As a graduate student in urban planning at Columbia, one of my required courses for fall 2022 was critical GIS, or geographic information systems. Inspired by working at Street Lab through the fall, particularly on the open streets program, my project partner and I created a priority metric to identify which of DOT’s existing open streets are among the most in need of funding/resourcing (similar to “priority A” designations). The Department of Transportation (DOT)’s Open Streets program is a form of tactical urbanism where previously car-dedicated streets are repurposed into pedestrian only spaces. However, it is widely criticized for reproducing spatial inequities, rather than resolving them. We argue that one way to address these inequities is resourcing existing open streets, with funding, programmatic support, full-time staffers like community organizers, infrastructure, and more. Our research created a metric (also known as a multi-criteria decision analysis) to determine which open streets the DOT should prioritize resourcing, through analyzing land use, street safety, and public infrastructure. Our analysis suggests prioritizing three streets: (1) Murray Hill’s Barton Avenue (due to proximity to vision zero traffic intersections + potential to expand into the LIRR plaza nearby), (2) East Harlem’s East 119th Street (due to proximity to PS 112 and a GreenThumb community garden), and (3) Mott Haven’s Alexander Avenue (due to proximity to four NYCHA complexes). Finally, as a proponent of critical feminist methodology and participatory GIS, our project also outlines how to reproduce this same research to incorporate other priorities (maybe you think that flood data or urban heat data should be part of the model!) Take a look at our research report here!
Our project was then selected to present at NYC’s Open Data Week!! We presented on our project and shared ways that community advocates can partner with civic designers and data analysts to create more equitable open streets! Feel free to watch our presentation here!
I continued this project into the spring semester, where I created a participatory street hierarchy to understand which streets in Queens should be transformed from car-only usage to fit the needs of people who live near them. Our project combined three spatial methodologies: participatory GIS, spatially constrained multivariate clustering analysis, and a networked street hierarchy. We suggest (1) transforming the 49th Lane to a Green Street, (2) transforming Roosevelt Avenue between 63rd Road and 69th Street to a Resource Street, (3) transforming 56th Avenue into a Programming Street, (4) transforming Skillman Avenue into a Transportation Alternatives Street, and (5) transforming 37th Avenue into a Food Justice Street.
We started with participatory GIS because quantitative data often fails to capture people’s lived experiences. Our team created a survey with open ended questions and a markable map to ask (1) how people use streets and public spaces (2) what people want more of from their neighborhoods, and (3) whether there are streets or areas in the neighborhood that reflect key public health indicators. After talking to almost 30 people at busy intersections and pedestrian plazas in Maspeth and Woodside, we qualitatively coded this data to create new “use categories” to typologize streets for our subverted street hierarchy. To begin, we created a survey with open ended questions, which broadly asked (1) how people use streets and public spaces (e.g. “what do you see people doing on this street?”) (2) what people want more of in their neighborhoods, and (3) whether there are streets or areas in the neighborhood that reflect key public health indicators (e.g. “mark where you see automobile congestion on the map”). For the first two sets of questions about use categories and community desires, we asked participants questions verbally. For the last set of questions, we asked participants to draw their perceptions on a map of Woodside. We used the results from our participatory GIS, both interviews and markable maps, to create new use categories for street transformation. We started by scanning all of our maps and questionnaires, and then created a table of topics participants mentioned (i.e. bikes). Using that, we went through each interview and noted the frequency with which people mentioned different categories (i.e. people mentioned bikes 12 times). We marked the frequency of each category regardless of positive or negative connotations, because regardless of the participant’s sentiment, it still gave us valuable information about how the street was used. Through our participatory research, we came up with five typologies of streets: green streets, resource streets, public programming streets, transportation alternatives streets, and food justice streets.
In order to create a hierarchy based on harm to pedestrians, we wanted to use spatially constrained multivariate clustering to create statistically significant and spatially co-located groupings of air pollution, flooding, and traffic incidents. While we were able to successfully use this method for air pollution through converting raster data into polygons, we found that it was less effective for flooding data and crash data. For flooding data, spatially constrained multivariate clustering didn’t tell us any new information about which streets flood because our data didn’t represent the amount of flooding, merely the presence of flooding (i.e. YES or NO). For crash data, we found it was more appropriate to use street specific data, rather than clustering. More broadly, crash data and flood data are already associated with a street, so clustering them provided us with less data than the original dataset. We also explored other forms of clustering, like Anselin Local Moran I and Getis Ord GI* which had similar limitations. A street hierarchy is a transportation planning technique that assigns each road’s “character of service” (i.e. primary road, local road) within the vehicular network. This definition only considers a road’s utility to drivers, so we reclassified each road’s use by incorporating a pedestrian-centric ranking system. We visualized this by categorizing roads that cause the least harm to pedestrians based on our variables of air pollution, flooding, and crash data. In order to create a network analysis, we used randomized origin points to simulate how a person may begin their journey from any point in our study area. We then routed these random points using our various hierarchies to subway stations to make sure that we are considering public transit routes as we remove roads from the street network. We iterated this method multiple times, and then combined our multiple routes in order to get a sense of what streets are necessary for the network. Basically, if the network analysis avoided a certain road, we know that it ranked high on our pedestrian harm hierarchy and that it was not absolutely necessary to the network. One major drawback to network street hierarchies as a methodology is that they require a set start and end point, even though our street hierarchy is not optimizing the “best” or “shortest” route.