Traffic Count Learning Curve
Although the CUP did not require Westmont to have permanent counters installed until the beginning of Phase I construction, we chose to use the time before that deadline to get experience managing daily trips by means of temporary counters. We purchased and installed three state-of-the-art tube counters in time for the beginning of the 2007-08 school year.
In the Fall of 2007, data collection went smoothly. As depicted on a chart of the daily trip patterns, traffic on Cold Spring Rd—and our capture of the data recording it—was quite regular and consistent.
In the Spring of 2008, we encountered several problems in collecting data. The chart for that period shows a number of days on which a variety of glitches interfered. Each of the four occasions when the Point 1 counters were down is described below.
Note that we often have two units counting simultaneously at the site below the college gates. This allows us to confirm the validity of the counts by comparing data from two machines. Where the counts differ between them over the course of a collection period (and they do, for numerous reasons [see footnote]) we have used the series with the highest of the two counts in calculating the averages.
In order to corroborate the accuracy of our units, in November we put in place a similar unit owned and calibrated by Associated Transportation Engineers, and compared its data against data from all three of our units. The test confirmed that our units are capturing counts within normal specifications.
Causes of Missing Data
Episode A (Jan 17-22)
On 1/17/8, while one counter at Point 1 was out for routine servicing (data offload and battery recharge), a maintenance worker inadvertently failed to properly restart the other counter at the paired count site, so we did not capture counts for these six days.
Episode B (Feb 15-21)
When the heavy rainfall in late January flooded two of our units, we were surprised to discover that the device housings are not waterproof. Although the counters continued to count normally for a few days, by some time in February water inside began to degrade the mechanism and the capture of data became intermittent.
Episode C (Mar 6-24)
One of the units exceeded the counting capacity of its memory chip. Unfortunately, this occurred about a week before the person responsible for harvesting the data left town on business for twelve days in March, resulting in the extended lapse shown. The backup unit was offline during this period.
Episode D (Apr 2-5)
In April, the battery ran out on one of the units a few days before it was next checked. Again, this occurred while the backup unit was down for routine maintenance.
Point 2 (above campus) Counter
For reasons not yet fully discerned but including most or all of the above, the counts of neighbors’ trips above La Paz Rd were found to be unreliable for several extended periods during the spring semester.
Substituted Data and its Implications for Cumulative Averages
For days on which data was missing, we have used substitute data. In doing so, we applied alternate data that results in higher daily counts chargeable to Westmont and thus, overall, higher CSR ADTs. Specifically, in light of the formula for counting Westmont's daily trips we have chosen to apply higher than expected values for Point 1 missing data, and typical or lower values for Point 2 missing data.
For Point 1 missing data, we substituted the highest raw daily count at Point 1 for the corresponding day of the week in the entire Spring semester (excluding special event days). For Point 2 missing data in the Spring semester, we substituted the average daily data from the Fall semester for the corresponding day of the week. (For the Spring Break period included in the Episode C timeframe we filled in for those missing days the highest daily values found in the Christmas-holiday extended break.) We have no reason to believe that our neighbors would generate markedly different daily traffic in the Spring than in the Fall.
For each day with missing data we checked the campus calendar to see if adjustments for special events on campus should be applied. Fortuitously, special events did not occur during any of the "Episode" periods.
Lessons Learned & Applied
Based upon manufacturer specifications and on our experience in the fall, we had initially determined that a monthly maintenance period would suffice. However, starting in April 2008, we have
- increased our vigilance in monitoring the temporary counters
- reduced to the minimum the down-time associated with re-charging the counter batteries (each needs to be taken offline for about a day every time it needs a battery recharge)
- ensured that both counters at Point 2 are kept in place and functional except when one of the two is routinely cycled out for maintenance
We look forward to installation of our permanent counters. Because they will be set up with a direct utility power source (with backup batteries) we do not expect power loss to impede the counts. And because data collection can be done wirelessly at the counter itself we will not need to take them offsite for routine maintencance. Also, the wireless connection allows for very quick, positive and frequent confirmation of the system status, so that whenever the system encounters an aberration we will thus be able to respond quickly to diagnose and repair it. We have purchased a complete third device to have on hand for backup, ready to swap in if either of the two main devices were to fail.
We're glad for this "shake-down" year to learn what to watch for in order to sustain the highest possible community confidence in the college's count of traffic on Cold Spring Road.
* - According to TimeMark, manufacturer of the Gamma brand devices, a variety of factors affect the data capture such that two identical units placed side by side will never yield identical counts. These factors include:
- the angle at which vehicles strike the tubes
- the affects of a northbound car striking one tube simultaneously with the strike from a southbound car (they can't both strike two tubes simultaneously