Bo Guo, Allen D. Poling, Mark J. Poppe
This paper summarizes the three types of real world applications of the GIS/GPS
A discussion of some real world issues such as GPS data accuracy, efficiency and effectiveness, and availability of quality maps and attribute data in applying GIS/GPS technologies are presented.
GIS is an information technology that analyzes, stores, and displays both spatial and non-spatial data. It combines the power of a relational database software and the power of a CAD package. It offers the potential to assemble and process data from diverse sources and present it in an easily understood graphical format. Geographic Information Systems (GIS) are recognized as a very useful tool in transportation planning, engineering and management. A Global Positioning System (GPS) is a positioning and navigational system. A GPS receiver receives the signal broadcast by satellites and uses information contained in the signal to calculate the position of the receiver. The positional accuracy that can be achieved with GPS ranges from 100 meters to millimeters, depending on the type of receiver used and if the data collected are differentially corrected either in real time or in a post process fashion. This space-age technology has found a variety of applications in natural resource management, urban development/analysis, agriculture, and social sciences. For the past several years, Lee Engineering has been actively exploring these new technologies in solving a variety of transportation problems. Our GIS/GPS experiences can be summarized into three general types of applications: 1) GPS as a data collection device, 2) GIS as transportation database manager, and 3) GPS as traffic design aid.
GPS AS A DATA COLLECTION DEVICE
Travel Speed Data CollectionGPS was used to collect travel speed data for a network travel speed study in the metropolitan Phoenix area. The study included data collection for 61 routes which comprised over 900 miles of arterials, freeways and HOV lanes. For each route, a minimum of three runs were made during a designated time period (either AM, Mid- day or PM). The data collection vehicles or test vehicles were equipped with GPS units which were set to record the date, time, vehicle position and velocity of the vehicle on a two-second sampling interval. Each test vehicle was also equipped with a portable computer used to record any event that caused a stop delay (defined as a speed of five miles per hour or less) other than general intersection delays. The data collection effort resulted in the collection of more than forty-mega bytes of GPS and event data. The GPS data were downloaded into a PC and differentially corrected using base station data which was provided by Maricopa County. The corrected GPS data were overlaid on a base map of the metropolitan Phoenix area for validity checking in ArcInfo before being further processed by ArcInfo, FoxPro and SAS packages. The full description of GPS data collection and processing can be found in "A GIS/GPS System Design for Network Travel Time Study" presented at the 74th Annual Meeting of the Transportation Research Board, preprint number 950332, 1995.
Traffic Sign Inventory Data CollectionGPS was used as the primary data collection tool for traffic sign inventories for the White House/Washington Mall areas and for Mount Rainier National Park. To accomplish the data collection activities, a program was developed, using the GPS vendor's software, to query the data collector for sign attribute information. The position of the sign was determined by the GPS unit while the data collector entered the attribute information into a hand-held computer which is part of GPS unit. At the end of each data collection day, the data was downloaded to the PC and differentially corrected (post-process). As part of the data collection activities for the White House/Washington Mall areas, a sketch was made and ultimately drawn in CADD, of each unique sign. For the Mount Rainier project, a color digital image of each unique sign was collected using an 8mm video recorder. For both inventories the sign image was related to attribute and location information through a sign code. A unique sign code was assigned to each unique sign legend and layout.
Roadway Inventory Data CollectionGPS was used to collect existing roadway features for several design projects which did not have complete as-built roadway inventories. In one data collection effort for a 17-mile-long scenic route, the existing no-passing zone striping and signing, location of turnouts and pull-outs, mile post markers and locations (beginning and end) of bridges were collected. To increase efficiency, bar codes were defined for different roadway features of interest before the data collection. To collect data, two data collection personnel used a vehicle equipped with a GPS unit. One data collector drove the vehicle and notified the second data collector of an approaching feature. The second data collector recorded the feature by scanning the matching feature code using a bar code reader as the vehicle passed by. To increase accuracy, more than one run was conducted. After differential correction, the GPS point data representing the events were processed through dynamic segmentation in ArcInfo. The analysis resulted in a mile-log report of all the events collected with satisfactory accuracy when checked against the available milepost log of the route.
GIS AS TRANSPORTATION DATABASE MANAGER
Identifying Roadway NeedsThe Arizona County Roadway Needs Study project recently completed by Lee Engineering created an improved system for estimating needed improvements on more than 20,000 miles of county roads in the state of Arizona. GIS was used primarily to establish the roadway inventories and to map the terrain conditions. While all the counties kept fairly complete roadway databases, only a few have their roadway database tied to GIS. In order to create a uniform GIS-based roadway database, Lee Engineering used TIGER maps from the 1990 census as the base map for those counties that did not have a GIS mapping of their own. Two major obstacles were encountered in creating the GIS county roadway inventory. The first was the map deficiencies. For this project the map accuracy was not as an important issue as the completeness and correctness of the map for the study. Manual effort went into updating the base map as well as connecting the unconnected roadways in the map. The second and the biggest challenge was mapping the existing roadway segments stored in the various database or spreadsheet formats into the base maps. Since there were no consistent county road naming conventions and very few roads in the TIGER base maps had names assigned, automatic segment mapping was out of the question. Lee Engineering had roadway inventory segments manually mapped out onto paper maps and had a unique identification number assigned to each segment. Then the segments identified on the paper maps were mapped into the base map manually through the help of AML (ArcInfo Macro Language) programs.
Generating Universe Roadway Segments for HPMSThe Highway Performance Monitoring System (HPMS) is a federally mandated program to track the features of roadways in the entire United States. The aim of the HPMS project for the Maricopa Association of Governments (MAG) was to set a spatial database of both the universe roadway segment and the sample panel segments in Maricopa County. The HPMS does not require state agencies to maintain the data on every segment of roadway within a state; however, it does require the state to maintain information on those segments which are not functionally classified as local roads. One major challenge was to inventory the universe roadway segments in a specified format. The two major segmentation criteria were homogeneity and the limited- length criteria. This required that functional and physical features, including ownership or jurisdiction, of a segment be uniform within a roadway segment of a limited length. The information available to Lee Engineering was a couple of boundary maps and two roadway network base maps for Maricopa County. The two base maps were maintained by the MAG Transportation Planning Organization and by the Maricopa County Department of Transportation. The maps had different file formats, different topological accuracies and different data attributes. The base map which contained the most important road segment attribute - roadway functional classification - had inferior accuracy. Adding further to the challenge, the two maps did not share the same street naming conventions and the physical roadway features such as number of lanes and presence of roadway medians were not available in either map. The aim of the inventory process was to create a list of segments conforming to the HPMS criteria using the best data and highest geographical accuracy available to Lee Engineering. A four step segmentation process was defined. The first step assigned jurisdiction codes to the nodes in the functional classification map by overlaying boundary coverages representing the three areas: urban and small urban, air quality, and rural areas. Having both functional classfication and HPMS area information, the map was then broken into segments homogenous in functional classification and jurisdiction through programming. The second step was the creation of a street name look-up table for converting street names used in one map to the names used in the other map. The segment reference names (in the form of on-street from-street to- street) of roadway segment database generated in step one were then converted to the names used by the more accurate map. The third step mapped the segments created and modified in the previous two steps into the more accurate map through a mapping program developed by the Maricopa County Department of Transportation. This mapping process resulted in a new coverage with both desirable topological accuracy and functional classification information. This also resulted in a smaller coverage in which each arc represented one segment homogenous in both ownership and functional classification. The forth step added the physical features along defined segments. Data collection personnel field checked the roadway physical features and the unpopulated database features were manually coded into the GIS database.
User Information Management SystemsFor the White House/Washington Mall and Mount Rainier projects, a GIS based sign inventory/management system was created to query sign attribute information, including sign location and sign images, on a base map. The management system allows the user to easily pan and zoom into areas of the coverage and display the attribute information and a picture. For the Mount Rainier project, a PC database management system was also created which allows the user to access and view the attribute information and issue maintenance work orders. For the HPMS project, Lee Engineering developed a user interface for maintaining and reporting the complicated and voluminous HPMS data. The main functions are a) geographical editing of the HPMS database. b) generating various roadway feature maps. c) reporting data in user-defined formats as well as the format required by the Federal Highway Administration. d) calculating vehicle-miles of travel for various boundaries. Figure 1 shows a screen of the HPMS interface system.
GPS AS TRAFFIC DESIGN AIDGIS and GPS are generally used as tools for inventory management systems. However, Lee Engineering recently experimented with integrating the traffic sign data captured through GPS with CADD for signing design. For a 17-mile long signing project, Lee Engineering used GPS to collect such sign information as sign dimensions, legends, sign codes, post types, number of posts, orientations and sign locations. Following differential correction, the GPS data (including the location information in the state plane system) were put in a dBase file.
In MicroStation, a MDL (MicroStation Development Language) program was developed to read this database table and place in the CADD file the cells representing signs on post or on posts and rotate the sign in the correct orientation. The program at the same time made the "hot link" between the database records and the graphic cells in the drawing. The "hot link" allowed the designer to make modifications to the database and have these changes automatically updated in CADD files. Another MDL program was developed to display the desirable attributes such as post ID and sign code for each individual sign. Any design changes, such as placement of new signs, removal or relocation of existing signs, were made both to the drawing and to the database. By maintaining a current sign database, generating construction cost estimates and signing summary design sheets became a more efficient and accurate process. For one striping design project, Lee Engineering experimented using GPS to develop as-build road geometry when as-built plans were not available and where field survey was difficult. The roadway geometries for this project consisted of two horizontal curves connected by a very short tangent coming into a T-intersection. The existing geometry was further complicated by the partial widening of the pavement both approaching and leaving the intersection. A GPS unit capable of real time correction was used to pin-point the locations of a series of "control points" along the edge of pavement. The data, when imported into MicroStation, showed a very satisfactory result. Figure 2 shows the GPS data imported into MicroStation. The blue points represent the real-time corrected GPS data and the red points represent the post- corrected GPS data.
DISCUSSIONS AND CONCLUSIONS
GPS Accuracy and GPS Signal ReceptionLocation data collected through GPS units have intrinsic random errors that cannot be totally eliminated. To obtain the higher level of accuracy for any receiver requires differential correction, a process of placing a receiver on a known location, called a base station, and using the collected satellite data to adjust GPS positions computed by other receivers at unknown locations during the same time period. The accuracy after differential correction in, Lee's experience, has been in the range of 2-5 meters. For most of Lee's applications, the level of accuracy seems to be adequate. However, the availability of base station data for post-differential correction has been a major issue. In the travel speed study, a few test vehicle runs had to be discarded and additional runs made because the base station failed to operate properly during the time the GPS data were being collected. Bad GPS signal reception results in missing data. This has been another major issue. The GPS signal broadcast by the satellites are susceptible to being blocked of tall buildings, bridges and even trees. Another type of reception error is due to unfavorable satellite configurations at particular times and particular locations of the GPS receiver. In the White House / Washington Mall and Mount Rainier projects, GPS reception errors made manual methods a necessary backup. It should be noted that through improvements to hardware and software algorithms, accuracy is increasing to 1 meter level for non-survey level GPS units. The potential reception problems are also being reduced.
GPS vs Manual Data CollectionDue to GPS reception problems mentioned above, it was necessary to collect some attribute information using manual methods in the White House/Washington Mall project. An analysis was conducted which compared the amount of time required using both methods, GPS and manual, to collect and post process the sign inventory data. It was found that signing data collection and post-processing with the GPS units required 4.6 man-minutes per sign verses 5.5 man-minutes per sign using the manual method. The signs collected using the GPS equipment required a longer data collection time but were post-processed with much less time. The longer field data collection time using GPS was due to the fact that 180 positions were collected at each sign location in order to obtain the best accuracy of the GPS equipment used for the task. At the rate of one position per second, GPS data collection required a minimum of 3 minute at each sign location. The full description of the analysis can be found in "Comparison of two Sign Inventory Data Collection Techniques for Geographic Information Systems", published in Transportation Research Record No. 1429, Planning and Administration, Multimodal Priority Setting and Application of Geographic Information Systems, Transportation Research Board, National Research Council. Three points should be noted:
- The improvements to GPS hardware and software have made GPS even more efficient than manual data collection. Not only does real-time differential GPS significantly reduce the post process time by eliminating the need for post-differential correction, it requires much less data acquisition time as well. One second per position with real-time differential GPS achieves the same level of positional accuracy as 3 minutes per position using non real-time differential GPS.
- While the above comparison was only made for traffic sign data collection and post-processing, it is felt that many aspects of the comparison are applicable to others types of data collection and processing.
- Other factors, such as equipment cost, personnel cost and availability should also be considered.
GPS Data Post Process and Base Map AccuracyHowever accurate the GPS data may be, they represent a discrete set of two or three dimensional points, But historically, the majority of transportation data have been in a linear model. The points only make sense in the context of a linear roadway network. The dynamic segmentation functions of GIS are often used to convert discrete GPS points to points referencing continuous line segments, i.e., the roadway network. While the dynamic segmentation functions have become standard features for most GIS packages, the accuracy of dynamic segmentation results rely very heavily on the accuracy of the base map. Unfortunately, generation and maintenance of accurate base maps require huge resources and therefore are not often available. Lee Engineering has been fortunate to have had access to a fairly accurate Maricopa County map for most projects. This road network map was digitized from 1:24,000 scale aerial-photographs in 1992, through a joint effort of the Maricopa County Department of Transportation and the Arizona Department of Transportation. The map claims a relative accuracy of 25 feet. Most of the TIGER-based maps have a reputation for having incomplete and inaccurate topological and attribute information. It is not recommended to use TIGER map to map GPS points in most engineering applications; however, TIGER maps can be adequate sources for many transportation planning applications. The results of using TIGER files for the roadway needs study were very satisfactory.
Truth in GIS DataWith GIS being the information integrator of data from different sources and different types, more than ever before, agencies are relying on each other for data. As a result, inaccurate data, together with accurate data are being spread across disciplines very easily and very quickly. Unfortunately, the quality and availability of data has been one of major obstacles in the use of GIS. Some of Lee's experiences indicate that the most difficult aspects of GIS are not necessary at the technical level, but at the data acquisition and validation level. Lee's experiences in using GIS/GPS technologies have been successful. It is Lee's believe that the GIS/GPS technologies are valuable tools that can help bring quality and efficiency to our profession. However, knowing the right technologies for the right applications, and knowing what is involved in using these technologies are the precursors to success.