Areal Interpolation

Areal interpolation is the process making estimates from a source set of polygons to an overlapping but incongruent set of target polygons.

Also called spatial resampling.

The interpolated values are typically counts or percentages.

Resampling requires an assumption about the distribution of values within the source polygons (i.e., uniformly distributed).

`areal`

areal (Chris Prener) is an R package tailored to perform areal interpolation.

provides a single function that performs the entire workflow
- arguments allow you to specify methodological details
tidyverse and sf friendly
additional functions for power-users who want to go step-by-step
good vignettes on data preparation and methodology
peer reviewed paper

For population weighted areal interpolation (i.e., Binary dasymetric method), see tidycensus::interpolate_pw().

Source - sf polygon object, preferably projected, with column(s) that you want estimated for the target polygons.

Target - sf polygon object, same CRS as the source.

Source and target polygons should overlap.

The source and target layer should both have a column with unique values (i.e., primary key)

Don’t include extraneous columns in the source that you don’t need in the target

Use ar_validate() to validate that your source and target objects are ready-to-go

The main function in areal is:

aw_interpolate(.data, tid, source, sid, extensive, intensive, weight, output)

.data - the target sf object
tid - a field in the target dataset with unique values
source - the source sf object
sid - a field in the source dataset with unique values
extensive - column(s) in the source that you want to interpolate and contain count data
intensive - column(s) in the source that you want to interpolate and contain proportion or density data
weight - should be "sum" or "total"
output - either "sf" or "tibble"

If the source and target polygons overlap completely (excluding small insignificant differences), use weight = 'sum'.

If the source and target polygons don’t cover the same area, use weight = 'total'.

The path of areal interpolation, more work and assumptions you carry. Follow only if you must.

Make sure you really have to do your own areal interpolation.

The Census Bureau summarizes a lot of census data by a lot of geographies.

Other data publishers commonly summarize their data by census enumeration areas and administrative boundaries.

In this exercise, we will:

import the 2020 voting tabulation districts (VTDs) for Camden County NJ from a Shapefile
import a CSV with voter turnout data from the 2020 primary election
join the tabular voter turnout data to the VTD polygons
save the data to disk (for use in other exercises)
map the voter turnout for the July 2020 primary
reshape the attribute table from a wide to long format
create facet maps (i.e., one map for each subset of the data)

https://posit.cloud/content/8521414