Journal of Maps, 2008, 108-118
Viewshed characteristics of urban pedestrian trails, Indianapolis, Indiana, USA JEFFREY WILSON1 , GREG LINDSEY2 and GILBERT LIU3 1 Department
of Geography, School of Liberal Arts, Indiana University - Purdue University Indianapolis, 425 University Boulevard, Indianapolis, Indiana, USA 46202;
[email protected] 2 School
of Public and Environmental Affairs, Indiana University - Purdue University Indianapolis, 801 West Michigan Street, BS 3027, Indianapolis, Indiana, USA 46202; 3 Children’s
Health Services Research, Indiana University School of Medicine, 410 West 10th Street, Indianapolis, Indiana, USA 46202; (Received 8th March 2008; Revised 24th May 2008; Accepted 27th May 2008)
Abstract: The map accompanying this brief report depicts spatial variation in viewshed characteristics of urban pedestrian trails in the city of Indianapolis, Indiana, USA. Visual openness and visual magnitude were modeled for approximately 50 kilometers of the Indianapolis Greenway Trail System using geographic information system and light detection and ranging (LiDAR) technologies. Viewshed data also were integrated with high resolution satellite imagery to estimate greenness perceived by trail users. The primary map was designed at scale of 1:32,000 and shows visual magnitude overlaid on aerial photography and LiDAR surface height layers. Inset maps illustrate viewshed integration with high resolution satellite imagery to estimate perception of greenness. We have reported statistically significant relationships between pedestrian trail traffic and viewshed characteristics along trail segments as estimated with multivariate regression models that control for other variables, including weather variation, month, day of week, and social and physical attributes of surrounding neighborhoods. In previous studies we also found higher satellite-based estimates of greenness were associated with improved self-reported perception of neighborhood walking environments by children and their families. We note here both current limitations of existing technologies and potential for emerging technologies to improve measurement of pedestrian viewshed characteristics.
108 ISSN 1744-5647 http://www.journalofmaps.com
Journal of Maps, 2008, 108-118
1.
Wilson, J., Lindsey, G. & Liu, G.
Introduction
Urban pedestrian trails offer alternative routes and support non-motorized modes of transportation within larger networks designed for and dominated by motorized vehicles. Trails can connect locations for utilitarian commuting and provide opportunities for leisure time physical activities such as walking, running, cycling, and skating. Proponents of urban trail development and improvement cite multiple benefits including beautification, positive environmental and economic impact, and potential to encourage physical activity in support of better public health (Frank et al., 2003; Saelens et al., 2003; Fabos, 2004). Analyses of social and environmental correlates of traffic on multi-use trails can inform planning, design, and management decisions. Trail managers can consider characteristics associated with the most heavily used portions of existing trail systems as potential features that, if added to other segments, could promote increased use. As a simple example, if trail managers observe more use on uninterrupted stretches of trail where users do not have to yield to vehicular traffic, then adding road overpasses or underpasses might be considered as a modification to increase use on other segments where trails intersect with vehicular traffic routes. Investments in trail infrastructure can be optimized if decision-makers consider social and physical attributes of neighborhoods surrounding trails, the places trails connect, and environmental characteristics perceived by users as they move through trail landscapes. Historically, efforts to describe trail-specific characteristics have relied on fieldwork and interpretation or rating of photographs (e.g. Pikora et al., 2002; Moudon and Lee, 2003; Reynolds et al., 2007), which can be costly and limited by low inter-rater reliability. More efficient methods of trail viewshed and landscape analyses are needed. Emerging forms of spatial data and analytical techniques provide opportunity to model visual characteristics of urban pedestrian trails more efficiently. Many cities are now being surveyed with high resolution light detection and ranging (LiDAR) instruments that provide detailed information on the three dimensional structure of urban landscapes (Priestnall et al., 2000; Lloyd and Atkinson, 2000). When combined with high resolution imagery and large scale GIS data, digital viewsheds can be constructed to represent visual properties of landscapes from pedestrian perspectives. We refer to this method as biophysical viewscape modeling because it integrates biological and physical approaches of environmental 109
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
remote sensing with viewshed analysis to estimate components of landscape perception. The accompanying map illustrates application of biophysical viewscape modeling for pedestrian trails in the city of Indianapolis, Indiana, USA. The main map shows visual magnitude overlaid on three dimensional surface data derived from an airborne LiDAR instrument. An aerial photograph mosaic is used as backdrop to aid the map reader with orientation and to provide context for the surface model and viewshed variables.
2.
Methods
A commercial contractor collected more than 3 billion LiDAR measurements across the city of Indianapolis in March and April 2003 using an aircraft-mounted OPTECH ALTM 2033 sensor. The sensor samples 33,000 points per second in a zig-zag pattern perpendicular to the line of flight beneath the aircraft. Data specifications report an average sampling density exceeding one point per square meter and accuracy of 15 centimeters in the horizontal direction and 30 centimeters in the vertical direction. First-return LiDAR measurements (the first laser pulse that returns to the sensor from a given location) were provided by the Indianapolis Mapping and Geographic Infrastructure System (IMAGIS) in tab delimited text format containing XYZ coordinates for each sampling point. Inverse distance weighted algorithms were used to interpolate a 1.5 meter (5 foot) resolution surface height model from the LiDAR points. A section of downtown Indianapolis is depicted using the LiDAR-derived surface height model in Figure 1. Trail centerlines digitized from high resolution (15 centimeter) aerial photographs were used to create 4,173 vertexes averaging 12.8 meters apart across 50 kilometers of the trail system. Viewsheds were generated at each vertex assuming an observer height of 1.8 meters above the ground and a maximum viewing distance of 805 meters (0.5 miles). We did not limit directional or inclination parameters of the viewshed because our purpose was to estimate general viewshed characteristics encountered by users traversing the trails in either direction. A viewshed is a delineation of landscape elements that are visible from a 110
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
Figure 1. Perspective view of LiDAR surface height model for a portion of the Indianapolis, Indiana central business district.
given vantage point. Viewshed algorithms were applied to estimate what can and cannot be seen from trails by modeling the interaction between lines of sight from each trail vertex and the three dimensional LiDAR surface model. The basic output of viewshed analysis is the binary viewshed - a map of visible and invisible areas from an observation point. Viewsheds generated from multiple adjacent observation points were combined to form a cumulative viewshed that models the visual environment encountered by a pedestrian traversing a route (Wheatley, 1995; Fisher et al., 1997; Llobera, 2003). Two other measures were derived from the binary viewsheds: visual openness and visual magnitude. Visual openness is a metric of the total visible area or percent area visible within the limits of the viewshed (Fisher-Gewirtsman and Wagner, 2003). For example, if we assume a viewshed from a single point is limited to a distance of 100 meters and there are no intervening terrain features to block lines of sight (e.g., trees, buildings, and terrain), the viewshed from that point would form a circle with an area of 31,416 square meters; the visual openness would be 100% because the entire area is visible. If, however, terrain elements block lines of site from the observation point so that only half of the total possible area is visible, then the visual openness at that location is 50% or 15,708 square 111
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
meters. Visual openness also can be estimated for multiple points along a route. Figure 2 presents a comparison of trail segments with low and high visual openness. For any given cell in the surface model, the number of observation points from which it was visible was summed to compute “visual magnitude” (Llobera, 2003). A visual magnitude of 0 indicates a location is not connected by a line of sight with any observation points along the trail. Low values of visual magnitude indicate portions of a landscape seen from one or relatively few observation points. High visual magnitude indicates areas that can be seen from many observation points. Figure 3 illustrates the derivation of visual magnitude for a portion of the White River Trail. Areas with high visual magnitude (shown as red in Figure 3b) can be seen more frequently by users traversing this trail segment compared to areas of low visual magnitude (shown as blue). In summary, visual openness gives an indication of how expansive the view is from a given trail segment, while visual magnitude is indicative of the interconnectedness of viewsheds encountered along the trail. Greenness within trail viewsheds was estimated using QuickBird satellite imagery (DigitalGlobe Corp.). The QuickBird sensor collects 2.4 meter multispectral and 61 centimeter panchromatic satellite imagery. We used a QuickBird image acquired on 23 July 2005 to compute Normalized Difference Vegetation Index (NDVI) values within trail viewsheds. NDVI is a well established method for estimating vegetation characteristics from multispectral imagery and has been shown to correlate with plant vigor, vegetative ground cover, and green biomass (Jensen, 2005). We estimated how pedestrians may perceive the greenness of the landscape as they move through the trail system by calculating the average NDVI within viewsheds from each vertex along the trail centerlines.
112
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
(a)
(b)
(c)
(d)
Figure 2. Comparison of visual openness for two segments of the Indianapolis Greenway Trail System. Cumulative viewsheds for a portion of the Monon Trail (a) and Pleasant Run Trail (b) are shown as overlays on aerial photography. Trail center lines are depicted with red lines and the semi transparent yellow overlay represents areas connected by lines of sight to observation points along the trail. Tall trees flanking this section of the Monon Trail contribute to a less open viewshed (c). (d) The photo illustrates the more open character of the viewshed along a section of Pleasant Run Trail, where lines of sight are less obstructed by vegetation.
113
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
(a)
(b)
(c) Figure 3. Example of visual magnitude for a segment of the White River Trail. (a) Aerial photograph with trail segment overlaid as red line. (b) Visual magnitude for the same area depicted using a color ramp typical of temperature maps; warmer colors (orange and red) indicate high visual magnitude, while cooler colors (blues and greens) indicate lower values. Areas where visual magnitude has zero values are transparent, allowing the aerial photography to be visualized. High values of visual magnitude occur on the White River because the trail is elevated and few tall trees block lines of sight to the water (c).
114
Journal of Maps, 2008, 108-118
3.
Wilson, J., Lindsey, G. & Liu, G.
Conclusions
In previous studies, we found that satellite-based estimates of greenness in urban neighborhoods were associated positively with higher levels of non-motorized trail traffic (Lindsey et al., 2006) and with more positive perceptions of neighborhood walking environments among children and their families (Liu et al., 2007). In later work involving multiple regression modeling, we also observed statistically significant relationships between trail traffic and visual magnitude, visual openness, and greenness after controlling for other variables, including weather, month, day of week, neighborhood socio-demographics and urban form, and trail segment characteristics (Lindsey et al., 2008). Our previous work used 30 meter spatial resolution imagery from Landsat satellites to derive greenness measures. The data associated with the map presented here extend our previous work by integrating high resolution satellite imagery from the QuickBird sensor to derive finer-scale estimates of viewshed greenness. Biophysical viewscape models can be extended to other applications where consideration of visual landscape characteristics is of interest. For example, our research group is part of an emerging multidisciplinary field that is applying advances in remote sensing, GIS, and spatial analysis to explore environmental correlates of physical activity and health (e.g. Troped et al., 2001; Sampson et al., 2002; Shonkoff, 2003; Sallis et al., 2005). In addition to health, our research has applications in the fields of transportation, recreation, urban planning, economic development, and, of course, geography. An advantage of the biophysical viewscape approach is the ability to estimate landscape perception at any location across large areas. Current limitations include discrepancies in scale and perspective of the satellite imagery compared to that of the human ground observer. New forms of photography and video (for example, oblique imagery sources becoming popular through internet providers such as Google Maps Street View and Microsoft Virtual Earth) permit interactive examination of landscapes from multiple perspectives. Indianapolis is currently one of about two dozen urban areas in North America with Google Maps Street View imagery available. However, only those portions of urban trails that share the right of way with roads are visible because the imagery is captured from motorized vehicles. As continuous ground-based imaging technologies become available for urban trails, the data they produce could be used to 115
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
provide photorealistic textures for LiDAR models from pedestrian perspectives to improve realism in viewshed analysis.
Software Data processing, analysis, and cartographic design were accomplished using components of ArcGIS 9.2 (Environmental Systems Research Institute, Redlands, California, USA) and ERDAS Imagine 9.1 (Leica Geosystems, St. Gallen, Switzerland).
References FABOS, J. G. (2004) Greenway planning in the United States: its origins and recent case studies, Landscape and Urban Planning, 68(2-3), 321–342. FISHER, P., FARRELLY, C., MADDOCKS, A. and RUGGLES, C. (1997) Spatial analysis of visible areas from the bronze age cairns of Mull, Journal of Archaeological Science, 24(7), 581–592. FISHER-GEWIRTSMAN, D. and WAGNER, I. A. (2003) Spatial openness as a practical metric for evaluating built-up environments, Environment and Planning B: Planning and Design, 30(1), 37–49. FRANK, L. D., ENGELKE, P. O. and SCHMID, T. L. (2003) Health and Community Design: The Impact of the Built Environment on Physical Activity, Island Press, Washington D.C., 253 pp. JENSEN, J. R. (2005) Introductory Digital Image Processing: A Remote Sensing Perspective, Prentice-Hall, New York, 526 pp. LINDSEY, G., HAN, Y., WILSON, J. and YANG, J. (2006) Neighborhood correlates of urban trail traffic, Journal of Physical Activity and Health, 3(S1), S134–S152. LINDSEY, G., WILSON, J., RUBCHINSKAYA, E., YANG, J. and ALEXA, C. (2008) Urban form, trail characteristics, and trail use: implications for design, Journal of Urban Design, 13(1), 107–132. 116
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
LIU, G., COLBERT, J. T., WILSON, J., YAMADA, I. and HOCH, S. C. (2007) Examining urban environment correlates of childhood physical activity and walkability perception with GIS and remote sensing, In Jensen, R., Gatrell, J. and McLean, D., (eds.) Geo-Spatial Technologies in Urban Environments: Policy, Practice, and Pixels, Springer-Verlag, Berlin, 2nd ed., pp. 121–140. LLOBERA, M. (2003) Extending GIS-based visual analysis: the concept of visualscapes, International Journal of Geographic Information Science, 17(1), 25–48. LLOYD, C. D. and ATKINSON, P. M. (2000) Deriving DSMs from LiDAR data with kriging, International Journal of Remote Sensing, 23(12), 2519–2524. MOUDON, A. V. and LEE, C. (2003) Walking and bicycling: an evaluation of environmental audit instruments, American Journal of Health Promotion, 18(1), 21–37. PIKORA, T. F., BUL, F. C. L., JAMROZIK, K., KNUIMA, N., GILES-CORTI, B. and DONOVAN, R. J. (2002) Developing a reliable instrument to measure the physical environment for physical activity, American Journal of Preventive Medicine, 23(3), 187–194. PRIESTNALL, G., JAAFAR, J. and DUNCAN, A. (2000) Extracting urban features from LiDAR digital surface models, Computers, Environment and Urban Systems, 24(2), 65–78. REYNOLDS, K. D., WOLCH, J., BYRNE, J., CHOU, C. P., FENG, G., WEAVER, S. and JERRETT, M. (2007) Trail characteristics as correlates of urban trail use, American Journal of Health Promotion, 21(4S), 335–345. SAELENS, B. E., SALLIS, J. F. and FRANK, L. D. (2003) Environmental Correlates of Walking and Cycling: Findings From the Transportation, Urban Design, and Planning Literatures, Annals of Behavioral Medicine, 25(2), 80–91. SALLIS, J. F., LINTON, L. and KRAFT, M. K. (2005) The first Active Living Research Conference: growth of a transdisciplinary field, American Journal of Preventive Medicine, 28(2S), 93–95.
117
Journal of Maps, 2008, 108-118
Wilson, J., Lindsey, G. & Liu, G.
SAMPSON, R. J., MORENOFF, J. D. and GANNON-ROWLEY, T. (2002) Assessing “Neighborhood Effects”: social processes and new direction in research, Annual Review of Sociology, 28, 443–478. SHONKOFF, J. P. (2003) From neurons to neighborhoods: old and new challenges for developmental and behavioral pediatrics, Journal of Developmental and Behavioral Pediatrics, 24(1), 70–76. TROPED, P. J., SAUNDERS, R. P., PATE, R. R., REININGER, B., UREDA, J. R. and THOMPSON, S. J. (2001) Associations between self-reported and objective physical environmental factors and use of a community rail-trail, Preventive Medicine, 32(2), 191–200. WHEATLEY, D. (1995) Cumulative viewshed analysis: a GIS-based method for investigating intervisibility, and its archaeological application, ˇ ic ˇ, Z., (eds.) Archaeology and Geographic In Lock, G. and Stanc Information Systems: A European Perspective, Taylor and Francis, London, pp. 171–186.
118
