United States Department of Commerce
National Oceanic and Atmospheric Administration
Coastal Services Center
2234 Hobson Avenue, Charleston, SC 29405-2413
Coastal Services Center Technical Report CSC/9-98/001 Version 1.0 September 1998
AN ASSESSMENT OF NASA’S AIRBORNE
TOPOGRAPHIC MAPPER INSTRUMENT FOR
BEACH TOPOGRAPHIC MAPPING
AT DUCK, NORTH CAROLINA
AN ASSESSMENT OF NASA’S AIRBORNE
TOPOGRAPHIC MAPPER INSTRUMENT FOR
BEACH TOPOGRAPHIC MAPPING
AT DUCK, NORTH CAROLINA
Andrew W. Meredith1 John C. Brock3,4 Asbury H. Sallenger, Jr. 4 William B Krabill5 Robert N. Swift6 Michael G. Hearne2 Jeff List9 Rob A. Holman8 Mark Hansen4 Thomas Reiss10 Karen L. M. Morgan4 C. W. Wright5 Earl B. Frederick6 Serdar S. Manizade6 Jim K. Yungel6
C. F. Martin7 John G. Sonntag7
1REMSA, Inc. 6EG&G Washington Analytical Services 2234 South Hobson Avenue Center, Inc. Charleston, SC 29405 Wallops Fight Facility
Wallops Island, VA 23337
2TPMC 7EG&G Washington Analytical Services 2234 South Hobson Avenue Center, Inc. Charleston, SC 29405 900 Clopper Road
Gaithersburg, MD 20878
3NOAA Coastal Services Center 8College of Oceanic and Atmospheric Sciences 2234 South Hobson Avenue Oregon State University Charleston, SC 29405 104 Ocean Administration Building
Corvallis, OR 97331
4USGS Center for Coastal Geology 9USGS Woods Hole Field Center 600 Fourth Street South 384 Woods Hole Rd. St. Petersburg, Florida 33701 Woods Hole, MA 02543
5NASA 10USGS Laboratory for Hydrospheric Processes 345 Middlefield Road Wallops Flight Facility Menlo Park, CA 94025 Wallops Island, Virginia 23337
TABLE OF CONTENTS
1.2 Airborne LIDAR Assessment of Coastal Erosion Project ---------------------------------------1
4.0 RESULTS AND DISCUSSION ------------------------------------------------------------------16
4.1.1 ATM to ATM Over a Stable Geomorphological Feature -------------------------------16
4.1.2 ATM to ATM Same Day Comparisons-----------------------------------------------------20
4.1.3 ATM to ATM Multiple Day Comparisons -------------------------------------------------37
4.2 ATM to Ground Survey Comparisons -------------------------------------------------------------43
4.2.1 ATM to 70-Kilometer List Buggy Survey -------------------------------------------------43
4.2.2 ATM to Holman Buggy Survey --------------------------------------------------------------49
4.2.3 ATM to Reiss Total Station Survey ---------------------------------------------------------59
7.0 APPENDIX: DETAILED COMPARISON RESULTS -------------------------------------69
LIST OF FIGURES
Figure 1. Map of ground survey locations --------------------------------------------------------------------3 Figure 2. Detailed maps of individual ground survey locations ------------------------------------------4 Figure 3. Airborne Topographic Mapper ---------------------------------------------------------------------5 Figure 4. Map of the September 26, 1997 ATM surveys --------------------------------------------------9 Figure 5. Map of the September 27, 1997 ATM surveys ------------------------------------------------10 Figure 6. Difference plot from intra-comparison of pass 135929 on September 26----------------13 Figure 7. Difference plot from intra-comparison of pass 182303 on September 27----------------14 Figure 8. Graph of relationship between the horizontal search radius and standard deviation ---15 Figure 9. Difference plot from comparison of September 26 and 27 ATM surveys over the
Wright Memorial-----------------------------------------------------------------------------------17 Figure 10. Histograms from comparison of September 26 and 27 ATM surveys over the
Wright Memorial-----------------------------------------------------------------------------------18 Figure 11. Graph of the statistics from individual ATM pass comparisons--------------------------19 Figure 12. Difference plot from comparison of all ATM passes to the 135929 pass for
September 26 along the 70-kilometer List buggy track ------------------------------------21 Figure 13. Histograms from the comparison of all ATM passes to the 135929 pass for
September 26 along the 70-kilometer List buggy track ------------------------------------22 Figure 14. Difference plot from the comparison of all ATM passes to 142303 pass for
September 27 along the 70-kilometer List buggy track ------------------------------------23 Figure 15. Histograms from the comparison of all ATM passes to the 142303 pass for
September 27 along the 70-kilometer List buggy track ------------------------------------24 Figure 16. Difference plot from the comparison of all ATM passes to the 135929 pass for
September 26 over the beach --------------------------------------------------------------------25 Figure 17. Histograms from the comparison of all ATM passes to the 135929 pass for
September 26 over the beach --------------------------------------------------------------------26 Figure 18. Difference plot from the comparison of all ATM passes to the 142303 pass for
September 27 over the beach --------------------------------------------------------------------27 Figure 19. Histograms from the comparison of all ATM passes to the 142303 pass for
September 27 over the beach --------------------------------------------------------------------28 Figure 20. Difference plot from the comparison of the 130739 ATM pass to the 182303 ATM
pass for September 27 over the beach----------------------------------------------------------29 Figure 21. Difference plot from the comparison of the 144126 ATM pass to the 182303 ATM
pass for September 27 over the beach----------------------------------------------------------30 Figure 22. Difference plot from the comparison of the 145204 ATM pass to the 182303 ATM
pass for September 27 over the beach----------------------------------------------------------31 Figure 23. Difference plot from the comparison of the 180419 ATM pass to the 182303 ATM
pass for September 27 over the beach----------------------------------------------------------32 Figure 24. Difference plot from the comparison of the 184941 ATM pass to the 182303 ATM
pass for September 27 over the beach----------------------------------------------------------33 Figure 25. Graph of the statistics from the individual ATM passes for September 26 compared
to ATM pass 135929 ------------------------------------------------------------------------------35 Figure 26. Graph of the statistics from the individual ATM passes for September 27 compared
to ATM pass 182303 ------------------------------------------------------------------------------36 Figure 27. Difference plot from the comparison of all ATM passes for September 26 to all
passes for September 27 along the 70-kilometer List buggy track -----------------------38 Figure 28. Histograms from the comparison of all ATM passes for September 26 to all passes
for September 27 along the 70-kilometer List buggy track --------------------------------39 Figure 29. Difference plot from the comparison of all ATM passes for September 26 to all
passes for September 27 over the beach -------------------------------------------------------40 Figure 30. Histograms from the comparison of all ATM passes for September 26 to all passes
for September 27 over the beach----------------------------------------------------------------41 Figure 31. Graph of the statistics from the individual September 27 ATM passes compared to
all September 26 ATM passes-------------------------------------------------------------------42 Figure 32. Difference plot from the comparison of all ATM passes for September 26 to List
buggy survey----------------------------------------------------------------------------------------44 Figure 33. Histograms from the comparison of all ATM passes for September 26 to List buggy
survey-------------------------------------------------------------------------------------------------45 Figure 34. Difference plot from the comparison of all ATM passes for September 27 to List
buggy survey for September 26 -----------------------------------------------------------------46 Figure 35. Histograms from the comparison of all ATM passes for September 27 to List buggy
survey for September 26--------------------------------------------------------------------------47 Figure 36. Graph of the statistics from the List buggy survey compared to individual ATM
passes -------------------------------------------------------------------------------------------------48 Figure 37. Difference plot from the comparison of all ATM passes for September 26 to Holman
buggy survey for September 26 constrained to the List buggy track.--------------------50 Figure 38. Histograms from the comparison of all ATM passes for September 26 to Holman
buggy survey for September 26 constrained to the List buggy track --------------------51 Figure 39. Difference plot from the comparison of all ATM passes for September 26 to
Holman buggy survey for September 26 over the beach -----------------------------------52 Figure 40. Histograms from the comparison of all ATM passes for September 26 to Holman
buggy survey for September 26 over the beach ----------------------------------------------53 Figure 41. Difference plot from the comparison of the 131058 ATM pass to the Holman
buggy survey for September 26 over the beach ----------------------------------------------54 Figure 42. Difference plot from the comparison of the 132917 ATM pass to the Holman
buggy survey for September 26 over the beach ----------------------------------------------55 Figure 43. Difference plot from the comparison of the 135929 ATM pass to the Holman
buggy survey for September 26 over the beach ----------------------------------------------56 Figure 44. Difference plot from the comparison of the 143020 ATM pass to the Holman
buggy survey for September 26 over the beach ---------------------------------------------57 Figure 45. Graph of the statistics from the September 26 Holman buggy data compared to
individual ATM passes for September 26-----------------------------------------------------58 Figure 46. Graph of the statistics from the September 27 Reiss survey compared to individual
ATM passes for September 27 ------------------------------------------------------------------59 Figure 47. Difference plot from the comparison of all ATM passes for September 27 to
Hansen beach surveys for September 27 ------------------------------------------------------61 Figure 48. Histograms from the comparison of all ATM passes for September 27 to Hansen
beach surveys for September 27 ----------------------------------------------------------------62 Figure 49. Graph of the statistics from the September 27 Hansen surveys compared to
individual ATM passes for September 27-----------------------------------------------------63 Figure 50. Statistics generated after a partitioning by latitude of the results from the
September 26 135929 ATM pass minus List buggy survey comparison ---------------66 Figure 51. Statistics generated after a partitioning by latitude of the results from the
September 27 184941 ATM pass minus List buggy survey comparison ---------------66 Figure 52. Graph of the statistics from the individual ATM passes for September 26 compared
to ATM pass 135929 constrained to the List buggy track ---------------------------------67 Figure 53. Graph of the statistics from the List buggy data compared to individual ATM
passes for September 26 --------------------------------------------------------------------------67
LIST OF TABLES
T able 1. Raw elevation data record format (“qfit” format) ----------------------------------------------11 Table 2. Intra-file comparison results------------------------------------------------------------------------12 Table 3. Summary of ATM to ATM comparisons for the Wright Memorial ------------------------16 Table 4. Summary of ATM to ATM comparisons for September 26 and 27, 1997 ----------------20 Table 5. Summary of between-day ATM comparisons for September 26 and 27, 1997---------37 Table 6. Summary of ATM to List buggy survey comparisons ----------------------------------------43 Table 7. Summary of ATM to Holman buggy survey comparisons -----------------------------------49 Table 8. Summary of ATM to Reiss total station survey comparisons -------------------------------59 Table 9. Summary of ATM to Hansen survey comparisons --------------------------------------------60 Table 10. Summary of non-ATM survey comparisons---------------------------------------------------63 Table 11. Summary of statistics from ATM to ATM survey comparisons---------------------------64 Table 12. Summary of statistics from ATM to ground survey comparisons-------------------------64
NOAA Coastal Services Center Technical Report CSC/9-98/001
AN ASSESSMENT OF NASA’S AIRBORNE TOPOGRAPHIC MAPPER
INSTRUMENT FOR BEACH TOPOGRAPHIC MAPPING AT
DUCK, NORTH CAROLINA
1.0 INTRODUCTION
This introductory section provides some background about the need and potential benefits from successfully applying LIDAR beach topographic mapping and presents a brief overview of the Airborne LIDAR Assessment of Coastal Erosion (ALACE) Project.
1.1 Background
Beaches are some of the earth’s most dynamic geologic features. Beach morphology fluctuates over a wide range of time scales, varying from periods of hours associated with diurnal tides and storm events, to years and decades associated with long-term erosional trends. Human actions, especially during the last 100 years, have created a situation in which beach erosion can have severe economic consequences. Currently 55 to 60 percent
of the U.S. population lives within the nation’s 772 coastal counties, with projections of 75 percent by 2025 (Hinrichsen 1998). Estimates reveal that approximately $3 trillion worth of U.S. coastal development is potentially vulnerable to erosion. It is also estimated that 70 percent of the world's beaches are undergoing erosion, with percentages approaching 90 percent along the Atlantic coastal plain (Bird 1985).
Accurate and timely assessment of erosion conditions and storm impacts is needed to assist decision making on land use, beach renourishment, erosion calculations, insurance compensation, and property value estimation. Proper storm damage assessment is an enormous task for emergency and disaster response agencies and personnel. Federal, state, and local agencies have traditionally used aerial photographs and land surveys to assess the overall impact of storms. Although these measurement methods provide valuable information, they are often not precise enough to describe specific coastal topographic changes that enable implementation of fully effective shoreline emergency response, or development planning along with beach renourishment programs.
1.2 Airborne LIDAR Assessment of Coastal Erosion Project
The ALACE project is a partnership between the National Oceanic and Atmospheric Administration (NOAA) Coastal Services Center (CSC) in Charleston, South Carolina; NASA Goddard Space Flight Center, Wallops Flight Facility (WFF) in Wallops, Virginia; and U.S. Geological Survey (USGS) Center for Coastal Geology in St. Petersburg, Florida. The project’s goal is to establish the capability of aircraft laser swath mapping to provide highly accurate, cost-effective information on coastal topography, erosion, and shoreline position. In working toward this goal, NOAA, NASA, and USGS have conducted several mapping missions along significant portions of U.S. coast using the NASA Airborne Topographic Mapper (ATM) flown aboard a NOAA Twin Otter aircraft.
During the fall of 1997, dense beach topographic data were collected with the NASA ATM mounted aboard the NOAA Twin Otter for a contiguous region of the east coast from Cape Henlopen, Delaware to the South Carolina/Georgia border and three sections along the west coast: Point Grenville, Washington to Cape Blanco, Oregon; Bodega Head to Big Sur, California; and Cayucos, California to the U.S./Mexico border. Coincident with the NOAA Twin Otter ATM overflights, along a 70-kilometer section of the Outer Banks in North Carolina extensive beach ground survey data were collected as a part of the SandyDuck 1997 coastal field experiment sponsored by USGS and U.S. Army Corps of Engineers (USACE). Data were collected from Corolla, North Carolina to Oregon Inlet, North Carolina using differential global positioning system (GPS), Total Station, and other survey techniques.
2.0 OBJECTIVE
The goal of this report is to validate the potential of airborne LIDAR as a technology for highly accurate and detailed topographic beach surveys. The primary objective is to verify both the accuracy of airborne LIDAR in relation to current ground survey data collection techniques and its ability to produce consistent repeat measurements over the same location. Krabill, et al. (in prep.), Krabill, et al. (1995), and Carter and Shrestha (1997) all have presented papers assessing the accuracy of LIDAR surveying techniques, but the evaluations were either not over beaches or they lacked intensive ground survey data to fully evaluate the LIDAR surveys. This report evaluates intensive ATM beach surveys in comparison to multiple ground surveys collected coincident with the ATM surveys and using a variety of accepted data collection methods.
3.0 METHODS
The following section describes the methods and instruments used for the collection of ground surveys and ATM data along with maps depicting the locations of each survey. The four investigators responsible for acquiring ground surveys included Rob Holman from Oregon State University (OSU), Jeff List of the USGS Woods Hole Field Center, Mark Hansen of the USGS Center for Coastal Geology, and Thomas Reiss of the USGS Western Region Coastal and Marine Geology Center.
3.1 Ground Survey Data Collection
Rob Holman’s group used their Trimble 4000 GPS survey system mounted on a beach buggy for the acquisition of a detailed grid of elevations about 3 kilometers north and south of the Duck Field Research Facility (FRF) pier (Figures 1 and 2a). The September 26, 1997 survey area extended from the dune line to waterline. Real-time differential corrections were obtained from a base station set up over an established benchmark on the FRF pier.
Jeff List used an all-terrain beach buggy outfitted with an Ashtech GPS receiver to obtain differentially corrected longshore elevation data for a 70-kilometer stretch of coast from Corolla, North Carolina to Oregon Inlet, North Carolina (Figure 1). The longshore pass was made northbound over the high beach on September 26, 1997. Four Ashtech base stations were established at locations along the beach to ensure the buggy was no more than 10 kilometers from a base station. The List raw GPS data were differentially corrected using the same processing methods as applied in the correction of the ATM data.
Figure 1. Map showing the location of ground surveys between Corolla, North Carolina and Oregon Inlet, North Carolina made September 26 and 27, 1997. Refer to Figure 2 for additional detail of the Holman, Hansen, and Reiss surveys.
(a)
(b) (c)
Figure 2. (a) Map showing the extent of the Holman buggy survey centered on the FRF pier. (b) Map showing the extent of the Hansen ground surveys around Corolla. (c) Map showing the extent of the Reiss ground surveys around Corolla.
Mark Hansen coordinated the collection of differentially corrected GPS beach and dune data using an Ashtech GPS receiver and an antenna mounted on a wheeled rod around Corolla (Figures 1 and 2b). Data were collected along several hundred meters of beach from the dunes to the waterline on September 27, 1997. An Ashtech base station was set up within several kilometers of the survey area.
Thomas Reiss of USGS collected beach and dune data around Corolla using a Total Station on September 27 and 29, 1997 (Figures 1 and 2c). The same GPS base station set up for the Hansen survey was used in the Reiss surveys.
3.2 LIDAR Data Collection
This section discusses details of the ATM instrument, the methods applied in ATM data collection and processing, and it presents details of the 1997 LIDAR beach surveys analyzed within this report.
3.2.1 Instrumentation The ATM is continually evolving as a result of technological improvements that
primarily result in reductions in the instrument size. Version 2 of the ATM (ATM-II) (Figure 3) was used during the fall 1997 beach surveys.
Figure 3. Airborne Topographic Mapper
The ATM is currently operated with a Spectra Physics TFR laser transmitter that provides a 7-nanosecond wide, 250-micro-joule pulse at a frequency-doubled wavelength of 523 nanometers (nm) in the blue-green spectral region. The laser transmitter can operate at pulse rates from 2 to 10 kilohertz (kHz). It was operated at 5 kHz during fall
1997 beach mapping surveys. The transmitted laser pulse is reflected to the earth’s surface using a small folding mirror mounted on the back of the secondary mirror of a 9-centimeter diameter Newtonian reflector telescope that views the laser footprint on the earth’s surface. The co-axial LIDAR transmit and receive path facilitates changing altitude above the topographic target without the need to realign the transmitter and receiver optics. The transmitted laser pulse and receiver field-of-view (FOV) are directed earthward by a nutating scan mirror assembly that is mounted directly in front of the telescope. The scan mirror, which is rotated as 20 hertz (Hz), is made from a section of 15-centimeter diameter round aluminum stock, machined to a specific off-nadir angle. A scan mirror with an off-nadir angle of 15 degrees was used in the ALACE beach mapping surveys, producing an elliptical scan pattern with a swath width equal to approximately 50 percent of the approximately 700-meter aircraft altitude. The ATM-II receiver is composed of the Newtonian reflector telescope, a single photomultiplier tube (PMT), and various other low cost, off-the-shelf optical components. The 2.1-milliradian FOV of the system is established by the thickness of a fiber optic cable situated at the focal plane of the telescope. The fiber transmits the reflected laser pulse to the photomultiplier assembly, which consists of a lens, a narrow band filter, and the PMT (Krabill et al., in prep.).
For the fall 1997 mapping missions, a passive channel sensor was added to the ATM. This sensor collects geo-referenced panchromatic (excluding 523 nm) data along the same elliptical scan path as the active laser. Images created from the passive channel data help identify ground features, and are used to assist in the delineation of the beach region.
The major components of the data acquisition system are a 133-megahetz (MHz) Pentium PC and a Computer Automated Measurement and Control (CAMAC) crate, which houses the time-interval counter, receiver power supply, pulse digitizer, inertial navigation interface, and pulse amplifiers. Output from an onboard Ashtech GPS receiver is collected by a separate PC (Krabill et al., in prep.).
3.2.2 Calibration Two types of calibrations are necessary for the topographic mapping system. The first is
necessary to develop a correction for the laser range determination. The ATM uses a leading edge discriminator in timing the laser range measurement. It must be calibrated
for a systematic error in range, which consists of a fixed part, or “zero-set,” and a part related to the amplitude of the received laser pulse, or “range-walk.” During pre-mission and post-mission calibrations, the outgoing laser beam is reflected horizontally via a folding mirror to a flat target board. Range measurements are then recorded while modulating the strength of the laser beam exiting the aircraft, which effectively produces a wide range of amplitude in the received laser signal. The distance between the scan mirror and the horizontal target board is measured both with a steel tape and independently with an electronic range finder. A correction table used in post-flight processing is developed from this ground calibration.
The second type of calibration is designed to determine the angular mounting biases of the ATM sensor relative to the inertial navigation system (INS) from which the aircraft attitude (roll, pitch, and heading) are determined. INS pitch and roll uncertainties are generally the limiting factors in ATM survey accuracy and are thus a primary source of concern. The roll and pitch orientation of the ATM scanner platform relative to the INS reference system must be determined to somewhat better than 0.1 degrees because, for an aircraft altitude of 700 meters and an off-nadir angle of 15 degrees, a 0.1-degree mounting error would introduce a height error of 32 centimeters and a horizontal displacement error of 131 centimeters. Because the ATM is a conical scanning sensor, the relative orientation between the ATM platform and the INS reference can be determined by flying over either a flat surface such as a water body or a known reference, and comparing the observed ranges with those computed on the basis of the determined position of the aircraft GPS antenna, the measured position of the scanner mirror relative to the GPS antenna in the aircraft (INS) coordinate system, the INS attitude measurements, and a model of the scanner measurement system. A large aircraft parking apron at WFF, which has been densely surveyed, served as the reference surface after installation of the ATM within the NOAA Twin Otter aircraft. It may be noted that these mounting biases can include small day-to-day variations in INS pitch, roll, and heading zero-set. Nonetheless, the ATM mounting biases are generally stable enough during a particular aircraft installation for a single set of numbers to be utilized for an entire campaign (Krabill et al., in prep.).
3.2.3 Navigation The ability to precisely follow specific flight lines is an important facet in beach mapping, both to ensure that data are collected over the desired site, as well as to ensure repeated coverage for change detection. Aircraft INSs are not sufficiently accurate to ensure that flights are precisely navigated along prescribed routes because of drift in their position estimates determined through accelerometers. Consequently, a navigation system based upon real-time GPS information was developed by the ATM group (Wright and Swift 1996). Associated software utilizes coarse acquisition (C/A) code positional output from the on-board GPS receiver that can supply data to an autopilot and provide the pilots with a real-time visual display of the flight line and current offset from the desired track. This system enables the pilot to maintain the aircraft within 30 to 50
meters of the desired flight track during missions lasting several hours and covering 100 to 200 kilometers of beach.
3.2.4 Aircraft Trajectory Determination In order to measure topography to the desired accuracy of less than 10 centimeters, the vertical and horizontal location of the GPS antenna mounted on the aircraft must be known to approximately 5 centimeters. This goal was achieved using kinematic GPS techniques (Krabill and Martin, 1987) that use the difference in the GPS dual frequency carrier-phase-derived ranges from the mobile receiver in the aircraft and from a fixed receiver located over a precisely known benchmark. Throughout the flight, the bank angle of the aircraft is limited to less that 10 degrees to avoid loss of carrier phase lock on the airborne GPS receiver. GPS data sets were obtained with the aircraft parked close to the fixed receiver for about 45 minutes before and after each survey flight. These stationary data sets are used to resolve ambiguities in carrier phase for each frequency between the fixed and mobile receivers for subsequent application during the processing of the in-flight data. Additionally, the local meteorological conditions (pressure, temperature, and humidity) were recorded for subsequent application during post-mission processing. These data are combined with a precise C/A code of the GPS constellation into a point-to-point range difference solution for the trajectory of the aircraft. Because of the relatively low noise in the phase data, no filtering or smoothing is required. The use of a precise post facto ephemeris is required for operations in which the baseline
between the aircraft and the fixed receiver exceeds 30 to 40 kilometers, and is recommended for all operations. These are available from several sources on the Internet within 2 to 10 days. The ATM surveys discussed in this report set up a base station using an Ashtech GPS receiver over an established survey mark located at the airport in Manteo, North Carolina, about 50 kilometers from the farthest end of the study area.
3.2.5 ATM Beach Surveys To ensure complete beach coverage over a section of beach, a typical ATM beach survey consists of two passes, a landward and a waterside pass. The flight lines are designed to produce an approximate 30 percent swath overlap between the two passes. The waterside
pass is conducted within an hour or so of low tide. Wider beaches may require additional passes to ensure complete survey coverage.
For purposes of this assessment, very dense coverage was obtained from 22 ATM passes acquired on September 26 and 27, 1997. Nine of the passes ran normal to the shoreline in an approximate east-west direction; the other passes ran shore-parallel, favoring either the landward or ocean side. The shore-parallel passes ranged in length from 70 kilometers, covering the entire study area, to less than 20 kilometers. On September 26, between 9:00 a.m. and noon, 10 passes resulted in over 16.5 million measurements for the study area from Corolla to Oregon Inlet (Figure 4). Over 25 million measurements were collected in 12 passes between 9:00 a.m. and 3:00 p.m. on September 27 for the same study area (Figure 5).
3.2.6 Generation of Raw Elevation File At the completion of the data collection, computer programs developed at the NASA WFF are used for mission post-processing to transform the data into binary files containing International Terrestrial Reference Frame 1994 (ITRF94) / World Geodetic
System 1984 (WGS84) geo-reference elevation measurements and additional ancillary data recorded at the time of the measurement. Post-processing involves the application of
(1) calibration corrections using previously collected pre- and post-mission calibration data; (2) differential corrections to each geo-referenced elevation measurement using both the aircraft GPS and base station GPS values; (3) INS measurements to correct for pitch, roll, and heading; and (4) aircraft mounting-bias parameters. The resulting file is composed of a variable length header followed immediately by the fixed length data records. Header information includes the number of words per data record and the header length in bytes. The format of the data portion of the raw elevation file, also
referred to as the “qfit” format, is shown in Table 1. For purposes of the analysis within this report, a subset of the data within the full 14-word raw elevation file was extracted, including latitude, longitude, and elevation.
Figure 4. Map showing the extent of the individual passes between Corolla, North Carolina and Oregon Inlet, North Carolina made during the September 26, 1997 ATM survey. The direction of each pass is listed in parentheses.
Figure 5. Map showing the extent of the individual passes between Corolla, North Carolina and Oregon Inlet, North Carolina made during the September 27, 1997 ATM survey. The direction of each pass is listed in parentheses.
| Word | Field description | Units |
|---|---|---|
| 0 | relative time | milliseconds |
| 1 | latitude | microdegrees |
| 2 | longitude | microdegrees |
| 3 | elevation | millimeters |
| 4 | transmit energy | counts |
| 5 | receive energy | counts |
| 6 | scan azimuth | millidegrees |
| 7 | pitch | millidegrees |
| 8 | roll | millidegrees |
| 9 | passive brightness | counts |
| 10 | passive latitude | microdegrees |
| 11 | passive longitude | microdegrees |
| 12 | rough elevation | millimeters |
| 13 | GPS time | hhmmss |
Table 1. Raw elevation data record format (“qfit” format)
3.3 Coordinate Transformation
ATM data is expressed in ITRF94 with the coordinates referencing the WGS84 ellipsoid. Survey data collected by List, Hansen, and Reiss were received within the same reference frame as the ATM data and required no conversion. The Holman buggy data required a transformation before direct comparisons between the data could be performed.
The Holman survey data were expressed in a local cartesian coordinate system used by USACE for research around the Duck FRF pier. The data were globally referenced to the North American Datum of 1983 (NAD83) with orthometric height based on the National Geodetic Vertical Datum of 1929 (NGVD29). The original Matlab script used to convert the data from geographic latitude and longitude coordinates to X and Y coordinates was rewritten to reverse the original transformation. The NGVD29 orthometric heights were converted to Geodetic Reference System of 1980 (GRS80) ellipsoid heights by adding
the geoid heights, calculated using the National Geodetic Survey’s “VERTCON” and “GEOID96” programs, to the orthometric heights. In the final step the NAD83 coordinates were converted to ITRF94 using published methods (Dana 1997).
3.4 Data Comparisons
To fully assess ATM beach surveys, intercomparisons of overlapping ATM swaths collected on the same day and on different days were conducted to assess repeatability of ATM measurements. In addition, intercomparisons were made to ATM and ground survey data collected using various beach survey methods to assess ATM measurement accuracy in relation to the ground surveys. Data comparisons were constrained to the subaerial beach between the dune line and waterline. This data delineation was accomplished by creating beach-only polygons for the study area. Any data falling outside the polygons were discarded. The data were further filtered to eliminate outlier elevation measurements beyond the bounds of reasonable beach elevations for the area (-45 to -32 meters WGS84 ellipsoid heights). ATM to ground survey comparisons were restricted to surveys collected on the same day in order to eliminate the influence of day-to-day beach change except for the List buggy survey, which was collected on the relatively stable upper beach.
Elevation measurements within two data sets were compared by selecting each point from one data set and locating all points in the second data set within a fixed horizontal radius of the point. Elevation differences were calculated between the locating point in the first data set and each of the identified points in the second data set. For these comparisons, a 1-meter radius was used because it closely correlates with the size of the laser footprint.
Three basic statistics are used to examine differences between surveys:
��0HDQ��GLIIHUHQFH�
�UHIHUV�WR�WKH�PHDQ�HOHYDWLRQ�GLIIHUHQFH�RU�RIIVHW
between compared data sets.
��5DQGRP��GLIIHUHQFH�
�LV�WKH�VWDQGDUG�GHYLDWLRQ�DERXW�WKH�PHDQ�RI�WKH
elevation differences between data sets.
(3) “Total” difference is the root-mean-squared (RMS) of differences, or deviation of the differences about zero, which reflect combined mean difference and random differences.
Statistics were generated for individual ATM pass intercomparisons as well as summary statistics generated by combining all matching points from all passes for a set of comparisons (i.e., September 26 to September 27 ATM comparisons). This process resulted in each overlapping point having an equal weight in the generation of the summary statistics.
As a verification of the validity of the comparison method and program, Table 2 shows the statistics from intra-file comparisons for two ATM passes.
| ATM Pass | ����� | ����� | RMS (cm) | # Points |
|---|---|---|---|---|
| 09/26 135929 pass | 0.0 | 4.7 | 4.7 | 6,638 |
| 09/27 182303 pass | 0.0 | 4.4 | 4.4 | 5,829 |
Table 2. Intra-file comparison results
The results of the comparison of points in a file to the same set of points are consistent with the expected results (i.e., mean difference = 0.0). The 4.4 to 4.7-centimeter variation in the data can be attributed to the 1-meter search radius and the data density, which cause the inclusion of points other than the exact matching point in the comparisons (Figures 6 and 7). It should be noted that this variation accounts for between 23 and 60 percent of the random differences seen in the ATM comparisons. Figure 8 shows the effects of reducing the search radius size on the random difference.
Figure 6. Difference plots from an intra-pass comparison of ATM measurements from pass 135929 on September 26. The two dashed lines bound elevation differences between two standard deviations. The gray line is the regression line calculated from the difference values.
Figure 7. Difference plots from an intra-pass comparison of ATM measurements from pass 182303 on September 27. The two dashed lines bound elevation differences between two standard deviations. The gray line is the regression line calculated from the difference values.
3
2
1
0
Figure 8. Graph showing the relationship between changes in the horizontal search radius and standard deviation (random difference) from the intra-file comparison of ATM pass 135929.
15
4.0 RESULTS AND DISCUSSION
The results from the three types of survey comparisons, ATM to ATM, ATM to ground surveys, and ground surveys to ground surveys, are presented in the following section. The labels identifying individual ATM passes refer to the time of the pass in Greenwich Mean Time (GMT) (e.g., 135929 equals 1:59:29 PM GMT).
4.1 ATM to ATM Comparisons
The ability of the ATM to provide consistent repeat measurements was evaluated by comparing overlapping ATM swaths. Comparing results from a survey over a stable flat surface with beach surveys provides the opportunity to assess the impact of the unique beach morphology on the survey measurements while overlapping ATM comparisons assess the capability to consistently repeat measurements over time.
4.1.1 ATM to ATM Over a Stable Geomorphological Feature A flat short-grass section of the Wright Memorial was selected as a stable geomorphological feature over which to compare elevation measurements from overlapping ATM passes. Two passes from September 26 and three from September 27 contained elevation data for the selected area. Table 3 presents the results of the pass comparisons for each day and between days. The mean difference for all comparisons is between 3.4 centimeters and 9.2 centimeters and the RMS range is 8.3 to 14.3 centimeters. Between 62 and 82 percent of all observations are within 10 centimeters of the mean. Figure 9 shows a plot of the elevation differences and Figure 10 presents associated histograms from the September 26 to September 27 comparison. A graph of the statistics from the individual pass comparisons (Figure 11) indicates excellent agreement between all passes except pass 145947 on September 27, which indicates an offset 10 centimeters higher than other intercomparisons. Discarding pass 145947
produces a mean difference of 4.6 centimeters and a standard deviation of 8.7 for between-day comparisons.
| Comparison | ����� | ����� | RMS (cm) | # Points |
|---|---|---|---|---|
| All 9/26 passes | 3.4 | 7.6 | 8.3 | 7,113 |
| All 9/27 passes | 8.0 | 11.4 | 14.0 | 21,085 |
| All 9/26 minus all 9/27 passes | 9.2 | 11.0 | 14.3 | 48,154 |
Table 3. Summary of ATM to ATM comparisons for the Wright Memorial
Figure 9. Difference plot from a comparison of ATM measurements for September 26 and 27 over a section of the Wright Memorial. The positive offset in difference values indicates measurements from September 26 tended to be higher than September 27 elevations. The two dashed lines bound elevation differences between two standard deviations. The gray line is the regression line calculated from the difference values.
(a)
(b)
(a)
0.25
0.20
Elevation Difference (m)
0.15
0.10
0.05
0.00
(b)
0.25
0.20
Elevation Difference (m)
0.15
0.10
0.05
0.00
(b) September 26 and 27 between-day comparisons.
4.1.2 ATM to ATM Same-Day Comparisons Intra-day comparisons of measurements over the beach surface were made with ATM passes conducted on September 26 and September 27. Two sets of comparisons were made. First, the ATM data were constrained to a 1-meter horizontal radius of all 70-kilometer List buggy data points. This produced a shore-parallel line along the high beach with results that are directly comparable to the results of all List buggy survey comparisons. The second set of comparisons was made for elevation measurements constrained to the beach surface between the waterline and dune line. Results from these comparisons provided an assessment of the ATM over a wider range of beach
morphology than the first method and greatly increased the number of points used within the comparisons.
After the data sets were delineated, all passes for September 26 excluding 135929, were compared with the September 26 135929 pass (Figure 4) and all passes for September 27 excluding 182303 were compared with the September 27 182303 pass (Figure 5). The 135929 and 182303 passes were selected for comparison against the other passes for the day because they generally included coverage along the entire length of the study area. Table 4 presents summary statistics of the results. Figures 12 to 19 show difference plots by latitude and their associated histograms for the two sets of comparisons for September 26 and 27. Between 54 and 62 percent of all observations agreed within 10 centimeters of the mean difference. The slope of the regression line in the September 26 plots (Figures 12 and 16) indicate a trend of increasing elevation differences with latitude (i.e., from south to north). This same trend is not as apparent in the September 27 plots (Figures 14 and 18) for several reasons. Firstly, each of the individual passes making up the summary plots cover varying portions of the entire latitude bounds presented in the summary plots. Secondly, each of the individual pass intercomparisons have a unique mean difference or offset. When summarizing the elevation differences in a single plot, these two conditions results in a canceling of the actual trends observed in separate plots of individual pass intercomparisons (Figures 20 to 24). Good agreement exists between September 26 and 27 beach and List constrained data sets, although there is a higher random difference for surveys over the entire beach surface.
| Comparison | ����� | ����� | RMS (cm) | # Points |
|---|---|---|---|---|
| 135929 pass minus all 9/26 passes along List buggy track | -6.1 | 12.1 | 13.5 | 8,939 |
| 182303 pass minus all 9/27 passes along List buggy track | 0.9 | 13.8 | 13.8 | 4691 |
| 135929 pass minus all 9/26 passes over beach surface | -6.7 | 14.8 | 16.2 | 1,092,692 |
| 182303 pass minus all 9/27 passes over beach surface | -0.8 | 19.0 | 19.1 | 535,803 |
Table 4. Summary of ATM to ATM comparisons for September 26 and 27, 1997
(a)
- (b) (a)
- (b) (a)
- (b) (a)
- (b)
Figure 25 presents the comparison statistics for each September 26 ATM pass compared to the September 26 135929 pass. The negative offset, or mean difference, indicates measurements for the 135929 pass were generally lower than the other passes for the day. The passes are ordered in chronological sequence, the pass identifier being the time of the pass in GMT; therefore, the graphs indicate trends over time. A trend can be observed in both graphs beginning with the 130755 pass, where there is an increase in the mean difference and RMS over time, peaking at pass 132917, then decreasing to mean differences between ± 5 centimeters for the remaining four passes. The time between the first and last pass is approximately 2.5 hours. There is little variation in the random difference from pass to pass.
Figure 26 shows the comparison statistics for each September 27 ATM pass compared to the September 27 182303 pass. The passes 144126, 145204, and 145947 show a general increase in the mean difference and RMS over the 19 minutes between the three passes. In sequential passes beginning three hours after pass 145947, less variation is seen in the mean difference. This trend is apparent in both the List track and beach constrained data sets, although the peak offset is lower (12.4 centimeters versus 16.5 centimeters) and occurs in the 145204 pass of the List track comparison (Figure 26a) as opposed to the 145947 pass in the beach surface comparison (Figure 26b).
(a)
0.25
0.20
0.15
Elevation Difference (m)
0.10
0.05
0.00