|
|
|
|
|
|
|
© GeoAcoustics 2008
|
Application Note 8 GeoSwath Wide Swath Bathymetry Survey of the Lavera and Caronte Areas in Port Autonome de Marseilles
Situated at the mouth of the Rhone on the Mediterranean Sea the Port Autonome de Marseilles is the leading commercial seaport in France. In 2003 Marseilles handled 61m tonnes of oil, 1.74m passengers and over 800,000 TEUs of container traffic. The international standing of the port and its high level of security was illustrated by the recent visits of the American nuclear aircraft carriers T. Roosevelt and H. Truman.
In April 2003 GeoAcoustics arranged a demonstration of the GeoSwath Bathymetry System for the Port of Marseilles, surveying an area around the crude oil moles at Lavera and the nearby Caronte canal. This involved mobilisation, system calibration and collection of bathymetry and side scan sonar data over a three day period around working berths.
Survey Objectives
The GeoSwath performance as a port survey and inspection tool was to be demonstrated, including:
In this report the GeoSwath survey results are also compared with data collected by a Reson Seabat 8101 beamforming multibeam.
The Survey
Mobilisation
The following peripherals were installed:
Data Collection
While on survey the GeoSwath software allows coverage to be monitored in real time, and several data displays are available for the surveyor to maintain a high level of data quality control. Side scan and bathymetry can also be displayed, allowing detailed inspection of sensitive areas in real time.
software, the GeoSwath screen (collecting data), the CSI DGPS Max, and the GeoSwath dry end box.
Data Processing
The GeoSwath is supplied with a complete set of software for data collection, calibration, filtering, generation of digital elevation models (DEMs) and side scan mosaics. Data quality control and data visualisation tools are an important part of the package.
Raw swath data was filtered to remove outliers, sound velocity corrections were applied and navigation data was checked and edited using the GeoSwath Swath32 software.
The calibration parameters were applied to the filtered swath data, giving georeferenced, calibrated data files on a line-by-line basis.
Gridder32 was used to bin the data in a 1m grid for DEM generation. A 1m bin size gives many soundings per bin when using the GeoSwath, increasing the depth accuracy of the bin and the depth confidence of the binned data.
After gridding the DEM was inspected using the GeoSwath vGrid and GridFly software, and Surfer was used for plots.
Survey Results
The diagram below shows the three survey regions. The main survey area was around the moles in the Lavera crude oil terminal, and this is the survey that is compared to the Reson 8101 data later in the report.
Lavera Crude Oil Moles
The main GeoSwath survey was in the Lavera Terminal around Mole1, the nearby quay wall, and up to Mole2. The moles are used for crude oil and refined products, taking 12m draft vessels up to 275m long (80,000dwt max).
Two views of the Lavera survey. On the left is a GridFly rendered image with a cross section across the end of mole 1. On the right is the soundings chart (generated in Hypack) on the mole 1 berths. This survey area is analysed in more detail in the comparison with the Reson 8101.
Caronte Canal
The Caronte Canal leads from the Gulf of Fos to the Etang de Berre. Along the north bank of the canal is the Port de Bouc – Caronte, which specialises in bulk traffic (mainly ore and chemicals). Maximum draft for the berths here is 9.1m.
The entrance to the canal from Lavera is to the left, the main ore berth is to the right on the north bank. Below are two screen captures from a flythrough.
Feature Detection
During the survey several features were found on the bottom of the port that illustrated the GeoSwath object detection and identification capabilities.
The GeoSwath bathymetry and simultaneous side scan provides the confidence of feature detection required in navigation and inspection surveys in a port. The real-time side scan displays can be used on the vessel to allow the surveyor to spot objects that might need further investigation during or after the survey.
Concrete Blocks in the Lavera Berths
Two concrete blocks were seen just off the quay opposite mole 1. These were first seen in the on-line side scan waterfall. Later investigation of the bathymetry and georeferenced side scan shows these to be two blocks about 2m across and 2m proud of the bottom, with several depth hits on the top of each block.
Three view of the processed data from the blocks
Cars in the Caronte Canal
Side scan and bathymetry showed several very clear, well defined features in the Caronte canal, about 4m long and 1.5m proud of the bottom. These were subsequently identified by the port authorities as cars known to have been lost in the canal.
Quay Wall Inspection at Lavera
The ability of the GeoSwath to image quay walls at high resolution is seen in the point cloud images. These show raw data from one line by the quay walls (to the south of the Lavera moles) being inspected in Gridfly flythroughs. The vertical walls are imaged up to the surface, and the GeoSwath resolves the dock floor up to the bottom of the walls. In the right image a hole leading into the quay wall can be seen.
Survey Accuracy
Positioning
The accuracy of the CSI DGPS Max system used in this survey was specified at ±40cm, so 1m was the highest resolution used for the DEM binning. Depth Accuracy
The survey accuracy is given by the depth repeatability combined with the errors in the static offsets.
Combined measurement error of the static height offsets to the water level was ±2cm. Depths were referenced to water level data provided by the port authorities.
The depth repeatability of the survey was measured using the standard deviation (SD) of depths within a 4m bin on a flat area of seafloor. This gives the combined effect of all the measurement error sources. As several lines contributed to the bin this also includes any error in the calibration parameters.
The SD of the depths in the bin shown is 10cm in 13.5m water depth, and the distribution follows a normal curve. The standard error in a meaned bin is given by the standard deviation of the data divided by the square root of the number of data points in the bin, for a normal distribution.
The 4m bin shown below contained 1335 soundings, or more than 80 per 1m bin. In each 1m bin in this survey there were typically 20 or more depth soundings (up to several hundred). Using the SD measured above, 20 points in a bin gives a standard error of ±2.25cm.
A 95% confidence level is required by IHO SP44 specifications; this is the confidence level given by two standard errors. This gives a binned survey repeatability of ±5cm with a 95% confidence level.
The high data density in the GeoSwath enables survey depth quality to be confirmed throughout the survey area, without having to use artificial error budget estimations that rely on manufacturer’s data sheets.
Comparisons with Reson 8101 data
Note the along track roughness from the Reson (arrowed), probably from poor bottom detect in the outer beams. The numbered blocks are analysed in more detail.
Overall the survey agreement between the two data sets was excellent. The shaded relief images clearly show the same scour features and bottom shape, and the contours match well.
For detailed comparisons four blocks were extracted from the two surveys. Statistical analysis shows that the mean survey depths agree to within a few cm, individual bins agreed to within 30cm at the 95% confidence level (2SDs), and none of the 34,500 soundings were over 90cm apart.
Statistical comparisons of the 4 extracted blocks. Note that the mean is not zero because of offsets in the grid centres on slops.
Many of the differences are from slight position offsets in terrain with slopes, but some can be traced to the different modes of operation of the two systems.
The GeoSwath Bathymetry System obtains several thousand data points per ping. In contrast the Reson 8101 electronically forms 101 beams each with a nominal 1.5ºx1.5º 3dB beampattern. The Reson bottom finding algorithm usually flags several outer beams with poor quality factor, often giving a useable centre sector of less than 130º Many of the outer soundings will be well outside IHO special order specifications for both coverage and bathymetry.
Comparison of colour coded depths (above) and sun illuminated image (below) for block 1. The GeoSwath survey is on the right. The overall agreement between the two surveys is apparent and the better resolution of the GeoSwath can be seen.
Comparison of colour coded depths (above) and sun illuminated image (below) for block 2. The GeoSwath survey is on the right. While the scour marks (10cm - 40cm deep and 1m - 4m across) can just be seen in the Reson data, they are clearly defined by the GeoSwath.
Contour plots of the surveys (GeoSwath on the right) from the entrance to berth 2 (the northern berth in the GeoSwath survey). Block 3 is the top left quarter of this area. The contours are 20cm apart, and the roughness from the outer beams of the Reson is apparent. The locations of the two cross sections below are shown as solid and dashed lines.
The poor accuracy of the outer beams of the 8101 is apparent in the survey data. This gives the bands of roughness running northwest-southeast that are apparent in the sun illuminated and contour plots. In contrast the GeoSwath data shows no along track artefacts, showing that the depth accuracy is high across the whole swath.
The depth errors caused by the outer beams of the 8101 can be found by looking at the variations in depth in these along track artefacts. In Block 3 a rough band runs diagonally across the whole area. Cross sections were taken from the 8101 data and the GeoSwath data: these show that the Reson soundings can vary by 50cm from each other and from the GeoSwath data. A nearby cross section taken from the 8101 inner beams shows excellent agreement with the GeoSwath data set (mostly to within 10cm). A plot of the difference between the two surveys clearly shows a band where the Reson’s outer beams cause problems.
Survey cross sections from above: red is Reson, green is GeoSwath. The upper sections (dashed lines on the contour plots) shows good agreement. The lower section (solid lines) coincides with the Reson outer beams. This gives depths up to 50cm too shoal with lots of variation.
Plot of the difference between the two surveys in block 3. Note the band of poor agreement coincides with the rough banks seen in the Reson survey.
The GeoSwath high data density allows small features to be accurately measured, and the side scan data can be used to confirm features seen in the bathymetry. This is shown in the plots from block 4, where scour marks can be seen in the sun illuminated views, colour coded bathymetric plots, and side scan waterfall. The cross sections through the two surveys show the better feature definition of the GeoSwath. True side scan images are not available from the Reson.
Reson (left) and GeoSwath (right) data from block 4. Scour marks A B C D can be compared with the side scan and sun illuminated below.
Sun illuminated view of block 4 (GeoSwath on right), with cross sections through the scours
In 15m water depth the 8101 outer beam across track footprint will be over 1.5m even when limited to a 130º sector. Quay walls will have very few soundings, and metre scale objects will often only have one hit per line. This low data density means that even 2m bins will often only contain one sounding, so depth confidence levels can not be verified and object detection capabilities are poor.
The Reson’s low data density also means that one noisy sounding will have a large effect on the mean depth in a bin, and the 95% confidence level required by IHO specifications must be calculated from manufacturers data sheets.
Conclusions
In the Port of Marseilles the GeoSwath demonstrated its capability as a port survey tool, giving survey data well within IHO special order specifications and allowing object detection and identification.
Total survey depth accuracy at the 95% confidence level was ±5cm, with ±2cm static offset error. This depth accuracy was measured using the survey data.
The GeoSwath survey results were shown to be an improvement on the data collected by a Reson beamforming multibeam, with GeoSwath being easier to mobilise and providing better depth confidence and enhanced feature detection.
In this survey the GeoSwath also showed:
More details on GeoSwath |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
Introduction
The
survey system was installed on a 7.5m survey launch operated by the port authority.
The GeoSwath Sonar wet end was mounted on a pole mount over the side of the vessel.
The installation and all offset measurements for the sonar, positioning system and
heading reference unit were completed in three hours on the first survey day.
Part
of the Side Scan record of the above scours