USE OF STATE FLEET VEHICLE GPS DATA FOR TRAVEL TIME ANALYSIS David P. Racca Center for Applied Demography and Survey Research (CADSR) University of Delaware Graham Hall, Rm 284, Newark, Delaware 19716 Phone: (302) 831-1698 E-mail: dracca@udel.edu
State of Delaware Public Vehicle GPS Data About 2400 vehicles broadcasting location every two minutes. 2 million point measures per month Providing 5 to 10 million travel way measures per month Does not include public safety, transit buses, or road maintenance vehicles. 40% passenger cars, 34% passenger vans, 23% pickups and SUVs Data as far back as year 2007
An Opportunity to Develop a Statewide Travel Time / Speed Survey Besides safety measures, travel time and speed throughout the day is the most important performance measure Addresses roads large and small Collection costs already covered Includes detailed trip data allowing for analysis of turning movement statistics
State Vehicle GPS Measurements
State Vehicle GPS Measurements Sample Trip
Northern New Castle County Weekday Observations 2012 to 2013
Sussex County Weekday Observations 2012-2013
Hourly 8am Observations
Processing Capture historical GPS data by querying Networkfleet web services for Delaware vehicles. Process GPS XML response data Create GPS point databases and GIS files Extract and associate GPS points with particular trips taken through time Build a trip and link based version of the GPS data - Estimate the path taken between GPS readings - Associate speed measures with a particular road link, direction, and tuning movement Screen the data for errors and anomalies.
The primary issues associated with traffic data and managing it and processing it in GIS are: Travel flow. Most of the data is directional Standard and effective ways of referencing traffic data to models of the transportation network Integrating data of different spatial types, point, line, and polygon. Examples: speed probe data, capacity data, device counters, travel demand Integration of data from different time dimensions. Dealing with large volumes of data. Aggregation and disaggregation. Integrating across various portions of the transportation network.
Desired Features For Referencing Traffic Data is related to an established standard can reference the smallest portions of road as well as the largest is not dependent on a particular cartographic source can relate data from various sources and measurement schemes can be generalized to relate information about small and large road segments can capture the direction of traffic flow. Traffic data for a particular portion of road is directional Provide a fixed identifier for use by those who cannot work with advanced DBMS or linear referencing systems (route and mile point)
Identification illustration To relate a measure to a particular turning movement a S, L, R, or U is appended to the LRSID, for example. Left turn from Sudlers Row LRSID = 0006160176000000L
Sample Output Weekday Hourly S Straight or Thru Shown Also available are Right & Left Segments statewide included
Example Detail Captured for Road Links Routing network is segmented at every major or minor intersection
Comparison of Segment Length, VehGPS vs NPMRDS State Vehicle GPS Data NPMRDS Data
Comparison of Coverage, State GPS vs NPMRDS For State GPS Weekday 2012-2013, NPMRDS in Black
Comparison of Coverage, State GPS vs NPMRDS For State GPS Weekday 2012-2013, NPMRDS in Black
Summary of Features of the State Vehicle GPS Data available for up to 6 past years Wide coverage, data for small and large roads Captures speeds and travel times relative to turning movement Measures available at great detail, road link breaks at all intersections, large and small Delay at intersections by turning movement, incorporated into road link speed / travel times. Ideal for generation of time sensitive routing network impedance. Cost of collection covered in existing program
Aggregations Over 100 million measurements each year available for very detailed road segments throughout the State creates a large data set that requires aggregations to examine conditions with respect to various factors of interest. These factors include: Time of day intervals, i.e. 30 minute, 60 minute intervals Periods of the day, AM Peak, Midday, PM Peak, Evening Day of week Season, i.e Summer or Non-Summer Holiday, non-holiday Year
Aggregation By Road Segment
Calculation of Free Flow Speed as the 75 percentile Of Hourly Averages (just major roads shown)
Calculation of Percent Degradation at 8am, weekdays Percent degradation = 100 * (freeflow75 speed) / freeflow75 Calculated from weekday hourly averages
Calculation of Percent Degradation at 8am, weekdays Percent degradation = 100 * (freeflow75 speed) / freeflow75 Calculated from weekday hourly averages
Travel Time Reliability
Intersection Study Left turns that are most effected ( > 40% degradation) by morning (8am) congestions
Other potential applications Establishment of a statewide routing network populated with DOW, time of day, impedances Before and after studies, land use and facility changes Examining delay at intersections Estimations of capacity and studies of volume speed relationship Relating traffic flow to land use and travel demand Multimodal studies Applications of a detailed time sensitive routing network, such as accessibility studies
Some observations Little experience in general working with this kind of data. Capabilities with huge amounts of traffic data are often lacking. the data accurate? Can we trust it? What accuracy do we need? How does it compare to other sources? Number of measures Resolution Time of Day, Day of Week, Season Turning movement Different data sets may measure differently Value depends on intended use. Corridor performance? congestion at intersections, effects of land use?
Blue Tooth Locations
Blue Tooth Travel Times by Hour of the Day Station 3 to Station 4, Sussex County
Blue Tooth Compared to Fleet GPS Estimates
Other CADSR Work Travel surveys Development of internet mapping and data query Population, employment, and housing projections and allocations Accessibility studies Markets for transit Environmental and land use studies Place and address files Network modeling and routing