|
|
|
|
|
|
|
© GeoAcoustics 2008
|
GeoSwath Plus Plymouth Data Set 2005 From the Common Data Set for Shallow
Survey 2005
In September 2005 the Shallow Survey Conference was held in Plymouth, UK. Prior to this conference sonar manufacturers were invited to collect a dataset of the bathymetry around Plymouth harbour. The survey areas covered were an offshore area south of Plymouth breakwater, and an inshore area around Plymouth Hoe. The GeoAcoustics GeoSwath Plus data collection for the Common Dataset is described in the first section of this report. The second section compares images from the GeoSwath Plus and from other sonar systems. The other systems’ data was used as supplied in the Shallow Survey 2005 Common Dataset disk.
The survey work was carried out during the week 20th June 2005 using the Sonardyne vessel “Marco”. The GeoAcoustics 250kHz GeoSwath system was deployed on a standard “V” plate, with a pair of transducers mounted at 30 degrees to vertical. Also fitted to the “V” plate were a Valeport mini SVS to record sound velocity at the transducers, and a Tritech echosounder which acts as an online depth QC. Sonardyne’s RTK GPS was used for positioning, whilst POS MV 320 gave attitude and heading. 1PPS synchronisation was used, synchronising the sonar system clock to UTC. A standard GeoAcoustics overside mount was deployed on the starboard side of the rear deck of the Marco. The vertical pole consisted of three screw together 75mm OD sections, with the V plate at the bottom and a cross piece at the top to mount the two RTK aerials. The POS MV IMU (attitude sensor) was rigidly attached to the transducer head in a pressure vessel fitted to the V plate. Mobilisation and calibration were carried out on the Monday, with data collection completed over the next 3 days. A total of 6 hours was spent in the Barn Pool area, 6 hours in the Hoe, and 10 hours outside the breakwater. De-mob was on the Friday morning. On the first full survey day (Tuesday) the outer section was covered, but the weather deteriorated and although not severe, created significant vessel roll and difficulty in holding line. The Barnpool section and most of the remaining outer section were completed Wednesday (fine and hot weather). The Hoe section and some fill in lines were finished on Thursday (also fine weather). Most of the survey was run at 5kts with swath widths set at 50m per side. Line spacing of 40 to 50m was used in the inshore sections, and a mix of alternate 90m and 40m spacing (side scan search pattern) in the outer sections. SVP dips were taken roughly every hour, generally at the end of lines. Tidal values were taken from the RN Hydrographic school (10 minute intervals) and the University of Plymouth (1 to 2 minute intervals, when available). All times are in UTC. Survey log notes are included in the Common Data Set disk in an Excel Spreadsheet. This also contains positions and times for SVP casts.
All data was processed using the GeoSwath Plus (GS+) acquisition and processing software. The GeoSwath Plus 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. In data processing the raw swath data was filtered to remove outliers, sound velocity corrections were applied and navigation data was checked and edited using the GeoSwath Plus software. The calibration parameters were applied to the swath data, and tide and SVP files were applied. This gave georeferenced, calibrated data files on a line-by-line basis. All this happens in an automated (but user accessible) way inside the GeoSwath Plus software system. Calibration was carried out using data as it was gathered day by day. Calibration parameters were found from the semi-automated calibration tools included in the GeoSwath plus software. Calibration results:
Data filtering consisted of setting filters in a wide window of +/- 2m around local mean depth along the swath, and processing swath by swath to reject outliers. The swath files were then gridded using mean binning.
GeoTexture software (also from GeoAcoustics Ltd.) was used to normalise the side-scan, allowing seamless mosaics to be made of the survey area, which could then be draped over the bathymetry to aid feature identification and to help classification of different bottom types.
Calibration offsets, tidal corrections, transducer face corrections and SVP corrections were all applied as the swath files were generated.
In the common dataset the GeoSwath 1m grids are supplied without interpolation or smoothing. The 50cm binned data is supplied in two formats: One unfiltered and one with 1m interpolation and single pass 2-D smoothing with centre weighting of 11.
The final bathymetric grids show good feature definition and line matching, and at 1m binning do not require interpolation, further cleaning or smoothing. The backscatter (side scan) mosaics can be processed either within GeoSwath Plus, or processed further within GeoTexture for textural analysis. They can then be overlaid onto bathymetry for 3-D views.
At 1m binning the data needed no further processing, with the high data density per bin resulting in very accurate and repeatable mean depths. This can be seen in the resolution of complex small features in the rocky region outside the breakwater. At 50cm binning the relatively reduced data density directly below the transducers is apparent in single swaths, but overlap is sufficient to give good coverage when gridding. The level of detail is very high with 0.5m binning, out to around 6 to 8 times water depth, where the effect of reduced signal to noise ratio is evident as noise at swath edges. It was apparent that in the shallow offshore section there are edge matching issues. It is thought that the most significant contribution to the artefacts comes from localised SVP variations which were not recorded at high enough density both in terms of position and time (even though SV dips were taken every hour). Though not large (a few cm at most), these artefacts are visible with sun illumination set at low angles. This is thought to be due to variability in the SVP during the course of each line – this variability is also seen in the recorded SV at the transducer face.
One of the objectives of the common dataset was to test the sonars’ ability to detect a 2m cube sonar target deployed in a deep part of the survey area (40m). This corresponds to the IHO SP44 ed4 Special Order and the LINZ Hyspec v3 criteria for object detection. A full analysis of the raw data from several of the sonars that detected the cube (including the GeoSwath) is included in the recent paper by Andrew Talbot of the UKHO, published in the Jan/Feb 2006 issue of Hydro International. The images below are from the GeoSwath Plus data in that area.
The images were created in GeoSwath Plus processing software using the processed data supplied by the sonar manufacturers in the Common Dataset disk. The sonars used to collect the Common Dataset were:
This set of images show the area surveyed in the Hoe. 50cm bin size was used.
The images below show a zoomed in 350m by 600m section of the 50cm grid of the Hoe.
These images show the extents of the outer survey area surveyed.
The three images on this page show more detail of rocks from the outer area, using 50cm bins.
The standard deviation of the difference grids between surveys in the two areas are shown in the table below. The Reson data was chosen as the reference because of the relatively flat SV seen on the day of the Reson survey. The table illustrates the consistency of the depths found by the GeoSwath, Reson and Kongsberg surveys.
The GeoSwath Plus was used to survey the common data set area in a total of 26 hours survey time over 3 days. The data products were 0.5m and 1m grids of depth, along with georeferenced and normalised side-scan mosaics. The images and charts produced demonstrate the ability of the GeoSwath to resolve small seabed features in complex harbour environments, and show the quality and consistency of the data collected. The GeoSwath Plus data set is self consistent and also appears to be consistent with previous surveys of this area and surveys using other sonars. The processing methodology described in the text appears to be successful for this data set, validating previous published work on this topic. The normalised backscatter data shows some interesting textural variations in the outer sections. Various objects and features of interest were picked up on both bathymetry and side scan. Comparison of the results of the GeoSwath survey with those of other manufacturers, both in this report and in other studies confirms the capabilities of the GeoSwath Plus sonar and the quality of the data produced in this kind of environment. The assistance of Sonardyne Ltd, and in particular the vessel skipper, should be mentioned, as this survey could not have been completed without their input. Many thanks are also due to the University of Plymouth and the RN Hydrographic School for making their tide information available, and to the Shallow Survey Team at MCA and UKHO for their help and patience. For further comparisons between the data collected by the different sonar systems see “Shallow Survey 2005 Common Dataset Comparisons”, by Andy Talbot, UK Hydrographic Office, in the January/February 2005 issue of Hydro International. More details on GeoSwath
|
