Adapting the advances in survey technology and visualization to provide new and better products

Transcription

1 Adapting the advances in survey technology and visualization to provide new and better products Duncan Mallace 1, John Dillon-Leetch 2, Lindsay Gee, Maurice Doucet 3, Rob Spillard, Alison Kentuck 4 and Peter Stewart, Richard Hill 5 1. NetSurvey Limited, Cropredy, UK 2. Port of London Authority, London, UK 3. IVS 3D Inc., Portsmouth, NH, USA 4. Maritime and Coastguard Agency, Southampton, UK 5. Applanix, Oswestry, UK Abstract Recent years have seen remarkable advances in techniques and systems for mapping both on land and in the oceans. Laser mapping and scanning systems for topography, sonar technology, positioning capabilities, and computer processing power have revolutionized the way we image our environment. The new techniques and systems produce massive and diverse data sets and the traditional approach to presentation and analysis of data is no longer adequate. These new techniques are being applied to many towns and cities with port, river and sea frontage that are in the process of development. The state of the dock walls and water frontage has historically been achieved by diver surveys and traditional land survey methods. Access to these areas can often be difficult and visibility is normally very limited for the divers. A good example of the scale of this new development is the Docklands region in the East of London, UK. Another example is the surveying of the wreck of the SS Richard Montgomery in the Thames Estuary that has a large quantity of unexploded ordinance and too dangerous for monitoring by diver. NetSurvey Limited and the Port of London Authority have undertaken a number of surveys for Civil Engineering companies to ascertain the state of dock walls using high-resolution multibeam data. The quality and accuracy of the results has astounded the Civil Engineering Companies involved. Recently a survey was also completed using Laser Scanning Data for the topographic features as well as multibeam data for the hydrographic features, both being collected at the same time from the same survey platform. The survey of the entire wreck of the Richard Montgomery was achieved using laser data combined with the multibeam data. Water column data was also collected which provided valuable additional information about the debris around the wreck. The latest time based visualization techniques were then used to show how the condition of the wreck and also the advances in survey technology had grown over the period of four multibeam surveys from 2002 to today. This paper describes the process and the results of these surveys. It details the issues involved with surveying, processing, analysis and products in these types of surveys and how they were overcome. Introduction The continuing advances in sonar technology and software capability have provided a platform for creating products not just for the traditional hydrographic survey market but also for new markets like coastal engineering, civil engineering and also fisheries research. They allow for structural monitoring as well as seabed monitoring over time. They enable us to visualize objects in completely new ways and to collect data that was hitherto unavailable.

2 The adaptation of multibeam technology has been one of NetSurvey's unique abilities from the start of the company and by partnering with the Port of London Authority the two companies have been able to educate and demonstrate to some of the largest engineering firms and government agencies the products that we are now capable of producing. Adapting technology also requires that companies work very closely with both the hardware and software suppliers and to enable this technology we work very closely with Applanix (on the Tightly Coupled GNSS Aided Inertial Navigation and laser technology) and IVS 3D Fledermaus (with the visualization and data processing technology). To achieve the breakthroughs in technology also needs a client who is forward thinking and who can see the advantage in the adoption of new methods. The Maritime & Coastguard Agency (MCA) have been at the forefront of adopting new methods of seabed mapping in the UK and have funded the use of the laser and water column data for hydrographic surveying. Advances in marine mapping technology This paper covers two of the recent advances in marine technology; vessel mounted 2-D scanning lasers and multibeam water column data. Laser data 2D and 3D laser scanners are becoming more accepted and used in land survey work and are easily adapted to be installed on a vessel. They can be interfaced in a very similar way to a multibeam system and produce a similar size dataset. 2D lasers are better for this purpose than the 3D lasers as the 3D lasers are not optimized for rapidly moving platforms. They are designed to be mounted onto a tripod and, using their built-in motor, they rotate collecting data as they go. This makes for efficient land survey data collection but the rotating motors need to be locked into 2D mode for use on a moving vessel. There also appears to be more of an internal delay in the 3D scanners than there is with the 2D ones. Fig1 shows a Riegl Q240i 2D scanner installed on the Port of London survey vessel Galloper. Fig 1 Riegl Q240i 2D laser scanner Combining the robust and accurate geo-referencing and motion compensation afforded by POS MV, together with the point cloud data from scanning laser and multibeam sonar, as well as georeferenced video, a seamless model of the marine environment, both above and below the water line can be built up. The LANDMark Marine solution serves to geo-reference video and laser data. It utilizes

3 videogrammetry (parallax) techniques to derive pixel locations in successive image frames thereby enabling precise measurement of objects in view and the location and orientation. The Applanix POS MV together with POSPac MMS, enables precision geo-rectification of video and laser data and allows the operator to output data in any coordinate reference system. POS MV provides the raw GPS and inertial observables in a form suitable for post processing with Applanix POSPac MMS. These logged observables are later merged with the GPS observables from a reference station(s) in post mission. Applanix s unique IAPPK (Inertially Aided Post Processed Kinematic) provides the most robust and accurate method for positioning to centimetric level. One major advantage of the IAPPK approach is the requirement for real time telemetry is removed thereby eradicating the line of sight and latency issues that frustrate RTK operations. Furthermore, the IAPPK approach allows the bridging of short-term GNSS data gaps (using forward and backward smoothing techniques) to ensure the most robust and accurate position and orientation possible. LANDMark Marine comprises a data acquisition and a data processing component. For real time data monitoring the system provides essential data related to camera exposure and frame adjustment, and position and orientation solution. The operator view allows not only the data being acquired but also system status which offers alerts to various anomalies and minimizes error on the first data pass virtually eliminating data re-acquisition. During post processing, the operator can scroll through the video and stop at a particular frame to zoom in on the area of interest (then a sub menu can be used to add metadata (type, height, width, condition etc.). Next, database based automatic recognition can be used to complete the analysis, aided by the laser data. Outputs can be in the form of video movie or stills and reports suitable for CAD/GIS systems, and laser point clouds with absolute coordinates. The laser data acquired has a slightly greater density than the multibeam data and therefore software that can handle billions of points in an easy spatial access structure is required. The Pure File Magic (PFM) data structure built into Fledermaus is the most efficient data storage structure for editing very large datasets. The single PFM can contain both laser and multibeam data. This enables easy identification of outliers (if a feature is shown on more than one line it is normally a real feature and not noise). Modern processing tools such as CUBE aren t very applicable to civil engineering structure surveys so having an intuitive 3D editor couple with a dynamically updating DTM structure is really essential. Fledermaus 3D editor enables coloring by line, attribute (for instance the intensity of the laser data) and when data has been rejected or accepted the surface updates automatically. It is also extremely useful to be able to bring into the 3D environment ancillary data such as aerial photos, CAD files and maps/charts to help with the data identification. Aerial photos in particular are a tremendous asset when editing the laser data. Fledermaus contain not only display this data but allows for project management of the ancillary data so that the location of the metadata in relation to the multibeam/laser data is shown. This speeds up the identification of the correct aerial photos for instance as only the one required needs to be converted for the 3D scene. Water Column multibeam data Specific multibeam sonars from Kongsberg Maritime and RESON have the ability to capture full water column data. By water column we mean acoustic imaging of the water mass and its contents. Sonars have previously been developed for fish tracking but these either had wide steered beams (beamwidth of 12 degs) or were forward looking with a broad transmit beam (+ 20 degs) but with narrow 1.5 deg receive beams. All of the above were focused on the mid water data at the exclusion of the seabed. Multibeam bathymetry sonars include not just the mid water data but also the seabed and the interaction of the acoustic beam with the seabed. In many ways this makes it easier to interpret but produces an enormous amount of data. A single head shallow water multibeam system operating in about 20 m of water produces about 20 GB of raw data per day if time series backscatter is also collected. With water column data the data collection rate is

4 3 GB a minute. Water column has been visualized before for fish mass (Mayer et al 2002) and also for looking at wrecks and its use for providing more information about the sonar performance (Hughes-Clarke 2006 a) but these were purely for visualization and not for any quantitative assessment. For fisheries research you need to be able to quantify the volume of the fish shoal and for hydrographic survey purposes you want to be able to ascertain the least depth and compare the water column data with the beam formed data. It is also very useful to be able to determine the location of rigging and other obstacles for further diver inspection surveys or just to see the deterioration of a wreck. Key problems to solve when working with water column data sets include the wide variety of file formats, the size of the data sets, potential needs for timely assessment, and the development of useful visualization metaphors that can be used to rapidly exploit the source data. A water column processing prototype being developed under a joint project between IVS and CCOM is attempting to solve all of these challenges. Firstly, the tool has the capability of parsing a wide variety of source formats. It also has the ability to integrate associated raw navigation files or processed navigation files (POSPac). In order to allow for initial rapid assessment, the tool converts all raw data to a generic water column format or 'GWC' file. This file contains resampled and indexed data that can be rapidly viewed in a variety of ways in the water column tool. Figures 2 and 3 show the Fan and Stacked views of a single processed survey line. The Fan view is the typical along track view one would see in an acquisition topside. The stacked view is similar to the beam view except that it 'stacks' all the beam data across the water column so that general target shape can be easily scene. Fig 2 Fan View Fig 3 Stacked View The operator then simply selects a bounding box around the desired area that they want to visualize. Exporting data from the water column tool then provides visualization and processing capabilities within tools such as Fledermaus or MatLab. Key object types that can be exported from the tool are the Beam Fan, Beam Curtain, Point Cloud and Volume object. The Beam Fan object (Figure 4) provides a 2-dimensional swath view of the acquired data at any point in time. The Beam Curtain (Figure 5) object provides an along-track, beam aligned view of the data from a specific beam. The Point Cloud object (Figure 6) provides the true 3-D (x,y,z) location of every target sample from the selected volume. The Volume object (Figure 7) provides an ISO surface based shape that is useful in fisheries studies. During export of any of these object types, the operator has the ability to select source level or resampled resolution.

5 Fig 4 Beam Fan Object Fig 5 Beam Curtain Object Fig 6 Point Cloud Object Fig 7 Volume Object

6 Adapting this new technology Tilting the sonar head Some of the adaptations have been made just by using the technology in a different way. For instance, a multibeam transducer is normally mounted horizontally to maximize the swath width. By tilting the transducer head by between 30 and 40 degrees you eliminate one half of the swath but you gain the ability to see further under structures, to look up walls, and to see beneath wreck debris. An example of this is shown below in Figure 8.This shows an image of what was reported to be cracks in the side of a wreck s hull whereas when surveyed with a tilted head it is clear that the hull is fully intact and the cracks were just shadows created from the overhanging debris. Fig 8: Tilted transducer Horizontal transducer note cracks in hull The further advantage of tilting the transducer head is that you can look up quay and dock walls. Many modern office and residential developments are created in old dockyards. The prime example of this in the UK is London Docklands, where huge office blocks are built right next to docks created in the 17 th and 18 th centuries. These civil engineering projects require very accurate survey work to gauge the retaining potential of the dock walls. The angle of elevation is also crucial. Traditionally surveys have been achieved using divers but these have been inaccurate at best and have missed many important features. Collecting an accurate point cloud of bathymetric data is good for showing objects on the seabed or on the wall but it is of limited use for engineers who need profiles and to see fissures and cracks in the wall. To achieve these two goals it is necessary to be able to build a model using the offset distance from vertical as the z component to be gridded into a model. NetSurvey has developed a small utility that converts an Easting, Northing and Depth dataset into a Chainage, Depth and Offset from vertical dataset. An additional advantage of this is that it is much easier for divers to further inspect contacts by laying a tape along the top of the wall and measuring the distance from the start of it, rather than trying to use a GPS position fix. Figure 9 shows the transformation utility.

7 Fig 9: NetSurvey Quay Wall Utility The model can then be used to extract profiles and images can be captured to show features of interest. These can then be included in CAD drawings and survey reports. However, we also produce 3D Fledermaus scene files of the walls, co-located with photography and seabed detail so that the whole picture can be seen in context. The wall models hang in the scene as a vertical image class and truly fuse with the seabed data. These scene files are now starting to be used more than the CAD drawings. Figure 10 shows a screen capture from a typical scene and Figure 11-A shows the point cloud data from the wall that is also delivered. What is missing from these datasets though is the continuation of the model above the waterline. Fig 10: Chatham Docks, UK: Fledermaus Scene showing Seabed DTM, Wall DTM and geo-referenced photos

8 Fig 11-A: Chatham Docks, UK: The same Fledermaus scene with the wall as a point cloud Integrating the laser scanner With the laser data we now have the ability to fill in the missing data from the purely bathymetric surveys but not just as a separate exercise but at the same time as the bathymetric data collection. This saves valuable vessel time not just as an expense but some of the areas are very hard to access. In the UK we are lucky enough to have tidal ranges all round the Island to be able to create a truly seamless dataset by surveying the bathymetry with a tilted transducer at high tide and the laser data at low tide. For surveying in London this gives approximately a 4 meter overlap between the two datasets. Figure 11-B shows the combined point cloud data on a wall Purple is the laser data and the other colors are unique bathymetric survey lines. Fig 11-B: Embankment, London, UK: Laser scanner data purple, other lines show individual multibeam survey lines

9 A further application of a combined multibeam and laser data is in its use for coastal monitoring survey programs. The laser data can collect good quality, high density data from the vessel at low tide when the vessel has to be further out from the shore. This has a number of advantages. It provides safety of navigation information if the survey area is particularly rocky/dangerous area as the protruding rocks can be captured by the laser data and then used for line planning and general online helmsman guidance. This data is also very valid data in terms of data collection and certainly for beach frontage this can save valuable vessel time by acquiring the data that is hardest and least productive by normal multibeam means, which is the shallowest inshore area. Many coastal monitoring programs are looking at cliff stability. This is impossible to get by normal aerial laser scanning and very hard to acquire by land survey techniques due to the inaccessibility of the cliffs. However with a sideways looking laser scanner mounted on a vessel going past the cliff collecting bathymetric data is easy to achieve. The SS Richard Montgomery - an example of combining all the technologies The SS RICHARD MONTGOMERY was a US Liberty Ship of 7146 gross tons. She was built in 1943 by the St John s River Shipbuilding Company of Jacksonville, Florida and was one of over 2700 of these mass-produced vessels built to carry vital supplies for the war effort. Fig 12: Richard Montgomery at low tide Fig 13: Richard Montgomery at high tide In August 1944 the ship was loaded with a cargo of some 7000 tons of munitions and joined convoy HX-301 bound for the UK and then on to Cherbourg. On arrival in the Thames Estuary, the vessel was directed to anchor in the Great Nore anchorage off Sheerness. The ship was to await the formation of a convoy to continue the journey across the Channel. However, on the 20 th August 1944, she dragged her anchor in the shallow water and grounded on a sandbank running east from the Isle of Grain approximately 250m north of the Medway Approach Channel. The vessel grounded amidships on the crest of the sandbank and intensive efforts began to unload her in order to lighten the vessel so that she could be refloated and also to save the cargo of munitions that were vital for the Allies post-d-day advancement. Unfortunately, by the next day, a

10 crack appeared in the hull and the forward end began to flood. The salvage effort continued until the 25th September, by which time approximately half the cargo had been successfully removed. The salvage effort had to be abandoned when the vessel finally flooded completely. The wreck of the SS RICHARD MONTGOMERY remains on the sandbank where she sank. The wreck lies across the tide close to the Medway Approach Channel and her masts are clearly visible above the water at all states of the tide. There are still approximately 1,400 tons of explosives contained within the forward holds. NetSurvey was contracted by the Maritime & Coastguard Agency to survey the wreck as part of the MCA's monitoring program. NetSurvey sub-contracted the Port of London Authority to provide the survey vessel Galloper and experienced helmsmen for the survey. They also supplied the survey vessel Yantlet that conducted the surrounding seabed survey and acted as a mothership. MCA were very interested in collecting data to ascertain the condition of the three masts as these had not been surveyed previously. The Galloper is equipped with an Applanix POS MV 320 Inertial System and a RESON Seabat 7125 multibeam system. Hypack Hysweep was used to acquire, sound velocity correct and to apply the Smoothed Best Estimate of Trajectory (SBET) out from the post processed inertial data. IVS 3D Fledermaus was used to clean the data of outliers, both for the bathymetry and laser data and also to process the water column data. Applanix POS MMS software was used to provide a post processed network RTK solution for the inertial data to produce the new SBET. This software was also used to merge the raw laser scanner data with the inertial data. The bathymetric survey consisted of a straightforward high resolution survey at equal line spacing to acquire data as to how the surrounding seabed had changed from previous surveys (this was conducted by the Yantlet using her RESON Seabat 8125 and POS MV 320). To survey the actual wreck itself numerous survey lines were run around the wreck either side of high tide by the Galloper, initially with the transducer mounted horizontally and then with it mounted 40 degrees to starboard to get better coverage of the wreck. While the head was in the horizontal mode the water column data was also collected. The Riegl laser scanner was installed on the Galloper a few days later and the data from this was collected of the exposed section at low tide with the Reigl laser scanner interfaced to an Applanix LANDMark Marine system, thus combining all possible sensors and the latest technology. MCA were very keen to see the state of the masts as these had not been surveyed before. The survey was performed over two days with the multibeam data being collected first at the high tide and then the laser a few days later at low tide. Figure 12 shows the visible portion of the wreck at low tide and Figure 13 shows the wreck at high tide with the SV Galloper conducting the survey. The laser and multibeam datasets were processed individually for both sensor merging and data cleaning. They were then combined together into a Fledermaus PFM to see if improvements in either dataset could be made by looking at the two together. The two datasets were very well matched in terms of resolution and data density and most importantly they were in the correct location. From the image in Figure 14 it can be seen that the multibeam dataset with the tilted head extended right up to the high tide water level and that the laser data extended as low as the shallowest davit. No features were missed by either system in the overlapping area.

11 Fig 14: Image showing the combined point cloud of multibeam data in white and laser data in yellow The water column data was collected for four survey lines, two in each direction either side of the wreck. You must make sure when collecting water column data that the data files are capped at less than 4 GB in size. If not it is almost impossible to extract them from the data acquisition computer. With the RESON Seabat 7125 the water column data is actually collected by the computer than controls the sonar (the 7P). As an indication to file size each survey line was approximately 100 m long and took about one minute to collect. The average file size was 3.3 GB. With the water column processing tools the Applanix SBET file is ingested into the water column data to correctly position and correct for motion the multibeam data. The beam, fan and point cloud objects were then created. The point cloud object was then brought into Fledermaus where it could be compared to the original beamformed data that had been processed in the normal way. Figures 15 and 16 show the combined point cloud with the bathy/laser data in yellow and the water column point cloud as a depth based color map. Fig 15: Bathy/laser & water column combined davit detail Fig 16: Bathy/laser & water column combined The next two images show the main view of the wreck from the water column data only and the bathy/laser data only. Note that the water column data was collected with the transducer in the horizontal position and therefore does not extend up the masts as much as the tilted head bathymetry.

12 Fig 17: Bathy/laser data only Fig 18: Water column data only

13 Conclusions The technological advances of recent years and their application has provided the hydrographic surveyor with valuable new tools to provide improved and completely new data products to existing clients and it has opened the door to many more potential clients. Mobile laser scanner data can be collected in exactly the same way as multibeam bathymetry data with the same software and peripheral sensors. It is also possible to collect the data simultaneously thus saving valuable vessel time. The laser scanners offer the ability to collect near shore seabed data in tidal areas for either coastal engineering projects or coastal monitoring projects. For civil engineering projects such as river side construction, bridge abutment monitoring or breakwater development, the ability to use both sets of data together to create a seamless model for analysis and visualization is a very powerful tool. The visualization enables the products to be rapidly and easily disseminated by an engineer, harbor master or government official. Water column data has many potential products and more will be found as the use of water column data becomes mainstream. For hydrographic surveying the main use will be with wreck investigations and hopefully this will be the final piece of technology that means that we do not require wire wreck sweeping. For archaeological investigations the more information that can be derived the better the final analysis will be. Due to the massive data volumes that water column data produces, the push will be to put these new tools into the acquisition stage so that it is possible to collect water column data all the time and not just at specific times as is currently the way. The good news for the hydrographic surveyor is that the current tools used to provide traditional hydrographic products can be modified, appended and enhanced to be able to capture data once and use it in multiple ways. This then creates a larger and more varied client base with the benefits that brings to business.