Background
The goal of our project is to develop information that facilitates more informed planning, better urban design and healthier life styles through objective measurement and analysis of urban greenway trail use, surrounding physical and socio-demographic environments and specific trail characteristics.  Few tools are available for forecasting use of recreation facilities like trails. We previously reported models that explain 80% of the variation in daily trail traffic. These models incorporate weather, time (month and day of week), socio-demographics and neighborhood urban form, but do not include measures of specific characteristics of individual trails. In the current research, we use satellite imagery and GIS to extend our previous analyses and examine the relationships between trail use and the physical characteristics of the trails themselves.

Objectives
Our primary objective is to specify and estimate a robust, theoretically complete model of trail use. To achieve this objective we identify elements of the trails or the physical environment immediately adjacent to them that correlate with trail traffic. An additional objective is to demonstrate the feasibility of using LIght Detection And Ranging (LIDAR) technology to develop objective measures of the visual environment that have implications for the planning, design, and management of trails for recreational and utilitarian physical activity.

Methods 
Daily Traffic Counts and Measures of Trail Neighborhoods: We are monitoring trail traffic 24 hours per day, seven days per week at 30 locations on five multiuse greenway trails in Indianapolis using Trailmaster© infrared monitors. Monitors are located approximately one mile apart to provide coverage of the entire 33-mile trail network. Our dataset includes results from monitoring at four locations on one trail from February 2001 through July 2005; two locations on a second trail between June 2002 and July 2005; and 24 locations on five trails between May 2004 and July 2005. The sample size for our estimates of daily trail traffic, the dependent variable in our model, is more than 15,000. Measures of independent variables (i.e., socio-demographic characteristics and urban form of trail neighborhoods) are developed from U.S. Census data and other sources using GIS.

Measures of Viewsheds and Other Trail Characteristics: We use GIS, aerial and satellite imagery and GPS field data to map variation in a variety of physical characteristics along the trails including: viewsheds, sinuosity, slope, greenness, surface type (paved, gravel, etc.), presence of art, water features, public facilities and land use. LIDAR data are used to develop three-dimensional surface models of the landscape along the trails. Viewshed analysis is implemented on the surface models to delineate areas visible from each trail segment, each of which is associated with an infrared traffic monitoring site. Variation in the physical environment within each segment’s viewshed is quantified. We use standard bivariate and multivariate models to explore how trail viewsheds and other physical characteristics correlate with trail use.

Results
Our results include additional counts of daily trail traffic, new measures of trail viewsheds and other trail characteristics and a new, more complete model of trail traffic that incorporates characteristics of trail segments in addition to neighborhood characteristics.  We report results for individual environmental variables and use model results to determine the relative importance of neighborhood and trail-specific variables with respect to variation in trail traffic.

Conclusions
We discuss the methods and technology used to develop our environmental measurements, assess challenges to generalizing them, and make suggestions for alternative approaches that incorporate next-generation spatial information technologies. Our results will inform planning and health policy by providing objective analysis of the characteristics of existing trails that are significantly related to use. Our results also will provide important information for new trail design and development by increasing the ability of analysts to estimate potential use of a proposed trail based on neighborhood-level and trail-specific characteristics.

Support
This research is supported by a grant from the Robert Wood Johnson Foundation Active Living Research Program.