{"id":179371,"date":"2012-08-16T23:47:25","date_gmt":"2012-08-17T06:47:25","guid":{"rendered":"http:\/\/www.esri.com\/arcgis-blog\/?post_type=blog&p=179371"},"modified":"2021-08-02T23:49:37","modified_gmt":"2021-08-03T06:49:37","slug":"using-spatial-analysis-to-measure-city-accessibility-by-intersection-density","status":"publish","type":"blog","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-desktop\/analytics\/using-spatial-analysis-to-measure-city-accessibility-by-intersection-density","title":{"rendered":"Using spatial analysis to measure city accessibility by intersection density"},"author":5101,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","format":"standard","meta":{"_acf_changed":false,"_searchwp_excluded":""},"categories":[23341,23351],"tags":[24321,33681,27631,30891],"industry":[],"product":[36991],"class_list":["post-179371","blog","type-blog","status-publish","format-standard","hentry","category-analytics","category-transportation","tag-geoprocessing","tag-network","tag-roads","tag-transport","product-arcgis-desktop"],"acf":{"short_description":"Past work has shown the more accessible an urban area, the more walking and public transit use is promoted, whilst car journeys and traffic speed","flexible_content":[{"acf_fc_layout":"content","content":"

Past work has shown the more accessible an urban area, the more walking and public transit use is promoted, whilst car journeys and traffic speeds are reduced.\u00a0 What makes an area accessible and how do we determine if an area is indeed accessible?\u00a0 One factor we can consider is intersection density which has shown to be a valuable measure in determining such accessibility (Ewing and Cervero, 2010)\u00a0 and correlates strongly with encouraging people to get out of their cars and cycle, walk, or take public transportation to work.\u00a0 All this suggests that good land planning and urban design could help reduce car use and, therefore, any related social and environmental costs.\u00a0 GIS analysis has a key role to play in shaping our towns and cities and our future environments. \u00a0This blog entry shows how I used geoprocessing and spatial analysis<\/a> in ArcGIS to gain an understanding of spatial patterns of urban accessibility using the USA road network as a case study.<\/p>\n

The current statistics on commuting to work in the USA show a picture of car dependency with 86% of workers traveling by car (and only 10% carpooling), 5% using public transportation and only 3.5% either cycling or walking to work (ACS 2009).\u00a0 These figures, however, do not show spatial variations across the country which is important for local planning.\u00a0 Within the country there are some very interesting differences in commuting patterns.<\/p>\n

The American Community Survey 2009 report (McKenzie and Rapino 2011, McKenzie 2010) shows the percentage of workers who commuted by public transport in 2009 at metropolitan and micropolitan statistical areas (metro and micro areas).\u00a0 These are areas used for data collection and publication by Federal statistical agencies.\u00a0 At this scale, public transit use is as high as 30% in the New York-Northern New Jersey-Long Island metro area to as low as 2% in the Louisville\/Jefferson County metro area.\u00a0 But how does this compare to the accessibility of these areas?<\/p>\n

The first step in performing this analysis involves collating appropriate data.\u00a0 I needed to find out the number and density of road intersections in metro and micro-metro areas, so I needed a street dataset and a feature class of Core Based Statistical Areas CBSA,\u00a0http:\/\/www.census.gov\/population\/metro\/<\/a>). CBSAs are comprised of both metro areas that contain a core urban area of 50,000 or more population, and micro areas that contain an urban core of at least 10,000 (but less than 50,000) population.\u00a0 Each metro or micro area includes at least the county that contains the core urban area plus any adjacent counties that have a high degree of social and economic integration (as measured by commuting to work) with the urban core.<\/p>\n

To calculate the number of intersections, in each metropolitan area, I developed a custom geoprocessing script tool named\u00a0Create\u00a0Junction Connectivity Features<\/a>\u00a0which you can download from the\u00a0Model and Script tool gallery<\/a>.\u00a0 This Python script tool counts the number of lines connected to each vertex in a line feature class.\u00a0 The only input to the tool is a line feature class (Figure 1).<\/p>\n

\"\"<\/a>Figure 1. Create Junction and Connectivity Features tool<\/em><\/p>\n

For each vertex in the dataset, the count of intersecting lines is attributed and an output point feature class generated (Figure 2).\u00a0 Checking \u201cIgnore junctions with only two connecting lines\u201d means that only junctions with three or more connecting streets\u00a0 (intersections) are output, as used in this analysis.<\/p>\n

In order to restrict the analysis to metro and micro areas I used the\u00a0Intersect\u00a0<\/a>tool with the road feature class and the metro and micro areas.\u00a0 This output of Intersect is only those roads that are within CBSA which linear feature class provides the input for the Junction Count tool. \u00a0A\u00a0Spatial Join<\/a>\u00a0can be used to find\u00a0intersections by CBSA and then using\u00a0Dissolve<\/a>\u00a0I calculated the total number of intersections by CBSA.\u00a0Finally, to calculate the intersection density by area, I used the\u00a0Add Field<\/a>\u00a0tool and then calculated density as (intersection count\/CBSA area)*100.\u00a0All these steps can easily be automated using ModelBuilder (figure 3) . Note that custom tools, such as the Create Junction Connectivity Features tool, can be\u00a0 used in models.<\/p>\n

\"\"<\/a>Figure 2: CBSA accessibility model<\/em><\/p>\n

Analyzing the Results<\/strong>
\n\"Road
\nFigure 3. Road intersection density for the USA at CBSA level<\/em><\/p>\n

If we look at the results of this case study, we can see some interesting patterns emerge.\u00a0 As would be expected from conurbations such as New York and Washington DC, transport usage is high (30.5% and 14.1% respectively) and intersection density is high (1.14 and 0.53).\u00a0 However, if we plot the results on a graph we see the analysis throws up some interesting anomalies which are worth further consideration.
\n\"Graph<\/p>\n

Figure 4: Intersection density and percentage of commuters using public transportation<\/em><\/p>\n

This graph, created in EXCEL, allowed me to plot the data together using two Y-axis; one for intersection density and one for the percentage of commuters using public transportation.\u00a0 This method allowed me to graph the two related variables together, although they use very different measures.\u00a0 In this example, we can easily see where the two measures show a different pattern.\u00a0 The graph bars were colored using the same RGB colors as the choropleth map for ease of comparison.<\/p>\n

By calculating the percentage difference (a unit-less measure), we can statistically compare the difference between the intersection density and transportation usage.\u00a0 Although the actual values are very different we expect the differences between the two to show a similar relationship for all metro areas.\u00a0 Plotting these, together with the mean, in ArcGIS allows us to clearly see the pattern. \u00a0We can see any selections made on the graph reflected in our map (Figure 5).<\/p>\n

\"Percentage<\/p>\n

Figure 5.Intersection density and transportation usage percentage difference.<\/em><\/p>\n

We would expect areas with a high intersection density to show high public transportation usage.\u00a0 Using one standard deviation from the mean (calculated using the\u00a0Statistics dialog box\u00a0<\/a>on our attribute table) we can see which areas do not follow this pattern.<\/p>\n

\"Metro<\/p>\n

From this analysis, we can pick out eight metro areas that significantly differ from what we would expect.<\/p>\n