|Geographic Data Set:||
Husdon River Estuary E-W, N-S Side Scan Imagery
Sidescan imagery is an acoustic analog to aerial photography. We use sound that reflects off the estuary floor to create an image of the estuary floor. The strength of the acoustic echo is governed by the composition of the estuary floor and the local relief of the estuary floor. We use sidescan imaging as a remote sensing tool to map features such as sediment waves on the estuary floor and to map variations in sediment type from hard rock to sand to fine silt and clay. As with aerial photography we must ground-truth sidescan images with sediment samples (cores and grabs) and close-up photographs (Sediment Profile Imagery) as part of the process of interpreting sidescan images.
In these images darker areas are more reflective than lighter areas. The more reflective areas are harder than the less reflective areas and are more likely to be dominated by rock or sand as opposed to fine sillt or clay found in lighter areas.
Hudson River Estuary
|Completion or Most Recent Revision Date:||
|Type of Data:||
SUN raster files.
The location of the ship during data acquisition was determined using real time differential GPS and after corrections for location and orientation of the sidescan sensors with respect to the ship pixel locations on the images are located to better than a meter.
|Projection and Map Units:||
UTM Zone 18 = NYTM in meters, NAD83 horizontal datum.
|Data Acquisition & Processing:||
The towed side scan system used for this program was the Edge Tech DF-1000 dual frequency sonar with and ISIS data acquisition topside. The system is designed to acquire data at two frequencies 100Khz and 384 kHz. The 384 kHz is often referred to as the 500 kHz for historical reasons. The six foot tow fish was deployed from a boom off the bow of the ship to place the system in quiet water for optimal instrument performance. The fish was towed at a depth of 6 feet. The fish has transducers and receivers on either side of the tow fish. The transducers transmit and receive both frequencies simultaneously. The acoustic signals are digitized in the tow fish and transmitted with a high-speed digital uplink to the onboard acquisition system. The tow cable is a single coaxial cable. A swath width of 200m was used so that together a total width of 400 m of riverbed was surveyed with a single survey track. The topside unit was an ISIS data acquisition supported by Triton Elic. The side scan data was time tagged in the ISIS system and recorded to hard disk. The Triton Elic system also recorded several auxiliary data streams including: the ship's compass heading, the single beam bathymetry and the real time navigation. The LDEO ship compass was mounted in magnetically quiet location amidships. The depth sounder used was the R/V Walford 's Raytheon DE-719C with a hull-mounted transducer. The transducer (Raytheon model 200TSHAD) operates at 208 kHz, with 8deg; beam width at half power points. The DE-719C system produces ananalog output and was interfaced with an "Odom Digitrace" system. The transducer was mounted amidships and a bar check was performed daily to monitor for any system offsets. The navigation data recorded in the ISIS system was the satellite corrected GPS positions from the Ashtech Z-12. The Ashtech Z-12 is a 12-channel dual-frequency geodetic caliber GPS receiver. The real time corrections were provided by Omnistar and received by a Trimble AG-132 unit. The Trimble unit was selected to enable the flexibility of using either the satellite broadcast corrections or the real time correction transmitted by the US Coast Guard. During operations only the satellite broadcast corrections were used to prevent the introduction of offsets between the two corrections. Duplicative JAZ disks were used to produce daily archives of the side scan data and the associated auxiliary data. The data were generally recorded into files which contained only the along track data. In areas 1 and 4 the data acquired during turns was saved into separate files while in area 2 and 3 the turns are often included in the main track files. The data acquired in Areas 2 and 3 was recorded with less resolution per pixel than the data acquired in the spring of 1999 in Areas 1 and 4. An improved version of the acquisition software from Triton Elic enabled the recovery of the full resolution.
(2) Survey Design
The survey design was targeted at full insonification of the river at 85m line-spacing in the north-south orientation and 180m line-spacing in the east-west orientation. The naming convention for the side scan files for the field files includes an area identifier, the direction of the track (i.e. east-west or north-south), and an indicator for the number of times the line was shot. During turns, data was recorded into a file using the naming convention of the just completed line with an A1 suffix. For example, A1N001b is the filename for the Area 1 file acquired in the north-south orientation along the track designated #1 for the second time as indicated by the b. Also, A1N001A1 is the filename for the Area 1 file acquired during the turn after A1N001A was complete. The numbering convention was designed so that low number lines are located in the south and west of the areas while high number lines are at the northern and eastern edges of the areas. A number of the East-West lines could not be collected due to the narrowness of the river specifically in area 3 in the region west of Kingston - Saugerties and close to the Hogsback. In two areas, Storm King and all of Area 4 the currents forced us to alter the orthogonal nature of the survey grid. In the Storm King region the north south lines run in a river parallel sense (SE-NW) and are labeled as A2SA### instead of A2N##3. The SA sequence is also the only set where the numbering convention begins with the low number in the east. For example SA001 runs along the eastern shore and SA005 images the western shore at the base of Storm King Mountain. We were not able to acquire the east-west lines in this area which have a normal orientation and normal naming convention. In Area 4, the narrowness of the navigable channel prevented any true east-west lines from being acquired. In this case we collected a suite of 128 lines oriented in NE-SW and NW-SE orientations. These are all labeled A4W###a.
(4) Navigation Merge
Following the field program the side scan data was processed and the track data merged to produce mosaics of the four areas. The principal steps from the field tapes to the final mosaics were to demux the data, merge with final navigation and produce the final mosaics. Each of these steps is described in detail.
(1) Demux the field tapes.
(2) Merge with the final navigation and incorporate layback correction.
GPS navigation data were acquired from a pair of Ashtech Z-12 (dual-frequency 12-channel Y-Code) geodetic-quality GPS receivers, operated continuously at 2 Hz. One receiver was located on the ship and one was located at a fixed basestation at LDEO. The GPS data are post-processed with the KARS (Kinematic and Rapid Static) software package developed by G.Mader at the National Geodetic Survey (NGS). We solve for the relative position of the ship at each epoch with respect to the fixed basestation, which has been precisely located from a static solution with nearby stations in the CORS (Continuously Operated Reference Stations) network, and precise satellite orbit data obtained from the Crustal Dynamics Data Information System (CDDIS) archive. Differential processing allows us to remove clock errors and orbit errors which are common to both receivers. The tropospheric delay is modeled, and the ionospheric error is reduced by using both the L1 and L2 frequency observables. KARS uses an optimized integer-search algorithm to resolve initial satellite ambiguities and detect cycle slips. Final pseudorange corrected positions are accurate to 0.5 meters. Once the final navigation is obtained layback corrections are calculated for the side-scan sonar vehicle. The layback is the offset between the GPS antenna, the position provided by the GPS positions, and the location of the side scan fish as it is towed off the bow of the boat. Fore-aft distances from the fish to the GPS antennae vary as a function of wire angle that depends on ship speed. During data acquisition, speed through the water was logged to determine tow angle from which offset between the transducers and antennae is calculated (ping by ping) to a much greater accuracy than is achieved with the common assumption of a fixed tow point.
(3) Merge track data to produce final mosaics. We have chosen this philosophy to provide minimally processed backscatter data in the final mosaics. This ensured the preservation of the maximum information. We have implemented a slant range correction for the data. All the mosaicking routines assume a flat bottom. The major corrections incorporated into the mosaicking routine are: to remove the nadir, to systematically seam the data and to equalize the backscatter values over each area. The nadir central stripe produces a final product which is difficult to interpret. Our survey design ensured that the nadir from one line could be filled with the outer beam data from the adjacent line. In the mosaic we have filled the nadir with the adjacent line data wherever possible. The nadir infill data is thus lower amplitude than the adjacent data. For each area we have equalized the backscatter values. The net result is that a backscatter value cannot be robustly compared between regions at this point. We evaluate in detail the utility of correcting the side scan sonar for the heading of the fish, which can differ from the heading of the boat. We concluded that an empirically derived correction of 4deg; reduced the swath to swath mismatch of riverbead features and improved the final mosaic. In generating the mosaics for each area we used a suite of large (5') and small mosaics (~2 km on a side) for each area. Each area was subdivided into 5' or 5 nautical mile tiles. One 5 tile was generated for Area 1 and 2 for each of the other areas. These 5' mosaics were produced at 2m pixel resolution. For detailed imagery, when the users zoom using the Arc View tool, we also generated a series of tile approximately 2 kilometers on a side. The bounds for each mosaic and a schematic of the locations are found below. For each tile 4 mosaics were produced using different azimuth data and different frequency data. The four mosaics are a north-south 100-kHz mosaics, an east-west 100-kHz mosaic, a north-south 384-kHz mosaic and an east-west 384-kHz mosaic.
(3) Quality Control
After each day of acquisition the data was returned to LDEO for overnight quality control. For the side scan data the goal was to ensure the proper number of bits were present in the data, that the navigation and time tagging was appropriate and that the preliminary imagery was correct. The goal was to identify any corrupt track while it was still possible to reshoot the data. The side scan quality control for each track included a page of text containing the critical acquisition parameters and a track plot of both the 100 and 384 kHz data. The track plot included an overlay of the ships heading and depth from the single beam. In addition to these track based quality control products, a daily mosaic was generated for each day's data, identifying missing data, poor quality data and badly steered lines. These products were reviewed by a project scientist early the next day. Summaries of the quality control results were transmitted to the field crew to enable acquisition of repeat data as necessary. Data quality was in general excellent. Field conditions which introduced noise into the side scan data included stratification of the water column in shallow water during calm, hot days, excessive ship's motion, turbulent water due to boat wakes, and sonar fish vibration due to build up of vegetation on the towing cable.
John W. Ladd
Benthic Habitat Coordinator
NYS Dept of Environmental Conservation
43 Hudson Watch Drive
Ossining, NY 10562