Background and Purpose
Background: Managers of urban multiuse trails historically have not measured traffic volumes or miles traveled by users of trail systems. With increased budgetary pressure and demands for accountability, trail managers need consistent information about trail use. Transportation planners and engineers have developed systematic procedures for monitoring motor vehicle traffic and estimating average daily traffic and miles traveled on road networks. These measures are used for a variety of purposes, including allocation of resources for capital improvements and maintenance. In this paper, we illustrate how local planners and engineers can estimate annual average daily traffic and user miles traveled on urban trail networks by implementing a coordinated monitoring system that includes a small number of continuously monitored reference locations and systematic short-duration counts on all trail segments within a network.

Methods
Continuous counts of non-motorized traffic were collected from 2011 at 6 locations on the off-street trail network in Minneapolis, MN. Using these data we developed a new approach – use of day-of-year factors – for estimating AADT from short-duration counts. In year-2013 we deployed 6 mobile counters (in addition to the 6 reference site monitors) to estimate non-motorized traffic on the entire off-street trail network (~80 miles). We collected short-duration counts (i.e., 1-week) at 78 locations and subsequently estimated (AADT) and User Miles Traveled (UMT) for each trail segment. We then use these estimates to map trail traffic and explore relationships between trail traffic and neighborhood design.

Results
We successfully deployed a non-motorized traffic monitoring program on the off-street trail network in Minneapolis, MN. We have 4 core results that may be useful for developing non-motorized monitoring programs in other areas:

• A system of short-duration and reference site measurements can yield spatially precise performance measures of trail traffic for an entire trail network.
• Day-of-year scaling factors have smaller error than the standard method used by transportation agencies (day-of-week and month-of-year) in estimating AADT, especially from shorter duration (<1 week) counts.
• Extrapolation error decreases with the length of the short-duration counts, with only marginal gains in accuracy with counts longer than one week.
• Error in estimating AADT is lowest when short-duration counts are taken in summer (or spring-summer-fall) months (April-October) in Minneapolis, MN.

Policy implications
Trail managers can develop performance indicators comparable to those used routinely in planning for motorized traffic networks. We were able to estimate annual traffic for an urban trail network (~80 miles) using a relatively small number of monitors (n=12) in 7 months. Spatially precise information on trail traffic may be helpful to policy-makers interested in planning for active travel. Analysts can use day-of-year factors to increase accuracy of estimates of AADT.

Next Steps
Our next steps include working with additional communities in the metropolitan region to develop regional performance measures.