Recall the scenario of our earlier analysis of the distribution of street trees across the city (Geoprocessing, Aggregation, & the Modifiable Areal Unit Problem: Trees in NYC).
Following this analysis, either you or members of your team conducted further analysis of the density of street trees surrounding McCarren Park in Brooklyn, comparing densities within a series of buffers constructed from the park’s center (Proportional Split Estimates & More Vector Geoprocessing: Trees and Land Use in BK).
After some consideration, you have decided that the earlier analysis of tree density surrounding the park was inadequate to describe the geography of street trees which are, by definition, placed along the street network. As a result, you have decided to revisit the issue, modifying one of the questions of that scenario:
Complete the exercise below, answering the two questions at the end of the exercise.
After completing the exercise, produce one figure suitable for inclusion within an academic paper that describes the analysis and findings (including a comparison of the calculations when determined by straight-line distance buffers versus networked distances along the street). Consider whether the density of street trees is best compared via maps or other types of infographics. Consider how you will graphically communicate the differences between methods used to calculate the findings. (Hint: the syllabus readings associated with this topic include a few academic papers that include similar figures.)
The figure should be no larger than 6.5 inches by 9 inches (portrait orientation)—the interior dimensions of a letter-size page with one-inch margins. Your answers to the two questions at the end of the exercise should be incorporated into the figure through annotation, callouts, labels, an infographic, or some other method.
Whereas a simpler analysis of the street tree density at distances from the park would be conducted with radial buffers, here we will construct “network buffers” that measure distances from McCarren Park along the street network in walkable intervals of 1/8-mile up to a half-mile from the center of the park. In most GIS software, a network-based “radius” is commonly referred to as a “service area.”
We will isolate the pedestrian accessible streets from the City’s LION street centerline dataset, and then use this centerline dataset as the basis for our network measurement.
Recall that networks are graphs, which are structurally different than our polyline vector feature classes. Depending on your software, some network analysis tools will create a graph from input polyline features each time the tool is executed. Others will require a premade graph as the input. Because of this difference, this exercise includes tutorials for calculating a service area in ArcGIS Pro which requires creating a graph (Esri calls this a “network dataset”) from the polyline feature class and calculating a service area in QGIS which does not require creating a graph beforehand (but does require setting the specifications of the graph within the service area tool). The two software approaches also differ in their output geometries, which is discussed further below.
Lastly, once we have generated service area polygons, we will calculate and compare the density of street trees as measured within the network service areas and within radial buffers, by spatially joining the street tree points to each polygon.
<aside> 💾 Download and unzip the exercise data package, saving it to your working directory.
Exercise data packages are accessible via Google Drive with a Columbia University UNI login. If you would like a copy of the full course materials, see here.
</aside>
This exercise will re-employ the datasets provided with the earlier street tree analysis—namely, the point locations of street trees, as collected in the 2015 street tree census. A subset of the original dataset (the 2017 updated version) including only the street tree points within the borough of Brooklyn is included in the geodatabase called Networks_PreppedDatasets. This subset was created to streamline our processing, given that our study area is located in Brooklyn. The dataset is named StreetTreeCensus_2015_BK.
We will also use a single, multipart polygon feature representing the boundaries of McCarren Park in Brooklyn. The feature was created by isolating the polygons representing McCarren Park in the City’s open space feature class, then subsequently dissolving the polygons together to create one multipart feature. It is prepared as the feature class called McCarrenPark_Dissolved in the Networks_PreppedDatasets geodatabase.
Lastly, we will need one additional dataset representing the street network in New York City: The LION dataset produced by the City’s Department of City Planning. The dataset and its metadata are available for download at the URL included in its citation below. The dataset as downloaded is also provided in the exercise materials along with a PDF of its metadata. (Recalling our original scenario, and maintaining the vintages of our datasets, we are using the 2017A version of the LION database.)
New York City Department of City Planning. 2017. LION v17A. [Esri File Geodatabase]. https://www1.nyc.gov/site/planning/data-maps/open-data/dwn-lion.page.
The LION database contains street centerlines (as well as administrative boundaries and other linear features) as polyline features with several attributes per feature segment. These attributes include address ranges on either side of the street, which we will use in a later exercise.