Contents

Scenario

Since the Bloomberg administration’s Million Tree initiative, large-scale, volunteer-based street tree census data has been intermittently collected in New York City. As more trees are planted (and counted), you are interested in examining the overall geographic patterns of the city’s tree cover relative to the city’s population and neighborhoods. Given known relationships between vegetation and quality of life indicators, you are specifically interested in understanding the spatial equity of the city’s street trees and in taking steps, if necessary, to advocate for more tree plantings in underserved neighborhoods. Thus, the primary questions your initial analysis must answer are

Where are the relatively highest and lowest street tree densities in the city?
And in which City Council Districts are these located?

The complete tree census data includes the coordinate locations of each counted tree in the city, and includes over 680,000. With such a high number of trees, variations between neighborhoods are imperceivable if the data is merely mapped as points across the five boroughs. As such, you will need to aggregate the point-level data to appropriate boundaries which raises secondary questions: Which boundaries are most appropriate for aggregation, and how does this choice change the results of your analysis?

You have decided to compare several options because of the relative advantages of each approach. In short, in addition to understanding the distribution of street trees by city council district, you have decided that…

Census blocks should be mapped because they are the smallest spatial unit available and therefore can reveal more granular variation in tree density. Larger areal units would obscure block-by-block differences and would, for example, hinder an analysis of tree density very near truck routes.
Census tracts should be mapped because they are the only sub-borough spatial unit for which several pertinent census demographic variables are available. If you would seek to directly compare household income levels to tree density, for example, you would need to conduct that comparison at the level of census tracts.
Similarly, the potential need to compare relevant health data—such as asthma rates—to tree density suggests that the tree census information should be analyzed based on the city’s health area boundaries (by which most health information is aggregated).
The eventual goals of organizing community leaders toward advocacy suggest that mapping tree density by community district would also be beneficial. The outcomes of your analysis may help inform communities and mobilize efforts within the scales of the city’s public planning process.

Confidently answering your primary research question will require calculating tree density levels when aggregated to four different relevant boundaries and comparing the results.

Assignment & Deliverables

Complete the exercise below, answering the embedded questions.

There are no map deliverables from this exercise. Rather, the scenario describes preliminary research questions—evaluating the impacts of choosing specific areal units as the basis for analysis. Still, several analytical maps will be generated through the process (although they are not required to be shared with others).

Data

<aside> 💾 Download and unzip the exercise data package from google drive, saving it to your working directory.

Exercise data packages are accessible with a Columbia University login. If you would like a copy of the full course materials, see here.

</aside>

The data needed for this exercise includes a table of tree-level (unaggregated) data from the 2015 Street Tree Census as well as boundary shapefiles for each of the five spatial units described in the Scenario.

<aside> ❔ QUESTION 1: Using the information included here, on the websites linked below, and/or the metadata files accompanying each dataset, write complete bibliographic citations for each of the five datasets required for this exercise. They are included in the exercise’s data package in the \tables\ and \vector\ folders (one table and four shapefiles, respectively).

</aside>