you had found the shoalest point. The mid 1970 s saw the introduction of the first sidescan sonars; towed behind the survey vessel, these systems for the first time allowed the surveyor to look between the lines of soundings, identifying any contacts for subsequent investigation and also highlighting areas of different seabed type and shape. Initially recorded on wet paper the results were dependent on where the wreck lay within the field of ensoniifcation and on the sonar s orientation in relation to the seabed, which had a significant effect on the image quality. Weather also had a significant part to play and as can be seen from Figure 2 it can be difficult to identify the wreck or obstruction.
Figure 2: Early Wet Paper Sidescan Image (UKHO). Dry paper recorders and eventually digital methods of recording data significantly improved the quality of the visual presentation of the data with the ability to mosaic adjacent lines, so data could be looked at from merged multiple lines, giving more information on the dimensions and other identifying features on the wreck or obstruction. Figure 3: Dry Paper Sidescan Record (UKHO). However, this did not remove the fundamental problems of identifying the wreck and defining the least depth and its accepted position. Sidescan sonar provided a good representation of the wreck that may allow positive identification of the vessel, but the calculation of height above the seabed using the shadow is difficult, and it was commonly accepted that two methods would be used to determine the depth. Close sounding involves running multiple lines of soundings over the wreck, correcting for tide, generally at around 10m line spacing in an attempt to define the shoalest position. The second was to use wreck sweeps and multiple passes to identify the shoalest points (Figure 4).
Figure 4: Echo Sounder Trace and close sounding plot (UKHO). This however would not provide identification or confirmation however that you were actually sounding over a wreck or sandwave if your sonar picture was degraded. In the late 1980 s the Royal Navy invested in Hydrosearch (a forward looking sonar developed for wreck investigations to remove the requirement to wreck sweep and aid identification and classification of contacts. Imagery however was difficult to interpret and although measurement tools aided in identification, it was still difficult to confirm vessel identity (Figure 5). Figure 5: Hydrosearch Imagery (UKHO). The latter part of the last century saw positional accuracies continue to improve with the advent of GPS and towfish tracking, along with the introduction of high frequency sonars it was now possible to identify and position the majority of contacts to a high accuracy.
Figure 6: High Resolution Sidescan imagery (UKHO). Coupled with continued developments in acoustics technology, sidescan sonars began to provide higher and higher resolution imagery (Figure 6), providing a hitherto unseen level of detail for classification and identification. This period also saw the introduction of the first multibeam on a commercial scale. The explosive growth in computer technology and software design meant that these systems rapidly became mainstream tools in surveying, allowing the full coverage of the seafloor and precise location of wrecks and other contacts. In fact in one survey conducted in UK waters, a search of wrecks and obstructions in the areas showed 250 contacts in the area. More than 60 new contacts were identified using this technology. The systems and processing packages now available, for the first time allowed us to obtain full seafloor coverage and for contacts; position, dimensions and least depth in almost the first pass giving a much higher level of detail using both the bathymetry and limited backscatter information now available (Figure 7). Figure 7: EM1002 Contact Imagery (UKHO). In many instances this has been sufficient to identify most wrecks and obstructions and define the shape of the seafloor. However, growing environmental and scientific interest in the seabed and its environs and in limiting damage to the environment from recently sunk vessels or to provide safety statements on current hazards to navigation is leading to calls for even more classification and detail. Early software processing packages allowed multibeam bathymetry to be processed in 2D format and to visualise data in sun illuminated views. In the instances of larger contacts and for topography this was fine, but what about the contact that was broken up or the topography was a more difficult rocky seabed with multiple returns and false echoes.
The traditional bathymetric comparison conducted by the surveyor against the standard nautical chart and source data from previous surveys also became extremely difficult. Traditionally, overlays from the chart would be blown up to survey scale and point assessments would be completed by the Charge Surveyor with reasonable confidence (Figure 8). Figure 8: SBES Survey overlaid on Admiralty Chart for bathymetric comparison (UKHO). Multibeam now meant that there may be several hundred soundings in the footprint of the charted sounding making comparison difficult. In fact early surveys rendered often missed key soundings due to the density and complexity of the seafloor that was being seen for the first time (Figure 9). Figure 9: The density of soundings despite showing the shoal does not show clearly the fact that several areas are up to 10m shoaler (UKHO). New software packages such as the IVS Fledermaus allowed data to be viewed in 3D and from any angle (Figure 10) and allowed both the surveyor and hydrographic office for the first time to consider other ways of comparing new and existing bathymetric data.
Figure 10: Visualisation of wrecks in Fledermaus (UKHO). Present Day Procedures This section contains the procedures and techniques that the British Royal Navy and also NetSurvey follow when performing chart validation. When a survey is issued a Hydrographic Instruction is sent out to those involved in the survey. Contained within this instruction is a listing of all known or possible wrecks within the survey area and also a geo-referenced image of the lastest chart for the area. During the course of the survey any significant features that are not on the chart and would pose a hazard to the safety of navigation of vessels in the area must be reported to the UKHO via a Hydrographic Note form called an H102. Significant features include un-charted or incorrectly position shoals or wrecks. At the end of a survey a Seabed Features and Contacts section of the Report of Survey details any major changes to seabed topography and a full wreck listing for the survey area. The multibeam data collected during the surveys is processed and quality controlled (QC) using IVS Fledermaus software. Even the smallest contacts are visible by displaying the standard deviation or CUBE uncertainty values for each cell in 3D. The geo-referenced chart is also brought into the 3D scene. The chart is referenced to see if the newly found feature exists on the chart or is in the correct location on the chart. Once a suspect area has been found the soundings are loaded up into the 3D editor and rogue soundings rejected. An option is then taken to select the shoalest sounding over the wreck or shoal. This sounding s depth and position can then be compared to the one on the chart and if significantly different an H102 can be filled in. The information is sent to the UKHO as a clearly described document with both chart and seabed together as well as 3D views of the feature. Figures 11 and 12 are taken from a recent H102. Fig 11: Image showing position and depth of shoal Fig 12: View from East As well as reporting the shoal a Target can be added at the shoal location. These targets are stored in a single XML file. A screen capture of the feature or sidescan image of it can be appended to the file. The target can file be created prior to survey operations with the existing wreck database. This makes it easy to see if wrecks are
previously uncharted or in the wrong location. If a wreck is in the wrong location the position can be updated by selecting the shoal sounding and moving the target location to the shoal sounding. Fig 13: Contact shown in sounding view These targets can be clearly viewed on the surface of the DTM as well as within the 3D editor. Fig 14: Targets shown on top of DTM When the Report of Survey is being written this target file can be easily incorporated in the report saving many hours of checking an entire survey area for contacts. One survey contract that NetSurvey currently holds with Marin Matteknik (www.mmtab.se) is the Routine Re- Survey contract in the UK. This contract s aim is to survey the continually changing sand banks that are predominantly located around the South-East of England. This area is also one of the busiest for shipping traffic around the UK. Significant movement of the seabed exists and this must be shown and documented. To rapidly and intuitively gain an understanding to the change in seabed topography a new technique was established that uses image visualisation techniques that were previously limited to high end computers to view and work with the chart in an interactive manner. The end result of the multibeam processing is a DTM that accurately defines the seabed topography; in our case a CUBE (Combined Uncertainty and Bathymetric Estimator) derived DTM. The geo-referenced image of the chart is draped onto this DTM. If the chart image was just texture mapped onto the DTM the seabed
features viewable using sun-illumination would be lost and all that would be left is a chart that has then shape of the DTM. Fig 15: Chart texture mapped By using a technique called Cut and drape the image can be draped over the DTM but the sun-illumination works with the image to show the seabed features, thereby giving you the best of both worlds. Fig 16: Chart draped over DTM with contours applied With the chart draped onto the DTM the object can be loaded into Fledermaus to view in 3D. Contours are then created at the same interval as on the chart and colour coded for ease of identification. The contours are automatically draped onto the DTM as well and this make comparison simple. Changes between the chart contours and those derived from the latest multibeam survey show the extent and direction of seabed movement and the creation or new shoals or deeps. In addition to the Report of Survey we now deliver a Fledermaus scene file that can be viewed using the free viewer iview3d showing the draped chart, contours, wreck locations and their number, grab samples and their description and also backscatter mosaics draped onto the DTM. It is hoped that rather than the scenes being used by the Bathymetric Data Centre (BDC) who validate the surveys, the scenes are used by the Chart compilers who don t have the same appreciation of the technologies that are involved as BDC possesses. The scene is intuitive, interactive and easily understandable to the non-surveyor. This technique is also being used by Trinity House in the UK to determine when navigation buoys need to be re-positioned due to the changing seabed. Conclusions Modern swath sonars collect massive amounts of data and it is the inherent density of swath sonar data that allows us to make the transition from numeric data to image data and thereby develop new visualisation techniques. It is now up to us to change the way that we work so that we can take advantage of these
developments. By utilising the ability of modern software to display and interact with both bathymetry and imagery information simultaneously the task of chart validation and feature inspection has been made much easier and very much quicker. Wrecks are easy to locate using the 3D visualisation and draping the standard deviation or CUBE uncertainty. The 3D editor make selection of the shoalest sounding trivial and allows intelligent editing of rogue soundings around the wreck/shoal. These features can be checked against the latest chart by displaying them both correctly geo-referenced together. The ability to keep a running database of all wrecks and features that is intuitive to create significantly reduces the amount of time spent collating the information for a Report of Survey. Draping the latest chart onto the DTM and then creating contours based on that DTM makes it easy to ascertain changes to the seabed topography and possible incorrect wreck positions. When combined with the 3D visualisation it makes it easier for the non-hydrographer to understand the technologies that underlay the data collection process by allowing them to view their traditional chart product together with a modern DTM.