<%BANNER%>

Accuracy Evaluation of Terrestrial Lidar and Multibeam Sonar Systems Mounted on a Survey Vessel

Permanent Link: http://ufdc.ufl.edu/UFE0042609/00001

Material Information

Title: Accuracy Evaluation of Terrestrial Lidar and Multibeam Sonar Systems Mounted on a Survey Vessel
Physical Description: 1 online resource (51 p.)
Language: english
Creator: Dix, Michael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: applanix, echosounder, error, gnss, gps, inertial, lidar, mbes, multibeam, pos, r2sonic, sbet, terrestrial
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: ACCURACY EVALUATION OF TERRESTRIAL LIDAR AND MULTIBEAM SONAR SYSTEMS MOUNTED ON A SURVEY VESSEL There is an emerging demand for data that can accurately describe the transitional environment between land and water. Survey equipment and techniques are meeting this demand but their resolution and repeatability capabilities are still being addressed. This research provides a performance test of terrestrial lidar and multibeam sonar systems integrated with Global Navigation Satellite System/Inertial Navigation System (GNSS/INS) positioning and orientation on a vessel platform. To measure the accuracy of the data, experiments were designed to allow the lidar and sonar systems to acquire scans of a control target that extended above and below the water surface. The scans were acquired under normal and induced conditions present in the marine survey environment such as variations in speed, range, motion, and orientation. The scan data was post-processed to create a blended GNSS/INS solution accounting for optimal sensor position and orientation dynamics throughout data collection. The scans from each data set were viewed in point cloud software and coordinates representing the center of the target were selected. These coordinates were then compared with the control coordinates for the target and errors in northing, easting, elevation, and planimetric dimensions were calculated. Standard deviation, Root Mean Squared Error (RMSE), and Mean were also computed across all data sets for each experiment. Horizontal RMSE values of 0.06m and 0.03m were achieved for the lidar and sonar data, respectively. Vertical RMSE results of 0.04m were found for both data types. These results were comparable with previous mobile mapping research involving similar systems. Contributing error sources are also discussed regarding expected and achieved results. The methodology and results of this research provide validation of lidar and sonar data acquired from a survey vessel.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Michael Dix.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Abd-Elrahman, Amr H.
Local: Co-adviser: Dewitt, Bon A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042609:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042609/00001

Material Information

Title: Accuracy Evaluation of Terrestrial Lidar and Multibeam Sonar Systems Mounted on a Survey Vessel
Physical Description: 1 online resource (51 p.)
Language: english
Creator: Dix, Michael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: applanix, echosounder, error, gnss, gps, inertial, lidar, mbes, multibeam, pos, r2sonic, sbet, terrestrial
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: ACCURACY EVALUATION OF TERRESTRIAL LIDAR AND MULTIBEAM SONAR SYSTEMS MOUNTED ON A SURVEY VESSEL There is an emerging demand for data that can accurately describe the transitional environment between land and water. Survey equipment and techniques are meeting this demand but their resolution and repeatability capabilities are still being addressed. This research provides a performance test of terrestrial lidar and multibeam sonar systems integrated with Global Navigation Satellite System/Inertial Navigation System (GNSS/INS) positioning and orientation on a vessel platform. To measure the accuracy of the data, experiments were designed to allow the lidar and sonar systems to acquire scans of a control target that extended above and below the water surface. The scans were acquired under normal and induced conditions present in the marine survey environment such as variations in speed, range, motion, and orientation. The scan data was post-processed to create a blended GNSS/INS solution accounting for optimal sensor position and orientation dynamics throughout data collection. The scans from each data set were viewed in point cloud software and coordinates representing the center of the target were selected. These coordinates were then compared with the control coordinates for the target and errors in northing, easting, elevation, and planimetric dimensions were calculated. Standard deviation, Root Mean Squared Error (RMSE), and Mean were also computed across all data sets for each experiment. Horizontal RMSE values of 0.06m and 0.03m were achieved for the lidar and sonar data, respectively. Vertical RMSE results of 0.04m were found for both data types. These results were comparable with previous mobile mapping research involving similar systems. Contributing error sources are also discussed regarding expected and achieved results. The methodology and results of this research provide validation of lidar and sonar data acquired from a survey vessel.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Michael Dix.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Abd-Elrahman, Amr H.
Local: Co-adviser: Dewitt, Bon A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042609:00001


This item has the following downloads:


Full Text

PAGE 1

1 ACCURACY EVALUATION OF TERRESTRIAL LIDAR AND MULTIBEAM SONAR SYSTEMS MOUNTED ON A SURVEY VESSEL By MICHAEL ESTEN DIX A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Michael Esten Dix

PAGE 3

3 To X, Y, and Z

PAGE 4

4 ACKNOWLEDGMENTS I would like to acknowledge my parents, Rob and Anne, for their enduring support throughout my adventures. I would like to thank my brother, Chris, and his family, Ivy and Riley, for their love and wisdom. Amr AbdElrahman, my advisor, has provided valuable guidance and input throughout this research. Bon Dewitt, Scot Smith, and the Geomatics program have provi ded quality resources and the opportunity to be part of a challenging learning environment. I would like to acknowledge my colleagues that have provided their time, thoughts, and efforts towards this project with s pecial thanks to Lou Nash, Keith Dixon, M ario Hernandez, John Smith, Donna Nash, and the Measutronics Corporation. I would also like to thank Bruce Francis from Applanix, Jimmy Green and Joe Revelle from Optech, and Charles Brennan from R2Sonic.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 ABSTRACT ..................................................................................................................... 9 1 INTRODUCTION .................................................................................................... 11 1.1 Applications ...................................................................................................... 13 1.2 Error Analy sis ................................................................................................... 14 1.3 Background ....................................................................................................... 16 1.3.1 Terrestrial Lidar ....................................................................................... 16 1.3.2 Multibeam Echosounder .......................................................................... 17 1.3.3 Sensor Positioning and Orientation ......................................................... 18 2 METHODOLODY .................................................................................................... 22 2.1 Site and Equipment ........................................................................................... 22 2.2 Calibration ......................................................................................................... 24 2.2.1 Lever Arm Offsets .................................................................................... 24 2.2.2 Lidar Boresight Procedure ....................................................................... 24 2.2.3 Multibeam Patch Test .............................................................................. 26 2.3 Variable Conditions ........................................................................................... 27 2.4 Data Collection ................................................................................................. 28 2.4.1 Experiment 1 ........................................................................................... 28 2.4.1.1 Control coordinate computations ................................................... 28 2.4.1.2 Navigation data computations ........................................................ 29 2.4.1.3 Lidar target scan coordinates ......................................................... 30 2.4.1.4 Multibeam Echosounder (MBES) target scan coordinates ............. 30 2.4.2 Experiment 2 ........................................................................................... 31 2.4.2.1 Contr ol coordinate and navigation data computations ................... 31 2.4.2.2 MBES target scan coordinates ....................................................... 32 3 Results And Analysis .............................................................................................. 42 3.1 Experiment 1: Terrestrial Lidar .......................................................................... 42 3.2 Experiment 2: MBES Sonar .............................................................................. 44 3.3 Lidar vs. Sonar .................................................................................................. 45 4 CONCLUSION ........................................................................................................ 48 LIST OF REFERENCES ............................................................................................... 49

PAGE 6

6 BIOGRAPHICAL SKETCH ............................................................................................ 51

PAGE 7

7 LIST OF TABLES Table page 2 1 Variable conditions for Experiment 1 .................................................................. 41 2 2 Variable conditions for Experiment 2 .................................................................. 41 3 1 Terrestrial lidar accuracy, Experiment 1 (in meters). .......................................... 47 3 2 Multibeam Echosounder (MBES) sonar accuracy, Experiment 2 (in meters) ..... 47

PAGE 8

8 LIST OF FIGURES Figure page 1 1 General georeferencing formula (Barber et al., 2008) for terrestrial lidar and Multibeam Echosounder (MBES) system. .......................................................... 20 1 2. Visual representation of georeferencing formula (from Figure 12). .................. 21 2 1 Experiment site location map. (Note: Image retrieved from Microsoft Bing Maps, 2010). ....................................................................................................... 33 2 2 Target system A) on its side to measure offsets before installation and B) installed on a concrete dock. .............................................................................. 34 2 3 Terrestrial lidar and MBES sonar instruments mounted on survey vessel .......... 35 2 4 Target locations for boresight procedure. ........................................................... 35 2 5 Experiment site showing vessel course and scan direction with respect to the face of the target. Angular orientations and linear features shown here are approximate and not to scale. (Note: Background image retrieved from Microsoft Bing Maps, 2010). ............................................................................... 36 2 6 Lidar target for Experiment 1 .............................................................................. 37 2 7 Sonar target for Experiment 1 ............................................................................. 37 2 8 Profile and plan views of the target system with vertical and horizontal offsets, respectively. Note: Angular orientations and linear features shown here are approximate and not to scale. .............................................................. 38 2 9 Example of XY (horizontal) coordinate selection of lidar target, Experiment 1 ... 38 2 10 E xample of Z (vertical) coordinate selection of lidar target, Experiment 1 .......... 39 2 11 Sonar target for Experiment 2 ............................................................................. 39 2 12 Example of XYZ coordinate selection for sonar target, Experiment 2. ................ 40 3 1 Mean error (in meters) in lidar and sonar data sets. ........................................... 46 3 2 S tandard deviation (in meters) in lidar and sonar data sets. ............................... 46

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ACCURACY EVALUATION OF TERRESTRIAL LIDAR AND MULTIBEAM SONAR SYSTEMS MOUNTED ON A SURVEY VESSEL By Michael Esten Dix December 2010 Chair: Amr Abd Elrahman Cochair: Bon Dewitt Major: Forest Resources and Conservation There is an emergi ng demand for data that can accurately describe the transitional environment between land and water Survey equipment and techniques are meeting this demand but their resolution and repeatability capabilities are still being addressed. This research prov ides a performance test of terrestrial lidar and multibeam sonar systems integrated with Global Navigation Satellite System/Inertial Navigation System ( GNSS/INS) positioning and orientation on a vessel platform. To measure the accuracy of the data, experi ments were designed to allow the lidar and sonar systems to acquire scans of a control target that extended above and below the water surface. The scans were acquired under normal and induced conditions present in the marine survey environment such as var iations in speed, range, motion, and orientation. The scan data was post processed to create a blended GNSS/INS solution accounting for optimal sensor position and orientation dynamics throughout data collection. The scans from each data set were viewed in point cloud software and coordinates representing the center of the target were selected. These coordinates were then compared with the control coordinates for the target and errors in northing, easting, elevation, and planimetric dimensions were calculated. Standard deviation, Root Mean Squared Error

PAGE 10

10 ( RMSE) and Mean were also computed across all data sets for each experiment Horizontal RMSE values of 0.06 m and 0.03 m were achieved for the lidar and sonar data, respectively. Vertical RMSE results of 0.04m were found for both data types. These results were comparable with previous mobile mapping research involving similar systems. Contributing error sources are also discussed regarding expected and achieved results. The methodology and results of t his research provide validation of lidar and sonar data acquired from a survey vessel.

PAGE 11

11 CHAPTER 1 INTRODUCTION In many co astal and inland waterways there exists a need for threedimensional (3 D) spatial data of features above and below the water surface to support construction, engineering, monitoring, and environmental projects. Many times this involves civil structures such as bridges, seawalls, and port areas where features necessarily extend through both environments. Current methods to acquire t hese data involve traditional tape measurements and the deployment of divers to more advanced combinations of land and hydrographic surveying techniques. This research evaluates the accuracy of a system that can acquire data above and below the water surf ace using a single survey vessel. The goal of this research is to test whether terrestrial lidar and multibeam sonar systems mounted on a survey vessel can achieve horizontal and vertical error values equal to or less than 0.05m as established by Barber ( et al., 2008) for error pre analysis estimates of a mobile terrestrial lidar system mounted on a van. Surveying and mapping has grown in scope along with technological advances in hardware and software. Governments are requiring more spatial data concerni ng their lands and underwater features in a format that is comprehensive, accurate, and capable of being referenced to historic and future data. The United States Army Corps Civil Works Program (2008 ) estimates a 2010 fiscal year budget of $1.89 billion t owards their Navigation business program which includes planning, construction, operation, and maintenance of channels, locks, dams, and other inland waterways. They also budgeted $1.32 Billion in 2010 for the Flood and Coastal Storm Damage Reduction program which includes the design and construction of dams, levees, jetties and seawalls. Shorelines and coastal areas represent a unique and changing boundary of

PAGE 12

12 traditional hydrographic and topographic maps. In many situations, the ability has now met the need for a seamless integration of recent, highresolution topographic and bathymetric data (Gesch & Wilson, 2002). Mobile lidar mapping from a terrestrial platform has been developing since the early 1990's ( Grejner Brzezinska, Li, Haala, & Toth, 2004). This method of data collection is effective in its ability to capture large amounts of data quickly and accurately. Conventionally terrestrial lidar instruments are used while mounted on a fixed tripod; o nly in recent years has the terrestrial lidar sc anner been considered on a mobile vessel platform ( Talaya et al., 2004). Due to Global Navigation Satellite System/Inertial Navigation System (GNSS/INS) positioning in real time and through post processed kinematic solutions, there is less of a requirement to scan control targets for georeferencing data. However, as mobile mapping environments vary significantly, scanning control targets provides a check for systematic errors as well as evaluating overall system performance in a survey area. Sonar plat forms for mapping applications were born in the mobile environment and have continued to increase in resolution and coverage. Multibeam echo sounders (MBES) have the ability to ensonify an area with spatial resolution dense enough to allow for minimal int erpolation when developing a surface or elevation model Depending on equipment configuration and range from the sensor to an object, the spatial resolution of an MBES sensor can be similar to that of a terrestrial lidar sensor. Sonar systems are difficult to test for absolute accuracy due to the difficulty in establishing underwater control points. Digital elevation models can be compared over multiple data sets and specific features can be analyzed in the same manner, but these

PAGE 13

13 methods fall short of measuring the accuracy of a specific point that has been established by independent control. The evolution of this methodology has come to a point where the accuracy and precision of the data should be measured to continue its development. This is cardinal for u nderstanding its suitability in many applications. 1.1 Applications Data collected from this type of terrestrial lidar and multibeam sonar configuration is advantageous in its ability to tie into existing r eference and control data. These data are currently being combined in Geographic Information Systems (GIS) but may suffer due to the variability in data sources. Gesch and Wilson (2002) note that, because existing topographic and bathymetric data have been collected independently for dif ferent purposes, it has been difficult to use them together at the land/water interface owing to differences in format, projection, resolution, accuracy, and datums. For example, bathymetric data of major rivers in a county may be added to a GIS of that county with polygons for property lines and setbacks. This shows in general, how the depth and location of waterways relate to property boundaries but it is typically not robust enough to derive survey measurements and interpretations. Monitoring appli cations of this set up are numerous. Current methods of assessing, categorizing, and reviewing the stability of structures that extend above and below the water surface are tedious, subjective, and many times not feasible. These structures include docks, bridges, seawalls, dams, levies, locks, and jetties. For large projects, a combined lidar and sonar system may decrease the time it would take to inspect these structures. These data can also be used for feature extraction (Manandhar & Shibasaki, 2002) and classification (McGonigle, Brown, & Quinn, 2010). E nvironmental applications involving coastal erosion and habitat monitoring will benefit

PAGE 14

14 from the processes that can be modeled. During marine construction, many operations can be observed. As support structures are being installed, their orientation can be confirmed by control points above and below the water. For engineering, as built surveys can also be generated and recorded. R epeatability and integration of these types of data confirm the value for its extended use in future applications. 1.2 Error Analysis Data acquired by r emote sensing instruments from a mobile platform have many sources of error. Most instruments, sensors, and equipment used in the field of Geomatics have published accuracy and precision tolerances that are used in evaluating performance ( Ernstsen et al. 2006). Many times error analysis is based on factory calibration values as well as field or "in situ" calibration techniques For mobile mapping systems, the primary sour ce of positional error is expected to be kinematic GNSS positioning (B arber, Mills, Smith Voysey, 2008). There have been significant strides in the field of surveying that established the integration of G NSS positioning with an INS for mobile surveying applications in the air, on water, and for mobility on land ( Mostafa, Hutton, & Reid, 2001) Many of these systems have been able to determine their real time position at the centimeter level while synchronizing to the timing of scanning instrument With m obile marine platforms the accuracy of a system is often assumed to this extent, but in many surveys the actual points measured are not verified against an absolute control system. Barber (et al., 2008) conducted an accuracy evaluation of mobile terrestr ial lidar experiments Their research was conduct ed in an urban environment with sensors mounted on a van driven on a paved street. In contrast to Barber, the lidar and sonar sensors in this research were mounted on a survey vessel driven in a port envir onment

PAGE 15

15 and thus subject to increased dynamic motion. U sing a formula built upon Ellum and El Sheimy (2002), Barber established preanalysis expected horizontal and vertical errors of approximately 0.05m at a range of 5 .0 m. They were able to achieve an av erage vertical RMS error of 0.046m with a standard deviation between 0.02m to 0.03m. Horizontally their results indicated an accuracy of approximately 0.10m. Their preanalysis estimates did not take into account the quality of the GNSS control for their targets or the accuracy associated with selecting coordinates of the target center from the scan. However, these components were addressed in their research as contributing sources of error and are attributed to the lower than expected horizontal accurac y from their experiments. While research to develop error budgets for a MBES system (Hare, 1995) exists there is little on research evaluating expected and achieved absolute horizontal and vertical error of MBES system s ( Ernstsen et al., 2006). In this research we use the same positioning and orientation system to test both the lidar and sonar sensors The lever arm calibration measurements were also determined using the same procedure and equipment for both lidar and sonar sensors The primary differ ence exists in sensor ranging methods and capabilities. MBES beam forming of an acoustic pulse through water is subject to sound velocity, objects in the water, and attenuation among other influences as compared to the relatively minimal influences a laser (from a terrestrial lidar instrument) encounters through the medium of air. The raw range resolution, or accuracy of the measured distance from sensor to an object of t he MBES used in this research is slightly (<0.01m) larger than that of the lidar. T he methods of calibrating the angular orientation offsets of each sensor reference frame to the Inertial Measurement

PAGE 16

16 Unit ( IMU ) reference frame are quite different as described in sections 3.4 and 3.5 of this paper. Recognizing these similarities and diff erences, the preanalysis horizontal and precision error estimates established by Barber (et al. 2008) were considered also against the MBES results of the experiments in thi s research. The preanalysis estimates involved a georeferencing formula (Figure 11) expanded from Ellum and ElSheimy (200 2) that can be visually represented (Figure 12) to descr ibe possible sources of error. The position vector of the GPS antenna in the mapping frame at time t describes the position solution from the GPS receiv er and is where the first measurement error source is found. The rotation matrix between the IMU frame and the mapping frame at time t will introduce error from the GPS heading solution determined between the two GPS antennas and from the dynamic motion m easurements of the IMU. Errors from lever arm offsets are present in the position vector of the sensor (terrestrial lidar or MBES) in the IMU frame and from the position vector of the GPS antenna in the IMU frame. Angular errors from the rotation matrix between the sensor frame and the IMU frame are a result of the boresight procedure ( for the terrestrial lidar) and the patch test ( for the MBES). The determination of the object point in the sensor frame involves error from the sensors (terrestrial lidar or MBES) ranging accuracy at a particular angle off nadir, or center 1.3 Background 1.3.1 Terrestrial Lidar An advantage of a terrestrial lidar instrument mounted on a vessel is that it can be operated using the same positioning and orientation system as a MBES. Some software programs offer real time viewing of lidar data during acquisition. This can

PAGE 17

17 provide instantaneous data coverage analysis. If the data collected is not enough or not correct, a quick determination can be made to do the survey again. The lidar instrument can be mounted on a survey vessel in a fixed position, programmed to repeatedly scan a vertical swath Thus, as the boat moves forward, the scanner is imaging in a fixed direction against the chan ging landscape as the vessel moves by. Stewart and Canter (2009) wr ote that, A marine vessel affords a unique perspective from which to capture these near shore structures and topography, providing a viewpoint im possible from the land or air. The scanner can be optimized for a particul ar project in its calibration and settings. The vertical swath width (angle) can be changed to cover more or less area. The sensor rotation angle between laser pulses can also be increased or decreased to control the amount of points along the swath. If the settings are too sensitive there may be too much noise in the data creating false edges or layers. Conversely, if too little data is collected, t here may not be enough poi nts to model a surface, line, corner and other types of targets 1.3.2 Multibe am Echosounder There are a few types of sonar instruments that can acquire a three dimensional target. For this research a MBES was selected for accurate range resolution, effective integration with a position and orientation system, and its comprehensive swath coverage similar to a terrestrial lidar sensor. A MBES transmits an acoustic pulse that generates multiple range returns from an object or objects. These returns, or "beams" are spread out to achieve an opening angle up to 160 degrees to cover a l arge area or they can be focused to within a few degrees. Similar to the lidar instrument, the angular range that the beams are spread over is referred to as the "swath width". As the boat moves forward these swaths collect dat a points within the coverag e area. While the

PAGE 18

18 lidar scanner pulse rotates along a swath width at an extremely high pulse rate of 10,000Hz (Applanix, 2009) the MBES determines all the points along its swath from one pulse at a relatively lower pulse rate up to 60Hz (R2Sonic, 2010) Mounted to the vessel underwater, the MBES can be positioned in many ways to face down, sideways, or forward to collect data from a desi red perspective. 1.3.3 Sensor Positioning and Orientation The scanning devices (lidar and sonar) were securely attac h ed to a moving vessel and thus subject to angular motion on all three axes and positionally with respect to a threedimensional coordinate system. Without correction for vessel movements on the water, the data obtained from the scanners would be disorient ed. An inertially aided GNSS system was used to provide the positioning and orientation of the mobile platform throughout data collection. To configure the system, each sensor was mounted securely to the vessel. Threedimensional distance offsets were m easured to relate each sensor to the IMU. The positioning and orientation system establishes a precise heading vec tor that was integrated with yaw motion measurements during data acquisition. A calibration for the lidar (boresight) and sonar (patch test) sensors were carried out to determine the angular orientation offsets with respect to the IMU. The value of an integrated G NSS/ INS system is in their complementary measurements. Mostafa, Hutton, and Reid (2001) described this by saying, the GPS positio n and velocity errors are bounded and noisy, while the inertial navigator errors grow unbounded but are essentially noise free. The GPS can thus be used to estimate and correct the errors in the inertial navigation solution. The system can compute a real time solution as well as a post processed smoothed best est imate of trajectory (SBET) that can be applied to the lidar and sonar

PAGE 19

19 ranging data. The post processed kinematic (PPK) solution was used in this research to achieve the highest accuracy. Another reason that this platform was chosen was that the lidar and sonar sensors were integrated with the same positioning and motion compensation system. The acquired data also went through similar post processing procedures. If there is a systematic error somewhere in the system it will be easier to find with shared system components. This may help in tying together the accuracies of both systems better than if they were acquired separately. While this may imply that the sonar and lidar data must be acqui red simultaneously during the same pass, many times the sonar may need to make multiple passes while the lidar may only need one. Or, the lidar instrument may need to be further away from the shore to capture a desired target while the MBES may need to be as close as possible for the same purpose. While simultaneous data collection may be ideal, it is not necessary and is sometimes impractical for the best and most comprehensive data coverage and resolution.

PAGE 20

20 Figure 11. General georeferenci ng formula (Barber et al., 2008) for terrestrial lidar and Multibeam Echosounder ( MBES) system

PAGE 21

21 Figure 12. Visual representation of georefe rencing formula (from Figure 12 ).

PAGE 22

22 CHAPTER 2 METHODOLODY 2.1 Site and Equipment The test site was a port area in Tampa, FL (Figure 21) off the northwest facing side of Harbour Island. The site was chosen because of accessibility to a low concrete dock structure to which a target coul d be mounted. The target system (Figure 22 ) was modified for each of the experiments and consisted of a long rod that extended above and below the water. On the section above the water a lidar target was attached to the rod to allow for coordinate extraction of the target center during post processing analysis Below t he water there was a sonar target to provide the same result. GPS receivers were used to help establish control for the project. A Trimble SPS882 Smart GPS Antenna was mounted on the target This unit has the receiver and antenna in the same housing and is designed to be mounted directly on top of a survey rod. This receiver was configured to log data internally to be post processed against a base station or network. The target was designed in this way so that only minor offsets needed to be applied t o the horizontal (X and Y) coordinates of the targets An offset for the Zaxis was measured from the Antenna Reference Point (ARP) down to the vertical center of the lidar and sonar targets. Another GPS receiver that was used as a local base station was a Trimble 5700 receiver with a Zephyr Geodetic antenna. A group of Continually Operating Reference Stations (CORS) was another key component in creating a control network for the experiments. The CORS network is comprised of GPS base stations set up over known points that continuously collect GPS data. This data is available for free through http://www.ngs.noaa.gov/UFCORS/ (2010) which is managed by the National Geodetic

PAGE 23

23 Survey (NGS) under the direction of the National Ocean Service office of the National Oceanic and Atmospheric Administration (NOAA). The lidar scanner was a LANDMark Marine instrument from Applanix. It is manufactured by Optech Inc. as an ILRIS 3D with modifications designed for mobile m apping in a marine environment. It was mounted starboard with the scanning sensor facing outward (Figure 23 ) approximately perpendicular to the centerline of the vessel. As mentioned previously, the instrument scans in a fixed direction with a vertical swath width of 40 degrees or less The multibeam echosounder (MBES) was a Sonic 2024 manufactured by R2Sonic. It was also mounted starboard and positioned approximately below the lidar scanner. A Sonar Interface Module was located inside the cabin and s erved as a junction box for the sonar head to the other sensors. The MBES faced downward (Figure 23 ) and produced a swath width across track (athwartship) of 160 degrees. Attached to the Sonic 2024 was a Valeport miniSVS sound veloci ty sensor. This dev ice provided real time sound velocity data. The position and motion of the vessel platform was observed using an Applanix POS MV system. For positioning, the system contained a GPS receiver utilizing two Trimble Zephyr Model 2 Rugged antennas. The prim ary antenna was used for position and a second antenna was used to determine a heading vector. The data from the receiver was output in a standard NMEA format that is used in real time to aid in navigation of the vessel and in processing the location and orientation of the vessel and its sensors. An inertial measurement unit (IMU), also integrated into the POS MV

PAGE 24

24 system, was used to measure pitch, roll, and yaw along with velocity movements horizontally and vertically (heave). The lidar software consist ed of Optechs ILRIS 3 D PC Controller interface (for sensor settings and data acquisition) Match V iew (for boresighting), and Parser (for exporting the data). The MBES data was collected and processed in H ypack's HYSWEEP software suite. The MBES setting s were controlled through R2Sonics Sonic Control 2000 software. The positioning and orientation data was controlled and acquired by Applanix's POSView and pos t processed in POSPac MMS. Quick Terrain Modeler and Pointools View Pro were used for 3D point c loud modeling. 2.2 Calibration 2.2.1 Lever Arm Offsets Lever arm offsets of each sensor (lidar, sonar, GPS antenna) relative to the IMU were needed to reference the lidar and sonar data to its true position and to allow for motion compensation to be appr opriately applied. A 1 second Trimble SPS930 total st ation was used to measure the coordinates of each sensor. The final coordinates were averaged from measurements taken from two different control points The coordinate system was then rotated to the c enterline of the vessel (parallel to the X axis of the IMU) and the origin (0,0,0) m oved to the reference point on top of the IMU 2.2.2 Lidar Boresight Procedure A boresight procedure was needed to accurately measure the angular orientation offsets of t he lidar sensor with respect to the IMU for all three axes To begin, 10 targets were set up at different locations within the sensor's field of view. The targets were spaced at varying depths, widths, and heights to create a diverse field of view

PAGE 25

25 (Figur e 24) The locations of each target were then measured by a robotic total station from two setups and the resulting coordinate sets were averaged. Next, the lidar scanner (mounted on the vessel) was situated to face the targets. The POS MV was turned on and began recording data. When the POS MV had recorded data for 5 minutes, the lidar scanner was used to acquire a 3D static scan of the targets This part of the process allowed the lidar sensor to scan targets with known coordinat es while the POS M V recorded its position and orientation. The POS MV was allowed to record data for 5 minutes after the scanning was completed POS MV data was recorded for extended periods of time to allow for the forward and backward processing involved in determining the present position and orientation solution at a given time. The lidar instrument was also receiving real time GPS (UTC time) information along with a 1 Pulse Per Second (1PPS) signal during the scanning process to accurately time tag each laser pulse. The data logged from the POS MV during that session was post processed in POSPac MMS to output a Smoothed Best Estimate of Trajectory (SBET) file. A SBET file is a post processed solution of the position and orientation of the IMU and GNSS sensors The lidar scan data was then loaded into a parser program, which allows georeferencing transformation and other parameters to be applied before outputting the data in a variety of standard file formats. Here the SBET file was applied to the lidar scan of the targets and then output as an ILRIS Exchange Format (IXF) file. The I XF file is loaded in to Match View along with the coordinate file containing the locatio ns of each target. The Match View program allows the center of each target in the lidar scan t o be visually selected and associated or registered with its respective

PAGE 26

26 control coordinates that were measured by the total station. This procedure was done for each tar get and these associations were used to compute a resection to determine the angular orientation of the sensor. This orientation in all three axes is compared to the orientation of the IMU provided by the SBET. The angular differences between the IMU and lidar sensor are the boresight angles. The boresight angles were applied to the lid ar scan data during post processing. 2.2.3 Multibeam Patch Test A patch test calibration for the MBES is similar to the boresight calibration for the lidar sensor in that it determines the angular orientation of the MBES with respect to the IMU. However the procedure is much different. The patch test is a sequence of separate procedures, each designed to isolate roll, pitch or yaw offset s between the MBES and IMU reference frames Latency in the timing of the MBES pulse to the position in g and orienta tion system can also be measured. While the procedures are separate, a calibration error in one parameter could adversely affect the results of another parameter. The data for this research was acquired using H YSWEEP software. HYWEEP has a patch t est mo dule that aides in determining these parameters by exploiting angular errors between certain data sets. Latency is for this system is assumed to be zero as the sonar data is UTC timetagged with a 1PPS signal directly from the POS MV system. The pitch t est consisted of two MBES data sets collected over the same line in opposite directions at the same speed over a relatively sloped bottom surface. The second test was for roll offset. Two MBES data sets were collected over the same line in opposite direc tions at the same speed over a relatively flat surface. The final test

PAGE 27

27 was for yaw offset. Two MBES data sets were collected over parallel lines at the same speed over flat area that included a sloping feature. 2.3 Variable Conditions The experiments designed for this study placed the measurement system s under varying conditions of speed, range, motion, orientation, and repeatability while the target was scanned. Some of the data sets were collected under normal or standard conditions. These standard conditions were slow speed (2 knots), close range (3m), normal boat motion, scan direction perpendicular to the face of the target (Figure 25) and consisted of one pass by the target. All data sets were assumed to be acquired under these conditions unless a variable condition was applied. The speed variable conditions were typical of hydrographic surveying; this research evaluated a speed range of slow (2 knots), medium (34 knots) and fast (56 knots). The ranges consisted of close (5m), medium (1 0m) and far (15m). Since the target was scanned from the side and at a shallow depth, the MBES was limited in range. The MBES swath width can be opened to a maximum of 160 degrees and the further away from the target it is, the lower its field of view wil l be towards an object extending vertically down from the water surface. With regard to angular motion on all three axes a vessel is subjected to pitch, roll, heading. As these motion parameter s are unavoidable, they constitute necessary conditions of each experiment. While many coastal areas are subjected to extreme amounts of motion, many inland marine environments are relativity stable. The latter environment was chosen for this research to allow for the most accurate achievable results. However, a condition variable was designed to create an extreme amount of roll motion. This was a simple variable to create by rocking the vessel from side to side

PAGE 28

28 as the ves sel passed by the target and would offer insight in to what effects this magnitude of moti on may have. Another motion condition to be evaluated was to scan the target immediately after the vessel finished making a turn. One of the experiments was also designed for the vessel to pass by the target at a 45 degree angle off perpendicular. Acqui ring multiple target scans during continuous data collection of one dat a set was also invoked to assess variability. 2.4 Data Collection 2.4.1 Experiment 1 The target was set up and attached to a concrete dock outcropping and leveled. It consisted of a 5m steel rod mounted to a dock. On the section above the water a 35cm x 43cm rectangular lidar target was attached to the rod (Figure 26 ) The target was made out of hard plastic and checkered with aluminum tape to give the surface contrasting reflect ivity. On the sectio n below the water, a metal disc 40cm in diameter was attached to the rod (Figure 27 ) as a sonar target The GPS receiver on top of the target was started and began logging a static session. Vertical offset measurements were taken us ing measuring tape from the antenna reference point (ARP) of the GPS receiver to the center of both the lidar and sonar targets. A base station was also set up nearby (<1km away) For the first experiment, 12 data sets were collected. All data sets were collected under standard conditions as described in section 2.3 unless a variable condition was applied (Table 21) 2.4.1.1 Control coordinate computations The first part of processing was to establish control coordinates (actual location) of the GPS r eceiver that was mounted on top of the target. This receiver recorded only L1 data and could not be processed with the nearest CORS stat ion (MCD5) due to the

PAGE 29

29 baseline length of approximately 13km A local base station had been set up < 0.5km away from the target receiver and a baseline between these two points was processed to determine the control coordinates of the target receiver. The next step was to apply offsets (Figure 2 8) from the ARP of the GPS antenna to the center of each target. For the lidar data, points scanned from the survey rod were used to determine the horizontal (XY) location of the target in each data set. The offset from the center of the ARP to the edge of the rod was 0.014m. The orientation of the offset was at an azimuth of 295 degrees, so a corresponding latitude (+0.006m) and departure ( 0. 013m) offsets were calculated to apply to the northing and easting control coordinates, respectively. The vertical center of the lidar target was used to determine the Z coordinate in the s cans. A tape measurement of that distance ( 0.417m) from the ARP to the center of the lidar target, was used for the vertical offset value. For the sonar target, the same horizontal offsets as the lidar target were used. Vertically, the distance measur ed from the ARP to the center of the sonar target was 4.690 m 2.4.1.2 Navigation d ata c omputations The POS MV began recording before the scanning and continued afterward for approximately five minutes. This was to ensure complete coverage for the ent ire scanning period. Using the POSPac MMS software, POS MV output files were imported. The base station data used above to process the target control was imported and designated as a single base station. The GNSS Inertial Processor was executed and the resulting SBET file was exported. The GNSS Inertial Processor combines the raw IMU and GPS data along with the base station data and incorporates a smoother along with the inertial navigator, Kalman filter, and error controller components ( Mostafa, Hutt on, & Reid, 2001)

PAGE 30

30 2.4.1.3 Lidar target scan coordinates Using the Parser software utility, the SBET and boresight files were applied to each raw data set file. They were then exported in the WGS84 UTM 17N coordinate system in ASCII format with northi ng, easting, elevation, and intensity fields. Each file was then imported into point cloud software (Quick Terrain Modeler). In this software, the data set was viewed as a 3D scene and the perspective zoomed in to focus on the target as defined by a group of points. Noisy data were removed and two points were selected from the image of the target One point was selected from the survey rod to g ive the best XY coordinates (Figure 29 ); another point was chosen from the vertical center of the target (Figure 210) For the vertical point selec tion the points were displayed to show intensity values making it easier to discern the center of the target from the checkered reflective tape pattern. 2.4.1.4 Multibeam Echosounder ( MBES) target scan coordinates Ea ch MBES data set was imported into HYPACK Multibeam Editor (MBMAX). Here the appropriate lever arm offsets and patch test angular offsets were entered. The SBET was applied and the data exported as an ASCII point file with northing, easting, and elevation fields. In HYPACK, depths (elevation) are typically positive in the downward Z direction so the elevation field was inverted during the export process. Each data set was then imported and viewed in Quick Terrain Modeler software so the horizontal and v ertical target coordinates could be selected. The sonar data produced unusable results. For each data set, there were not enough points acquired to produce an image of the target for analysis. This was likely due to a few factors. The circular target had a slight concaving edge and may have caused errant returns back to the sonar receiver resulting in sections of target to be

PAGE 31

31 recorded at incorrect distances. The point density captured by the scan was also lower than desired for analysis. At an average spacing of 5cm horizontally and vertically between points, it would be difficult to determine and select a point at the center. 2.4.2 Experiment 2 A second experiment was conducted at a later date to collect data for sonar only. The experiment procedures were similar to Experiment 1. Settings for the MBES were changed to reduce the swath wi d th to 40 degrees (instead of 160 degrees) and steer it toward the target. These adjustments increased the vertical resolution of data points ensonified on the target. The pulse rate was also increased to provide increased horizontal point density. The target structure was modified to have only one target mounted underwater for the sonar. The actual target was changed from a metal disc to a metal X mounted to the bottom of the rod (Figure 21 1 ) Aluminum tape was used along with zip ties to secure the target. For the second experiment, 14 data sets were collected. All data sets were collected under standard conditions as described in section 2.3 unless a variabl e condition was indicated (Table 22) 2.4.2.1 Control coordinate and navigation data c omputations The target was not set up in the same location as the first experiment so new control needed to be established. A GPS baseline was processed between the GPS receiver from the top of the target and a local base station. Offsets were then calculated by taking the distance of 0.028m from the center of the ARP to the edge of the target and applying it to an azimuth of 295 degrees to produce a departure of 0.025m and latitude of +0.012m. The vertical offset measured from the ARP of the GPS antenna to the center of the target was 4.00m. Three SBET files were created for this survey period. Data sets 1 to 10 were processed with the first SBET file; data sets 11 to

PAGE 32

32 13 were processed with the second SBET file; data set 14 was processed with the third SBET file. The POS MV system was restarted between the three successive data sets to create more independence between the data. 2.4.2.2 MBES target scan coordinates Each SBET file was applied to the cor responding data sets in HYSWEEP along with the lever arm offsets. The data was exported and viewed in point cloud software to select coordinates for the center of the target. While the arms of the X target di d not return complete results, the center of the target that was wrapped in aluminum tape returned consistent results. As this was ultimately the desired section of the target, this was used to select the center point of the target in each data set (Figur e 21 2 )

PAGE 33

33 Figure 21. Experiment site location map. (Note: Image retrieved from Microsoft Bing Maps, 2010).

PAGE 34

34 A B Figure 22 Target system A) on its side to measure offsets before installation and B) installed on a concrete dock.

PAGE 35

35 F igure 23. Terrestrial lidar and MBES sonar instruments mounted on survey vessel Figure 24. Target locations for boresight procedure.

PAGE 36

36 Figure 25. Experiment site showing vessel course and scan direction with respect to the face of the target. Angu lar orientations and linear features shown here are approximate and not to scale. (Note: Background image retrieved from Microsoft Bing Maps, 2010).

PAGE 37

37 Figure 26 Lidar target for Experiment 1 Figure 27 Sonar target for Experiment 1

PAGE 38

38 Figur e 28. Profile and plan views of the target system with vertical and horizontal offsets respectively Note: Angular orientations and linear features shown here are approximate and not to scale. Figure 29 Example of XY (horizontal) coordinate select ion of lidar target Experiment 1

PAGE 39

39 Figure 210 Example of Z (vertical) coordinate selection of lidar target Experiment 1 Figure 211 Sonar target for Experiment 2

PAGE 40

40 Figure 21 2 Example of XYZ coordinate selection for sonar target, Experiment 2.

PAGE 41

41 Table 21. Variable conditions for Experiment 1 Data Set Variable Condition Abbreviation 1 Standard (ST AND ARD) 2 Three passes of target during continuous data recording (MULT PASSx 3) 3 Medium speed (ME D S PD ) 4 Maximum speed (MA X S P D ) 5 Medium range (ME D R ANGE) 6 Far range (F A R R ANGE) 7 Scanned after making a slow turn, medium speed (S L TRN M E D) 8 Scanned after making a hard turn, maximum speed (H D TRN M A X) 9 Scanned after making a hard turn, maximum speed (H D TRN M A X) 10 Heavy side to side rolling motion (S IDE ROLL ) 11 45 degree angle off perpendicular ( 45 DEG OFF) 12 Standard (S TANDARD) Table 22. Variable conditions for Experiment 2 Data Set Variable Condition Abbreviation 1 Standard (ST AN D ARD ) 2 Standard (S T AN D ARD ) 3 Standard (ST AN D ARD ) 4 Back and Forth (BA CK F RTH ) 5 Two passes of target during continuous data recording (M ULT P ASSx 2) 6 Three passes of target during continuous data recording (MULT PASSx 3) 7 Medium Range (M E D R ANGE ) 8 Heavy side to si de rolling motion (S IDE ROLL ) 9 Heavy side to side rolling motion for three passes (S IDE ROLLx 3) 10 45 degree angle off perpendicular ( 45 DEG OFF ) 11 Standard after POS MV restart (S TANDARD ) 12 Three passes of target during continuous data recording (M ULT PASSx3 ) 13 Standard with modified swath to 20 degrees (ST ANDARD ) 14 Standard after POS MV restart (ST ANDARD )

PAGE 42

42 CHAPTER 3 RESULTS AND ANALYSIS Coordinates of the target center were selected for each of the data sets. The error between those coordinates and the control coordinates for northing, easting, ellipsoid height, and horizontal distance are expressed as North, East, Vert and Plan respectively (Tables 31 and 32). The Plan errors repres ent horizontal accuracy while Vert errors repre sent vertical accuracy. Statistical mean, standard deviation (1 ), and RMSE were also calculated. While the data sets for each experiment were acquired under similar variable conditions, collectively they were not the same. An example of this is that the lidar data sets included two sets acquired under standard conditions and four data sets acquired under a motion variable condition. The sonar data sets included four sets acquired under standard conditions and only two under motion variable conditions. The mean, standard deviation, and RMSE represent valuable metrics to measure the overall quality of the lidar and sonar data acquired by this system, but are difficult to use in a same/same comparison with each other. 3.1 Experiment 1: Terrestrial Lidar The average horizontal RMSE for all data sets in Experiment 1 was 0.06 m with a standard deviation of 0.02m. This was 0.01cm larger than the predicted error established by Barber (et al., 2008). The average vertical RMSE was 0.04m with a standard deviati on of 0.01m. The sets acquired under standard or normal conditions were 1 and 12. These were expected to be the most accurate as they were not subjected to any extra conditions of speed, range, motion, etc. The horizontal and vertical errors for both of these data sets were equal to or lower th a n the expected errors Data set 2 shows a horizontal error of 0.05 m and a vertical error of 0.0 5 m. For

PAGE 43

43 this set, the target was scanned three times during continuous data collection and is effectively the com bination of three scans under standard conditions. Therefore it is not surprising that the error values for this data set were consistent with the mean errors. Both of the medium and far range data sets (5 and 6) show below average horizontal and vertical errors. If large errors existed in the boresight calibration, these errors would be magnified with increased distance from the target. No bias is shown at these ranges suggesting adequate boresight calibration. Data set s 7, 8 and 9 were recorded after coming off a turn and had error values approximately consistent with the expected error except for the Easting error in data set 8. This error was 0. 10m which was more than double the Easting errors in data sets 7 and 9. There is insufficient data avai lable to determine if this error is random or a result of the motion variable. Data set 10 involved heavy side to side rolling of the vessel and sensors and created a difficult environment for the GPS and motion sensors. The GPS antennas were subject ed t o side to side movements at high velocities The IMU would be subject to extreme motion along the roll axis which would subject the lidar sensor to rapid vertical motion. The horizontal error for this data set was 0.05m while the vertical error was 0.02m revealing no apparent degradation of accuracy from this variable condition in this instance. The Easting error values for all data sets in Experiment 1 are positive suggesting a systematic error. This is confirmed by a t test at a significance level of 0.05 (t /2, 0.025, 11) rejecting the null hypothesis that the mean Easting coordinate is equal to the Easting control coordinate. This phenomenon was also found in the negative vertical error values from Experiment 1. As described by the georeferencing formula in section

PAGE 44

44 1.2, possible error sources could be attributed to the terrestrial lidar sensors dynamic ranging accuracy, GPS position and heading errors, and the errors introduced in establishing control for the target. 3.2 Experiment 2: MBES Sonar For Experiment 2 the horizontal and vertical RMSE of 0.02m and 0.03m, respectively, are lower than the predicted error established by Barber (et al., 2008). The horizontal and vertical standard deviations for Experiment 2 are between 0.01m and 0.02m and were similar to the variability from the lidar data sets in Experiment 1. Sets 1, 4, 10, and 13 produced little or no returns on the target and could not be evaluated for this experiment. The data sets acquired under standard conditions were sets 2 3, 11, and 14. The POS MV system was restarted just before set 11 and again before set 14. The horizontal and vertical error ranges for these data sets extend out to two standard deviations (95%) from the mean error. This suggests that the majority of the variability between all data sets was not due to the variable conditions. Sets 5 and 6 were acquired by scanning the target two and three times, respectively, during continuous data collection. The absolute horizontal errors were between 0.01m and 0 .02m. Set 5 had a higher than average vertical error of 0.05m. A higher than average vertical error of 0.06m was also seen in data set 12 which consisted of three scans of the target. The lidar sets in Experiment 1 acquired under multiple passes did not show this degradation in vertical precision. Data sets 8 and 9 were acquired while the vessel was subjected to heavy sideto side rolling. Set 8 had the highest vertical error (0.07m) in all the data sets in Experiment 2 while set 9 produced an error equal to the mean error for all sets in Experiment 2. All of the vertical errors, except for one (set 11), were positive and

PAGE 45

45 suggests a systematic error This is confirmed by a t test at a significance level of 0.05 (t /2, 0.025, 9) rejecting the null hypothesis that the mean Vertical coordinate value is equal to the Vertical control coordinate. The Easting errors in Experiment 2 suggest a similar systematic error. 3.3 Lidar vs. Sonar The average error s in the lidar and sonar data sets (Figure 31) reveal positive systematic errors in Easting values. As the average errors are both positive, there may be a common source that was contributing to these results. This researcher believes that during the system calibration procedures, orientation errors betw een the lever arm offset measurements, IMU, and GNSS receiver reference frames could have been introduced. At a 5 meter range to target, a 0.25 degree error in this rotation would introduce a 0.02m systematic error in horizontal position. The standard dev iations for the lidar and sonar data sets (Figure 32) cannot be proven to be significantly different using an F test at a significance level of 0.05 (F /2, 0.025, 11, 9)

PAGE 46

46 Figure 31 Mean e rror (in meters) in lidar and sonar data sets F igure 32. Standard deviation (in meters) in lidar and sonar data sets.

PAGE 47

47 Table 31 Terrestrial lidar accuracy, Experiment 1 (in meters). Data Set Conditions North East Vert Plan 1 STANDARD 0.03 0.04 0.04 0.05 2 MULT PASSx3 0.02 0.04 0.05 0.05 3 MED SPEED 0.00 0.04 0.03 0.04 4 M A X S PEED 0.03 0.03 0.03 0.04 5 M E D R ANGE 0.00 0.04 0.03 0.04 6 F A R R ANGE 0.00 0.04 0.04 0.04 7 S L TRN M E D 0.01 0. 04 0.03 0.05 8 H D TRN M A X 0.01 0.10 0.04 0.10 9 H D TRN M A X 0.01 0.04 0.04 0.04 10 SIDE ROLL 0.02 0.04 0.02 0.05 11 45 DEG OFF 0 .03 0.05 0.06 0.06 12 STANDARD 0.01 0.0 2 0.05 0.02 Mean 0.01 0.0 5 0.04 0.05 Standard Deviation 0.02 0.02 0.01 0.02 Root Mean Squared Error 0.02 0.05 0.04 0.06 Table 32 Multibeam Echosounder ( MBES) sonar accuracy, Experiment 2 (in meters) Data Set Conditions North East Vert Plan 1 STANDARD 2 STANDARD 0.01 0.02 0.03 0.03 3 STANDARD 0.02 0.00 0.05 0.0 2 4 BACK FRTH 5 MULT PASSx2 0.01 0.01 0.01 0.02 6 MULT PASSx3 0.01 0.01 0.05 0.01 7 MED RANGE 0.00 0.02 0.03 0.02 8 SIDE ROLL 0.01 0.02 0.07 0.02 9 SIDE ROLLx3 0.01 0.02 0.03 0.04 10 45 DEG OFF 11 STANDARD 0.02 0.01 0.00 0.02 12 MULT PASSx 3 0.01 0.02 0.06 0.02 13 STANDARD 14 STANDARD 0.02 0.03 0.04 0.04 Mean 0.00 0.02 0.03 0.02 Standard Deviation 0.01 0.01 0.02 0.01 Root Mean Squared Error 0.01 0.02 0.04 0.03

PAGE 48

48 CHAPTER 4 CONCLUSION Under varying conditions of speed, range, motion, orientation, and repeatability the lidar data from Experiment 1 and the sonar data from Experiment 2 achieved horizontal and vertical average errors equal to or lower than the error preanalysis estimates of 0.05m by Barber (et al., 2008) The RMSE values are also within the preanalysis estimates with the exception of the horizontal RMSE (0.06m) for lidar data in Experiment 1. For both experiments, the SBET solution for position and orientation has average horizontal and vertical RMSE values of approximately 0.01cm. The remaining average error, 0.01cm 0.05 cm, is attributed to calibration procedures ( le ver arm, boresight, and patch test), dynamic sensor ranging, and the quality of the target control coordinates It is recognized that more data sets will help to establish trends caused by the condition variables and may indicate the cause of systematic calibration errors. Future research that can capture and analyze scans of the same target center by both lidar and sonar will also be helpful in strengthening accuracy evaluation for these systems.

PAGE 49

49 LIST OF REFERENCES Applanix Corporation. (2010). LANDMARK Marine specifications Ontario, Canada. Barber, D., Mills, J. & Smith Voysey, S. (2008). Geometric validation of a groundbased m obile laser scanning system. ISPRS Journal of Photogrammetry and Remote Sensing, 63, 128 141. Canter, P., Brennan, R., & Van Den Ameele, E. (2008). Proceedings of the Canadian Hydrographic Conference and National Surveyors Conference: Tightly integrated i nertially aided post processed virtual reference station technique for marine hydrography Victoria, BC, Canada. Department of the Army Corps of Engineers Civil Works Program ( 2008). Five year development plan: Fiscal year 2009 to fiscal year 2013 (p p. 19 24). Washington, D.C. Department of the Army Corps of Engineers. (2002). Engineering and design Hydrographic surveying. (EM 1110 2 1003). Washington, D.C. Ellum, C., ElSheimy, N., (2002). Landbased mobile mapping systems. Photogrammetric Engineeri ng and Remote Sensing, 68 (1), 13 17, 28. Ernstsen, V.B., Noormets, R., Hebbeln, D., Bartholom, Flemming, B.W., (2006). Precision of high resolution multibeam echo sounding coupled with highaccuracy positioning in a shallow water coastal environment. GeoMar Lett 26 (3), 141 149. Hare, R., (1995). Depth and error budgets for multibeam echosounding. International Hydrographic Review, 72 (1). Monaco. Gesch, D., & Wilson, R., (2002). Development of a seamless multisource topographic/ bathymetric elevation model of Tampa Bay. Marine Technology Society Journal, 35 (4), 58 64. Grejner Brzezinska, D.A., Li, R., Haala, N., & Toth, C. (2004). From mobile mapping to telegeoinformatics: Paradigm shift in geospatial data acquisition, processing, and management. Phot ogrammetric Engineering & Remote Sensing, 70 (2), 197 210. Manandhar, D., & Shibasaki, R., (2002). Proceedings of Symposium on Geospatial Theory: Auto extraction of urban features from vehicleborne laser data. Ottawa, Canada. McGonigle, C., Brown, C. J., & Quinn, R. (2010). Insonification orientation and its relevance for imagebased classification of multibeam backscatter. ICES Journal of Marine Science, 67. (1010 1023).

PAGE 50

50 Microsoft Bing Maps. (2010). [Digital aerial photograph of Tampa, FL]. Retrieved from http://www.bing.com/maps/ Mostafa, M., Hutton, J., & Reid, B., (2001). Proceedings of Photogrammetric Week '01: GPS/IMU productsthe Applanix approach. Heidelberg, Germany: Wichmann Verlag. National Oceanic and A tmospheric Administration. (2010). User friendly CORS [Data file]. Retrieved from http://www.ngs.noaa.gov/UFCORS/ R2Sonic, LLC. (2010). Sonic 2024/2022 broadband multibeam echosounders operation manual, V2 .0 Santa Barbara, CA. Stewart, P., & Canter, P., (2009). Feature: Creating a seamless model. Professional Surveyor Magazine, 29 (8). Retrieved from http://www.profsurv.com/magazine/a rticle.aspx?i=70304 Talaya, J., Alamus, R., Bosch, E., Serra, A., Kornus, W., & Baron, A. (2004). Proceedings of XXth ISPRS Congress, Commission 5: Integration of a Terrestrial Laser Scanner with GPS/IMU Orientation Sensors Istanbul, Turkey.

PAGE 51

51 BIO GRAPHICAL SKETCH Michael Dix graduated from the University of Florida in 2004 with a bachelors d egree in business administration. He has worked in the residential construction and land surveying industries and most recently has been involved with systems integration for marine construction and hydrographic surveying. He began his graduate studies at the University of Florida in 2009.