My interest in conducting a bike score evaluation for Philadelphia stems from my enthusiasm in biking. Many media ranks Philadelphia high in bikeability. However, as I tried a few times biking in the downtown area, the experience was not smooth. There are too many cars, very frequent signal controls, disconnected bike lanes, occupied bike lanes and more. All these factors undermine bikers’ experience on streets. Therefore, I hope to build a new bike score rating system and create an interactive tool for urban planners to retrieve bike score data.
I. Metrics for score rating
I believe that not only should bike score reflect bike infrastructure but also bikers’ experience. Based on my biking experience, important variables are the type of bike lane, the width of bike lane, bike network connectivity, adjacent lane average car speed, density of trees, frequency of intersections and slope. Based on the data that are available, I incorporated four variables in my rating system, shown as follows. For all metrics, each is scored using a 100 scale.
a. Lane Type Score
Categorized by different types of lanes. Since bike trail data was not available at the time, there is no 100 score, because I perceive bike trails as the best biking medium.
|Conventional with Sharrow||60|
|Contraflow with Conventional||70|
|Buffered with Conventional||90|
b. Slope Score
Slope(%) is obtained by analyzing the Digital Elevation Model (DEM). Based on literature studies, slopes greater than 10% are not bikeable, therefore with a score 0.
|9.1 – 10||10|
|8.1 – 9||20|
|7.1 – 8||30|
|6.1 – 7||40|
|5.1 – 6||50|
|4.1 – 5||60|
|3.1 – 4||70|
|2.1 – 3||80|
|1.1 – 2||90|
|0 – 1||100|
c. Tree Score
Tree_Score = (observed tree density – min tree density) / (max tree density – min tree density) *100
d. Intersection Control Score
Ctrl_Score = (observed Ctrl density – min ctrl density) / (max ctrl density – min ctrl density) *100
e. Total Score
I scale these criteria in different weights, and the total bike score is also on a 100 scale. The formula is shown below:
Bike Score = LaneTypeScore * 0.5 + SlopeScore * 0.3 + TreeScore * 0.1 + ControlScore * 0.1
II. Result and application
1. Total score overlay
From figure 1 we can see that the maximum score only goes up to 77 out of 100. This is due to the fact that the bike trails are not included, many streets do not have enough street trees and etc. Overall, the score shows that the bike condition in Philadelphia is not too ideal.
2. Interactive Tool
I hope to make this tool useful to urban planners, who often use ArcGIS for spatial analysis. Written in ArcPy script, it takes a pair of latitude and longitude from the user and returns a bike score of that location. Take 3600 Chestnut St, Philadelphia as an example: I enter the location information in Figure 2, and then the tool returns a point feature, as shown in Figure 3, along with an attribute table containing scores of different criteria and the overall score in Figure 4.
This tool provides a new definition to Philadelphia’s bike score. It sets up a framework for the score measurement. There is a lot potential in making this a better tool. However, there are some limitations that I am improving in my capstone project currently.
1. geocoding limitation
First and foremost, I would hope to overcome the technical difficulty and be able to provide the opportunity for users to enter an address, or be able to click on the screen and be able to know the score around that area.
2. absence of bike network connectivity research and routing possibilities
Second, there is no bike connectivity measure in this research. For this, I want to argue that the main point of this tool is to tell local bike scores. But I do recognize the value of having a ranking for the best bike routes down the road.
3. missing useable information about bike trails
One of the most shortcomings of this study is that there is no bike trail information. The bike network does not include it. Although I found an API that shows the bike trails, I was not able to get it as a shapefile and work with it. Had I had this piece of information, the bike score calculation would have been more comprehensive and powerful.
1. bike score and ranking source
2. bike score source
3. Code help
4. Script Credit to:
Paulo Raposo, http://gis.stackexchange.com/questions/199754/arcpy-field-mapping-for-a-spatial-join-keep-only-specific-columns
Example script 2 from: http://pro.arcgis.com/en/pro-app/tool-reference/analysis/spatial-join.htm
5. Long lat conversion:
6. Thanks to Professor Dana Tomlin and Jill Kelly for their help.