The raw data comes from the historical traffic data warehouse of the INRIX Smart Dust Network. Since 2006, INRIX has acquired billions of discrete “GPS-enabled probe vehicle” reports from commercial fleet vehicles – including taxis, airport shuttles, service delivery vans, long haul trucks – and cellular probe data. Each data report from these GPS-equipped vehicles includes at minimum the speed, location and heading of a particular vehicle at a reported date and time.

INRIX has developed efficient methods for interpreting probe vehicle reports that are provided in real-time to establish a current estimate of travel patterns in all major cities in the United States. These same methods can aggregate data over periods of time (annually in this report) to provide reliable information on speeds and congestion levels for segments of roads. With the nation’s largest probe vehicle network, INRIX has the ability to generate the most comprehensive congestion analysis to date, covering the nation’s largest 100 metropolitan areas.

The US Census Bureau definition of Core Based Statistical Areas is used to define metropolitan areas. This report uses the latest 2007 census estimates to identify the top 100 areas.

This report focuses on the major limited access roads in the top metropolitan areas in the United States. In all of its products, INRIX utilizes an emerging industry convention known as “TMC location codes” developed and maintained by the nation’s leading electronic map databases vendors, including Tele Atlas, to uniquely define road segments. The typical road segment is the interchange and the portion of linear road leading up to the interchange across all lanes in a single direction of travel. The length of a segment will depend upon the length of the distance between interchanges. For this report, over 47,000 road miles in nearly 31,000 discrete road segments have been analyzed.

The focus of this report is the calendar year 2007. In some cases, calendar year 2006 data is utilized to enable year over year comparisons.

There are two key building blocks for the different analyses included in this report:

• Reference speed (RS): For each road segment, all probe vehicle reports obtained in overnight hours (where congestion is usually unlikely) in 2007 are analyzed. The 85th percentile of those data points is identified as the “reference speed” for that particular road segment. This is typically the speed of “free flow” traffic if and when no congestion exists. Each segment has a single reference speed.
• Hourly average speed (HS): All probe vehicle reports for each road segment are grouped by hour of day, day of week (e.g. Monday from 3 to 4pm) and an “average speed” for each time slot is established for each road segment. Thus, each segment has 168 corresponding hourly average speed values – representing 24 hours of each day times the seven days in a week.

To assess congestion over a metropolitan area, INRIX utilizes several concepts that have been used in studies.

• Travel Time Index (TTI): TTI is the ratio of peak period travel time to free flow travel time. The TTI expresses the average amount of extra time it takes to travel in the peak relative to free-flow travel. A TTI of 1.3, for example, indicates a 20-minute free-flow trip will take 26 minutes during the peak travel time periods, 6-minute (30 percent) travel time penalty. For each road segment, a TTI is calculated for each hour of the week, using the formula TTI = RS/HS.
• “Drive Time" Congestion: To assess and compare congestion levels year to year and between metropolitan areas, only “peak hours” are analyzed. Consistent with similar studies, peak hours are defined as the hours from 6 to 10 am and 3 to 7 pm, Monday through Friday – 40 of the 168 hours of a week.
• For each Metropolitan Area, an overall level of congestion is determined for each of the 40 peak hours by determining the extent and amount of average congestion on the analyzed road network. This is easy to compute once TTI’s are calculated for each segment:
  • Step 1: For each of the 40 peak hours…
  • Step 2: All road segments analyzed in the CBSA are checked, each segment where the TTI > 1 is analyzed further…
  • Step 3: With each segment contributing a congestion factor that is, the average congestion (the amount the TTI is greater than 1) multiplied by the length of the segment
  • Step 4: For a given hour, overall metropolitan congestion is the sum of these congestion factors.
  • Step 5: To establish a Metropolitan Travel Time Index, this metropolitan congestion factor is divided by the number of road miles.

With the unique ability to examine in detail nearly 31,000 urban highway road segments, INRIX identifies the specific locations in each metropolitan area – and can compare locations across the country – that are consistently congested. These are “Bottlenecks. ”

Congestion – and how to measure it – can be in the eye of the beholder. Is congestion defined as how bad a road segment is at its worst or is it how often the segment gets “congested” (and what is the threshold for “congestion” anyways – tapping the brakes, stop and go conditions, etc.)? INRIX has developed a method that combines both the amount of time a road segment is congested with the intensity of congestion during those periods. The process used to analyze each of the nearly 31,000 road segments is as follows:

• The same RS and HS values are utilized as in the overall congestion by metropolitan area portion of the study;
• All 168 hours of the week are considered, not just the 40 “peak hours.” As will be evident in the data, severe Bottlenecks aren’t just limited to peak hours;
• For each hour of the week that the average speed is less than 50% of the reference speed, the hour is considered “congested;”
• For all “congested” hours, the average intensity of the congestion is determined by establishing an average travel time ratio;
• The total Bottleneck factor equals the number of hours of congested by the average travel time ratio.
• Each road segment’s Bottleneck factor can be compared with others in a metropolitan area and against all Bottlenecks nationally. It can also be compared year-to-year, as we will do going forward.