SUNGEO / Sub-National Geospatial Data Archive: Geoprocessing Toolkit

R package for integrating spatially-misaligned GIS datasets.

Version 0.2.22 (April 8, 2021)

Jason Byers, Marty Davidson, Yuri M. Zhukov

Center for Political Studies, Institute for Social Research

University of Michigan

Feedback, bug reports welcome: zhukov-at-umich-dot-edu

To install in R:

library(devtools)
devtools::install_github("zhukovyuri/SUNGEO", dependencies = TRUE)

Load package:

library(SUNGEO)

Read help files:

?poly2poly_ap
?utm_select

Example: geocode an address with OpenStreetMap

# Get geographic coordinates for the Big House (top match only)
geocode_osm("Michigan Stadium")
geocode_osm("Big House")

# Return detailed results for top match
geocode_osm("Michigan Stadium", details=TRUE)

# Return detailed results for all matches
geocode_osm("Michigan Stadium", details=TRUE, return_all = TRUE)

Example: geocode multiple addresses

# Geocode multiple addresses (top matches only)
geocode_osm_batch(c("Ann Arbor","East Lansing","Columbus"))

# ... with progress reports
geocode_osm_batch(c("Ann Arbor","East Lansing","Columbus"), 
                  verbose = TRUE)

# Return detailed results for all matches
geocode_osm_batch(c("Ann Arbor","East Lansing","Columbus"),
                  details = TRUE, return_all = TRUE)

Example: area-weighted polygon-to-polygon interpolation

# Load legislative election results (from CLEA)
data(clea_deu2009)

# Visualize voter turnout at constituency level
plot(clea_deu2009["to1"])

# Load 0.5 degree hexagonal grid
data(hex_05_deu)

# Interpolate
out_1 <- poly2poly_ap(poly_from = clea_deu2009,
                      poly_to = hex_05_deu,
                      poly_to_id = "HEX_ID",
                      varz = "to1"
                      )

# Visualize voter turnout at grid cell level
plot(out_1["to1_aw"])

Example: population-weighted polygon-to-polygon interpolation

# Load population raster (from GPW v4)
data(gpw4_deu2010)

# Interpolate
out_2 <- poly2poly_ap(poly_from = clea_deu2009,
                      poly_to = hex_05_deu,
                      poly_to_id = "HEX_ID",
                      varz = "to1",
                      methodz = "pw",
                      pop_raster = gpw4_deu2010)

# Visualize voter turnout at grid cell level
plot(out_2["to1_pw"])

Example: point-to-polygon interpolation using tessellation method and area weights

# Load point-level election results
data(clea_deu2009_pt)

# Interpolate
out_4 <- point2poly_tess(pointz = clea_deu2009_pt,
                         polyz = hex_05_deu,
                         poly_id = "HEX_ID",
                         varz = "to1")

# Visualize voter turnout at grid cell level 
plot(out_4["to1_aw"])

Example: point-to-polygon interpolation using ordinary Kriging

# Ordinary Kriging with one outcome variable
out_5 <- point2poly_krige(pointz = clea_deu2009_pt,
                          polyz = clea_deu2009,
                          yvarz = "to1")

# Compare observed values to predictions
par(mfrow=c(1,2))
plot(clea_deu2009["to1"], key.pos = NULL, reset = FALSE)
plot(out_5["to1.pred"], key.pos = NULL, reset = FALSE)

# Ordinary Kriging with multiple outcome variables
out_6 <- point2poly_krige(pointz = clea_deu2009_pt,
                          polyz = clea_deu2009,
                          yvarz = c("to1","pvs1_margin"))

# Compare observed values to predictions
par(mfrow=c(1,2))
plot(clea_deu2009["pvs1_margin"], key.pos = NULL, reset = FALSE)
plot(out_6["pvs1_margin.pred"], key.pos = NULL, reset = FALSE)

Example: point-to-polygon interpolation using universal Kriging

# Universal Kriging with one outcome variable and one covariate
out_7 <- point2poly_krige(pointz = clea_deu2009_pt,
                        polyz = clea_deu2009,
                        yvarz = "to1",
                        rasterz = gpw4_deu2010)

# Compare observed values to predictions
par(mfrow=c(1,2))
plot(clea_deu2009["to1"], key.pos = NULL, reset = FALSE)
plot(out_7["to1.pred"], key.pos = NULL, reset = FALSE)

Example: line-in-polygon analysis

# Load road data (from Digital Chart of the World) and extract highways
data(highways_deu1992)

# Basic map overlay
plot(hex_05_deu["geometry"])
plot(highways_deu1992$geometry, add=TRUE, col = "blue", lwd=2)

# Calculate road lengths, densities and distances from each polygon to nearest highway
out_8 <- line2poly(linez = highways_deu1992,
                   polyz = hex_05_deu,
                   poly_id = "HEX_ID")
                   
# Visualize results
plot(out_8["line_length"])
plot(out_8["line_density"])
plot(out_8["line_distance"])

# Replace missing road lengths and densities with 0's, rename variables
out_9 <- line2poly(linez = highways_deu1992,
                   polyz = hex_05_deu,
                   poly_id = "HEX_ID",
                   outvar_name = "road",
                   na_val = 0)

# Visualize results
plot(out_9["road_length"])
plot(out_9["road_density"])
plot(out_9["road_distance"])
dev.off()

Example: Automatically find a planar CRS for a GIS dataset

# Visualize original geometries (WGS1984, degrees)
plot(clea_deu2009["geometry"], axes=TRUE)

# Find a suitable CRS and re-project
out_10 <- utm_select(clea_deu2009)

# Visualize transformed geometries (UTM 32N, meters)
plot(out_10["geometry"], axes=TRUE)

# proj4string of transformed data
utm_select(clea_deu2009, return_list=TRUE)$proj_out

Example: Rasterization of polygons

# Transform sf polygon layer into 1km-by-1km RasterLayer (requires planar CRS)
out_11 <- sf2raster(polyz_from = utm_select(clea_deu2009),
                   input_variable = "to1")
# Visualize raster
raster::plot(out_11)

# 25km-by-25km RasterLayer (requires planar CRS)
out_12 <- sf2raster(polyz_from = utm_select(clea_deu2009),
                   input_variable = "to1",
                   grid_res = c(25000, 25000))
# Visualize raster
raster::plot(out_12)

Example: Create cartogram

# Cartogram of turnout scaled by number of valid votes
out_13 <- sf2raster(polyz_from = utm_select(clea_deu2009),
                  input_variable = "to1",
                  cartogram = TRUE,
                  carto_var = "vv1")
raster::plot(out_13)

Example: reverse rasterization

# Polygonization of cartogram raster
out_14a <- sf2raster(polyz_from = utm_select(clea_deu2009),
                    input_variable = "to1",
                    cartogram = TRUE,
                    carto_var = "vv1",
                    return_list = TRUE)
out_14 <- sf2raster(reverse = TRUE,
                   poly_to = out_14a$poly_to,
                   return_output = out_14a$return_output,
                   return_field = out_14a$return_field)
plot(out_14["to1"])

Additional examples in help files of individual functions.