Skip Maine state header navigation
Skip First Level Navigation | Skip All Navigation
|MaineDOT | Home | Contact Us||
mainedot survey and right of way
Home >Airborne GPS Controlled Photogrammetry
Airborne GPS Controlled Photogrammetry
In the fall of 1997, the Photogrammetry and Control Unit of the Survey Section requested authorization to research the use of Airborne GPS controlled Photogrammetry for engineering and design purposes. The research was to provide data toward solving questions of cost savings, accuracy and acceptability of this technology in terrain of changing topography. Since accuracy and procedural standards are still being developed for this new technology, and few practical papers showing results have been published, the MDOT decided to try it and publish the results. Two different projects with different terrain patterns and different scopes were chosen and targeted traditionally. The number of control points used in the final aerotriangulation was determined by resolving error propagation of the remaining targeted control points. Once current map accuracy standards were met within the control target coordinate set, the photogrammetric mapping was completed. Each project was mapped by a different consultant, using different aerotriangulation methods. The results indicate that current standard map accuracy standards can be met using Airborne GPS controlled photogrammetry with changing terrain and changing topography. The cost savings of the Airborne GPS controlled photogrammetry is mostly in the targeting stage of the projects, as would be expected.
Airborne GPS has a number of connotations dependent upon the context of its usage. In the photogrammetry and surveying industries two uses come to mind, flight planning and photogrammetric mapping control. The use of a GPS receiver in an aircraft is a given with many of the photogrammetric companies in the world. The GPS data can be used in real time to aid in flight line planning and execution to allow for more accurate flights. This use has proven itself effective over the past decade or so and therefore will not be explored any further in this paper. The use of airborne GPS for photogrammetric mapping is in the theoretically proven yet not fully implemented stage still. Many companies can and should dispute this statement. The use of this technology for high altitude mapping (e.g., USGS Quad mapping) is in a production stage and globally accepted. Unfortunately, the use of airborne GPS controlled photogrammetry for engineering and design purposes has not been completely proven, published, standardized and implemented. There are hints of standards for this application / technology floating around the photogrammetry community, but nothing definite has been published.
Photogrammetry is defined by the American Society of Photogrammetry and Remote Sensing as the art, science, and technology of obtaining reliable information about physical objects and the environment through processes of recording, measuring, and interpreting photographic images and patterns of recorded radiant electromagnetic energy and other phenomena. Webster's Dictionary simply defines it as 1. The use of photography in surveying, 2. Map-making from photographs. The use of aerial photography to create topologically correct maps has been around for ages. The Department used it extensively during the years of location analysis and studies such as the interstate. With today’s technology, we can obtain these maps quicker, more accurately, and electronically. By obtaining electronic photogrammetric mapping files, designers can use this mapping like it would use a typical mapping file obtained by a survey crew except that it is larger and typically covers more area. This type of mapping is primarily used for new location and/or major alignment change projects.
Photogrammetric mapping is an underutilized resource at the Maine DOT. The primary lack of usage is based on the need for very few location studies. Another part of the problem is the lack of understanding of the resource by the existing designers. Prior to this research, the last mapping project was done in 1992. The last three projects mapped in 1991 and 1992 were done with an electronic file as the final mapping medium. These files were submitted in MOSS GENIO format and, unfortunately, used very little as paper copies were used to design the alignment. MDOT has very few employees with experience in photogrammetric map manipulation using electronic media. The research projects have been submitted to the designers in electronic format, again in MOSS GENIO form.
Photogrammetric Control for Mapping
All photogrammetric control is set and located by Maine DOT personnel. The control survey crew is responsible for the geodetic quality of the coordinates used for photogrammetric mapping control points. For two of the last three mapping projects, the photo control was set using GPS methods and the third was done using traditional survey methods. The GPS equipment currently consists of two Trimble 4000 SSse’s and one Trimble 4000 SSi. The Trimble 4000 SSi was purchased after the ‘91-’92 mapping projects.
THE RESEARCH PROJECT SITES
Portland is the largest city in Maine, and is located on the coast in the southwestern portion of the state. It is the most populous city in the state. The terrain is undulating from sea level to about 50 meters above sea level. The topology is urban varying between residential and commercial buildings. Most of the major interstates cut through the center of the city. Being the major port in the state, you can imagine the amount of commerce that crosses the local highways and byways. One of the bigger problems of the area is access from the ports themselves to the hardtop. The project that we are developing is a connector that will link US Route 1 with Interstate 295 and then continue into State Route 9 (see Map 1). Some of the features in the project area include a port facility, a county jail, a shopping mall, rail lines, and a number of bridges. Because of the size of the area involved and the traffic counts on the roads, the Design Unit inquired about the availability of photogrammetric mapping. The decision was made to obtain new photography and new photogrammetric mapping at an engineering and final design scale
Augusta is the capital of Maine, and is the sixth most populous city in the state. The terrain is greatly undulating from sea level to 150 meters above sea level. The city is located forty miles north from the Gulf of Maine at the head of tide and is bisected by the Kennebec River. The topology ranges from rural to urban with a good mix of residential and commercial properties. One of the bigger problems in the city is the road capacity on the highways that connect the surrounding towns to the commercial center of Augusta. The project being developed is a partial widening / partial new alignment in an environmentally constraining area (see Map 2). Some of the features in the project area include numerous car dealerships, a cemetery, a sensitive watershed, multiple businesses, and numerous streams. The project was a good candidate for mapping for environmental reasons.
In the spring of 1997, the Photogrammetry & Control Survey Unit of the Survey Section requested projects to add to its spring photography list. At that time, our unit was approached about some possible mapping projects for the fall flying season. These requests came at an opportune time to try something new. As noted above, our department hasn’t done photogrammetric mapping since 1992. Our section met with project leaders Steve Landry (Portland) and Holly Anderson (Augusta-Manchester) to discuss the merits of the mapping and the facts surrounding the methodology of photogrammetric mapping. The two projects had differing specification needs to be met for the different map uses. It was at this time that we discussed Airborne GPS. Being a new technology to us and, for practical purposes, unproven, we needed a way to assure the accuracy of the mapping and still attempt something new. The Survey Section approached the Transportation Research Division of MDOT about obtaining research money to buffer the cost of the two projects to allow for the implementation of the research. The plan was to target the projects traditionally, obtain the coordinates to all of the targets, fly the projects with Airborne GPS and map the projects using a few targets as needed to provide accurate maps. The research
moneys were approved in early fall 1997, and the project leaders agreed to the terms of the research.
The Geodetic Control Crew of MDOT was (and is) responsible for setting all of the ground targets and obtaining all of the coordinates for control purposes. They used three geodetic grade Trimble receivers in fast static mode to generate the base control for the photography. This same crew was responsible for operating the ground units during the aerial photography aspect of the projects. The receiver in the plane was a Trimble 4000 SSi being operated by James W. Sewall personnel. During the flight planning, the James W. Sewall Company proposed collecting the GPS data at one second epochs according to industry suggestions. Being newcomers, we did not disagree. Two base stations were occupied during the time of photography, one per project.
Portland Project Specifics
A set of two parallel flightlines were created with a thirty percent sidelap to cover the initial project scope. Thirty-six targets were set throughout the project area and, using fast static GPS, we obtained the coordinates to all of them. We planned to do all of the design engineering directly from the mapping data, so the Portland project was flown at a scale of 1:2500 to create mapping at 1:500 with 0.25 m contours. The project scope was extended a season after the initial flight was completed, an additional seven targets were set and the area flown
Augusta-Manchester Project Specifics
For this project a single flightline was needed at an altitude that would generate mapping at 1:1250 with 0.50 m contours. The photography scale was set at 1:6000. The control crew set 30 targets throughout the project area. This particular project has been a high priority project since its conception, the mapping, however, was to be used primarily for environmental purposes only. That being noted, the decision had been made to ground survey the primary project area. This provides another layer of quality checks of the Airborne GPS mapping!
Requests for proposal were solicited and evaluated by the Survey Section of MDOT. Adjustment of the ground control data was the responsibility of the Photogrammetry & Control Unit of the Survey Section. The coordinate accuracy of targets for both projects was held at +/- 1.5 cm in the XY plane and +/- 2 cm in elevation All map checks were the responsibility of MDOT personnel. The decision on acceptable Airborne GPS map coordinates as compared to observed coordinates rested solely with the Photogrammetry & Control Unit.
Bradstreet Consultants, Inc., of Manchester, Maine, was awarded the Portland mapping contract(s) and James W. Sewall Co., of Old Town, Maine received the Augusta-Manchester mapping contract. Each of these contracts included: calculating the GPS exposure station coordinates, using the exposure station coordinates and holding a minimum number of known targets to solve for the coordinates of the rest of the targets, submitting the solved target coordinate data for comparison, and compiling complete photogrammetric mapping that meets national map accuracy standards.
Bradstreet Consultants chose to have the flight data computed and reduced by Geosurv in Ottawa. Computing the exact (within 2-4 cm) position of the aircraft was absolutely essential to the overall accuracy of the project. The company used Albany from ERIO Technologies as their aerotriangulation adjustment software. After many different calculations using four target coordinates, and many constraint alternatives, we determined that two additional coordinated targets in the center of the project area would be needed to tighten the control for the mapping to within 5 centimeters both horizontally and vertically. Final constraints decided upon by MDOT and Bradstreet Consultants were 4 cm for the Airborne GPS, 1 cm for the surveyed ground control and 5 microns for the photogrammetric measurements. The mapping was compiled by Bradstreet Consultants using an Intergraph / Zeiss P1. Planimetric and topographic data were collected using the KORK TRACK program. All planimetric features needed to be collected according to standards set forth in the "Surveying and Mapping Manual" developed by the Aerial Survey Group, Design Division, FHA in November of 1985. Topology was obtained by creating a digital terrain model (DTM). Masspoints and breaklines were also located to create a Triangulated Irregular Network (TIN). The TIN was then modeled to create the index and intermediate contours. The DTM and TIN enabled MDOT to generate our own contours.
The James W. Sewall company was able to compute the flight data in-house using Albany by ERIO Technologies as their adjustment program of choice. After an initial data error in the surveyed ground control was discovered and corrected, we again discussed adding another coordinated point to the adjustment for a total of five for a single flight line. The results of the adjustments seemed to indicate that the National Map Accuracies could be met for this project also. Unfortunately, due to time constraints and scope changes, the mapping area had been reduced. This did not affect the research or the results of the GPS adjustments. The adjustment constraints mutually decided upon were 3 cm for the Airborne GPS, 1 cm for the surveyed ground control and 5 microns for the photogrammetric measurements. The James W. Sewall Company compiled the mapping using direct data capture methods on a Kern DSR-11 and/or a Zeiss P3. Again all planimetric features needed to be collected according to standards set forth in the "Surveying and Mapping Manual" developed by the Aerial Survey Group, Design Division, FHA in November of 1985. All of the contours were compiled through direct stereophotogrammetric compilation (also known as "chasing" the contour).
The results of the aerotriangulation from the Portland project exceeded MDOT’s expectations. Bradstreet Consultants, Inc. / GeoSurv was supplied with the ground coordinates of seven targets, five on the original project site and two on the extension. Of the 42 targets set, only 31 showed up in the photography in a usable fashion. The eleven missing targets were from the original project site, some were covered by parked vehicles, some were off the edge of the photography and some were destroyed. The aerotriangulated coordinates for the twenty-six "unknown" targets were submitted to MDOT for comparison. The absolute value of the difference between the aerotriangulated coordinates and the field GPS generated coordinates did not exceed 0.148 meters in the Northing, 0.114 meters in the Easting and 0.106 meters in elevation. These were the maximum errors. The facts are that nearly seventy-four percent (74%) of the points had elevation differences at 0.05 meters (5 centimeters) or less. Sixty-eight percent (68%) of the Northing values came within 0.06 meters and ninety-four percent (94%) were within 0.10 meters. In the Easting, coordinate values were under 0.08 meters seventy-four percent (74%) of the time and ninety percent (90%) were within 0.10 meters (see Table 1). The extension aerotriangulation incorporated the data from the initial aerotriangulation and the results were very similar. The resulting mapping from Bradstreet Consultants met, and in some areas exceeded, National Map Accuracy Standards. According to Mark Bradstreet, ASPRS Certified Photogrammetrist and President of Bradstreet Consultants, Inc., "excellent flight data reduction, crisp photos from Sewall's Zeiss RMK and flexible aerotriangulation adjustment software" allowed us to meet the specifications with 6 known targets, eliminating the need for the remaining 30 targets. We have since determined that we can generate comparable results using only five known targets.
The results from the Augusta-Manchester aerotriangulation were encouraging even though many professionals in the industry expressed pessimistic views on using Airborne GPS on a single flightline project. The James W. Sewall Company was ultimately supplied the coordinates to five of the ground control targets. Of the thirty targets set, only one was unusable for calculations as it appeared monoscopically. The aerotriangulated coordinates for the twenty-four "unknown" targets were submitted to MDOT for comparison. The absolute value of the difference between the aerotriangulated coordinate and the field GPS generated coordinate did not exceed 0.218 meters in the Northing, 0.161 meters in the Easting and 0.377 meters in elevation. These were the maximum errors. The facts are that nearly seventy-five percent (75%) of the points had elevation differences at 0.20 meters (20 centimeters) or less. Sixty-three percent (63%) of the Northing values came within 0.10 meters and eighty-three percent (83%) were within 0.12 meters. In the Easting, coordinate values were under 0.10 meters seventy-nine percent (79%) of the time and eighty-three percent (83%) were within 0.12 meters (see Table 2). The final mapping from the James W. Sewall Company met, and in some areas exceeded, National Map Accuracy Standards.
In finalizing a cost analysis many variables came into play, the reconnaissance of the area to be mapped for target placement, target setting, GPS observation time, the additional cost of obtaining GPS data during the photography, and the additional cost for flight reduction.
The cost of recon, target setting and observation for the Portland project(s) totaled $6813.19. Because of doubling up the observations on four targets for the extension, the average per target cost was $145.00. The same cost associations for the Augusta-Manchester project totaled $4352.81, also averaging $145 per target. The cost of the obtaining GPS during the flights totaled $133.12 for the initial Portland site, $284.16 for the extension, and $90.00 for the Augusta-Manchester project. The cost for Airborne GPS flight reduction was $1400.00 per site in Portland, and $2000.00 for the Augusta-Manchester project. Using a five target arrangement (two on either end and one in the middle per flight path set), the cost for the Augusta-Manchester project would be a total of $2815.00 and the Portland project would total out at $2977.00 (a total of eight targets used -- the three additional used on the extension flightline set). The difference between mapping traditionally and using Airborne GPS techniques is $1538.00 for the Augusta-Manchester type project (single flightline set) and $3836.00 for the Portland type project (two flightline sets). If we chose to buffer the number of targets used for ground control when using Airborne GPS, it would take an additional ten (10) targets in an Augusta-Manchester type project and an additional twenty-six (26) targets in a Portland type project just to hit the break even point between traditional mapping techniques and Airborne GPS techniques. This calculates to an exceptional cost savings, not to mention the man-hour savings that would allow for the targeting of more projects for Airborne GPS mapping. Factors not considered include GPS adjustment of the ground control data, travel time to the project sites and the cost of mapping the project, as these costs are associated on an equal basis.
Both the Portland and Augusta-Manchester research projects provided mapping that met National Mapping Accuracy Standards proving to MDOT that Airborne GPS is a viable alternative. Engineering and design grade mapping can be achieved with the right conditions and for a more affordable price. We have, however, found a glitch. The good news is that the glitch is not with the Airborne GPS photogrammetric mapping. The bad news is that the glitch we encountered was with the hardware / software that the Maine DOT uses to put this data into production usage. We found that the existing mixed CADD network of Digital workstations including Apha series and model 90 VAXstations that the Preliminary Engineering Processing Unit of the Survey Section uses in conjunction with MOSS cannot handle the MOSS GENIO files supplied by the contractors. We recently acquired some Compaq 4000 personal computers that run MOSS for Windows, these machines do handle the files to our satisfaction. However, the Design Unit uses the same variation of computer equipment, and most of the users are still in the training stages of using MOSS for Windows and our new 2D drafting package, Microstation. Due to the perseverance of the staff within the Survey Section and assistance from the Highway Design Divisions computer expert, we were able to overcome any hurdles necessary to complete the research project and associated project tasks.
The results from both mapping projects look extremely promising. Both mapping scales and mapping uses look attainable using Airborne GPS. The big question of practicality has been answered and shows that Airborne GPS is a practical alternative. Timing and coordination are key factors in obtaining the right information the first time. Most flying seasons in the Northeast are short, therefore, including turnaround time, any mistakes could prove fatal. Keeping in close communication with both the ground unit and the aircraft was tantamount to the success that we encountered. We were extremely lucky to have gotten good results our very first time out. Our research shows that Airborne GPS photogrammetric mapping will work for highway engineering / design purposes. Our research has also shown that five targets seem to be the minimum required for similar projects. A safety net of three targets seems to be a reasonable offering in the off chance that something happens to any of the primary targets. By discussing the research with our photography contractor and other professionals, we have determined that kinematic GPS collected with one-second epochs used in the airborne data and fast static collection with one-second epochs in the ground receiver are the widely accepted "industry standards". We have ground-truthed the targeted points and the results are acceptable and meet accuracy standards. Our results show that the research will pay dividends for both MDOT and the industry. It is my hope that the department will begin to consider to use more photogrammetric mapping to accomplish our mission. Photogrammetric mapping has always been a valuable resource in location design and can be used for engineering purposes. The advent of electronic files that are fully design capable have made this resource even easier to use. Now that we have research that validates the usability of Airborne GPS controlled photogrammetry, the process is even more affordable. Given the right circumstances, I feel that photogrammetric mapping can meet or exceed our expectations and should therefore be used more often by the department. This research substantiates the validity of Airborne GPS controlled photogrammetry and therefore makes it a cost effective alternative to traditional photogrammetric mapping. I conclude that the department can and should make every effort to use this technology at every viable opportunity.
The author would like to express his sincere gratitude to the staff of the Survey Section of Maine DOT for without their assistance, patience, and perseverance, this research may not have been completed. Mark Bradstreet and his staff at Bradstreet Consultants Inc., and Gerry Reymore and his staff at James W. Sewall Company, were extremely helpful and dedicated to the results of the research project. I am grateful to Dale Peabody and the Research Advisory Committee for believing in the benefits that this project may provide. For additional information, you may wish to review an article in POB that was written in conjunction with this research project.
This page last updated on 9/18/13
|Copyright MaineDOT © 2010 All rights reserved.|