Michael Flaming by HC1112140287

VIEWS: 6 PAGES: 9

									Michael Flaming
Project Proposal
GIS in Water Resources
Fall 2005

Evaluation of GIS Basemap Accuracy for GPS Tolling

Introduction
Electronic toll collection technology has advanced to the forefront of the discussion on highway
finance as state and federal policy makers seek a stable revenue source for funding maintenance
and construction of the nation’s highways. Construction costs have escalated and the demand for
safe and efficient highways has steadily increased. Highway capacity has increased
insignificantly over the last 30 years while at the same time vehicle miles traveled have increased
dramatically. Furthermore, our 40 to 50 year old interstate and state highway system is in dire
need of rehabilitation. Tax payers and politicians alike have refused an increase in fuel taxes for
fear of slowing the economy or causing political windfall. Increasing cost coupled with
decreased revenue per mile driven has caused policy makers to move towards mileage based user
fees rather than indirect fees in the form of a gas tax and registration fees.

A GPS based tolling system relies on digital maps rather that overhead gantries or embedded
loops to determine a vehicles’ location and miles traveled for the purpose of assessing a charge.
There is no infrastructure on the ground other than what is used for enforcement; the basemaps
and GIS systems serve as virtual cordons where fees are assessed. Therefore, systems must be
thoughtfully developed and extensively tested to ensure their reliability and accuracy.

State departments of transportation maintain digital basemaps of the center lines of all of their
highway system. Most have been digitized from existing USGS 7.5-minute quadrangles maps.
The horizontal accuracy of these maps cannot be expected to exceed National Map Accuracy
Standards for 1:24,000 scale maps (+/- 40 feet or 12 meters). In fact, one would have to assume
some error in the digitizing and processing of these maps, so the accuracy would be even less the
USGS standard. TXDOT has just recently developed a second generation of digital maps that are
based on aerial imagery rather than USGS paper maps. The imagery has a published accuracy of
+/- 4 feet and a scale of 1:3000. Based on the experience of TXDOT GIS users, the accuracy of
the map is on the order of +/- 5 to 10 meters.
Aerial Photo of west Austin.
GIS features digitized from aerial photo

Objective
The objective of this project is to use GIS to evaluate the accuracy of TxDOT’s digital highway
basemaps for potential use in a GPS based road user charging (tolling) system. The particular
aspects of different tolling scenarios are left out in this report; the focus is on the GIS.

The accuracy of the TXDOT map is considered unknown in this project. In order to find the
accuracy, line elements of the TXDOT map are compared to GPS lines that were collected over
the summer. In reality, the GPS lines also contain some error but they are nonetheless considered
as the gold standard for this project.

      Objective is to compare digital maps against more accurate standard
      Accuracy of digital maps on the order of 5-10m
      Accuracy of differentially corrected GPS on the order of < 0.5m

Data
The main data source used in this project is the basemap provided by TxDOT’s Information
Services Division (ISD). This map includes centerlines of all state highways as well as
centerlines of each roadbed. Accuracies are in the range of ±5 to 10m. The maps were
constructed from aerial photography that has an accuracy of +/- 4 feet.

Over the summer I collected highly accurate differentially corrected GPS data to evaluate the
accuracy of the basemaps. I selected a range of highway types around the Austin area to
determine if there area significant differences in their accuracy. The test routes include I-35 both
on the north and south sides of Austin, all of Loop 360, a portion of north Mopac, 2222, portions
of US 183 both in and out of Austin and FM 163 outside of Lockhart.


Project
To analyze the accuracy of the line elements, a simple buffer analysis was used. A line string
buffer was applied to the GPS data to see if the line elements of the TXDOT map fall within the
range. Buffers of 1, 2, 3, 4, 5, 7, and 10 meters was used. In previous similar testing done on
USGS based maps the buffers selected were much larger, from 40’ to 150’. Statistical analysis of
the data was used to determine a measure of the linear accuracy of the digital maps.

The steps to compete these objectives are:

      Select test routes
       Urban Interstate Highway (I-35)
       Urban State Highway (US 183, Mopac)
       Rural State Highway (US 183 Lockhart)
       Loop 360, 2222, FM163
      Collect high accuracy GPS data
      Actual road data compared to same road segments of digital maps
      Buffers used to compute map coverage ratio
       Buffer represents range in which results are accurate

For a given coverage ratio, the larger the buffer the lower the accuracy
For a given buffer, the greater the coverage ratio, the higher the accuracy

The following images demonstrate the process of buffering and intersecting. First a buffer is
created around the GPS data. Next an intersect command is carried out and the remaining line
segments give a measure how much of the TXDOT line elements fall within the buffer.




GPS line with a 3 m buffer
Resulting intersect with TXDOT centerlines




Close-up image of 3m buffer around GPS data.
Resulting intersection of buffer.

Once all the buffers and intersects for each GPS sample had been calculated, all the bits and
pieces of line segments associated with different buffers had to be analyzed. This is where a
problem arose. The attributes of the line intersect that was created was tied to that of the original
line element in the TXDOT map. The lengths given were the original lengths, not the
accumulated lengths of line portions. To address this problem, a field was added to each of the
attribute tables called New_Length. A field calculation was performed using a VBA script to
carry out the calculation.
VBA Script for accumulated line segment length

After the calculation had been performed, the New_Length field was populated with the new
lengths of line segments for each buffer.




Attribute table of line intersect
After the new lengths of each line were calculated for each roadway and each buffer setting, the
data was exported to MS Excel for analysis. This included 9 total attribute tables, one for the
original TXDOT file, and 8 for each of the buffers. The line segment lengths associate with each
buffer were added up and divided by the original length to arrive at a “coverage ratio” for each
buffer. The coverage ratio is considered the percent of the TXDOT centerline that are contained
within each buffer. For example, 30% of the centerline of Loop 360 is contained within the 1
meter buffer as shown in the graph below.




                                            Basemap Accuracy


                     1.00

                     0.90

                     0.80

                     0.70
    Coverage Ratio




                     0.60
                                                                                               LP360
                     0.50

                     0.40

                     0.30

                     0.20

                     0.10

                     0.00
                            0   1   2   3     4       5        6   7     8       9      10
                                                  Buffer (m)

Graph coverage ratio for Loop 360

Conclusions
When all the data for various roadway types are plotted out on a single graph, it becomes clear
that the accuracies are independent of roadway type. The graphs for each roadway generally
follow a similar trend and lend well to a regression analysis. Based on my findings, it can be
concluded that the linear elements of the TXDOT basemap have an accuracy of at least +/- 7
meters assuming a confidence of at least 90%. These results are significantly better than similar
testing done on the digital maps that were based on USGS 7.5 minute paper maps.
                                                    Basemap Accuracy


                     1.00


                     0.90


                     0.80


                     0.70

                                                                                                   LP1
    Coverage Ratio




                     0.60                                                                          LP360
                                                                                                   I-35
                     0.50                                                                          US183 Urban
                                                                                                   2222
                     0.40                                                                          FM672
                                                                                                   US183 Rural
                     0.30                                                                          Mean
                                                                                                   Upper Bound
                                                                                                   Lower Bound
                     0.20


                     0.10


                     0.00
                            m   1m   2m   3m   4m        5m        6m   7m   8m   9m     10 m

                                                      Buffer (m)




Recommendations
These conclusions are only good for maps that are based on aerial photos with accuracy of +/- 4
feet. Statewide, maps may not be constructed to the same accuracies especially in rural areas.
Further testing needs to be done to fully evaluate the accuracy of digital maps statewide.

New technologies are continuously being developed to capture linear features digitally. As
requirements for more accurate data emerge, it may be necessary to create and evaluate maps
with sub-meter accuracy. This is possible, but of course the costs and benefits would have to be
analyzed.

References:
Max Donath, Pi-Ming Cheng, Shashi Shekhar, Xiaobin Ma. A New Approach to Assessing Road
User Charges: Evaluation of Core Technologies. Minnesota Department of Transportation Office
of Research Services. September 2003.

Edward J. Regan. Moving Off the Gas Tax: Implications for the Toll Industry. Tollways IBTTA.
Autumn 2004.

								
To top