Zev Ross President, ZevRoss Spatial Analysis

Transcription

1 SPATIAL ANALYSIS IN R WITH SF AND RASTER Buffers and centroids Zev Ross President, ZevRoss Spatial Analysis

2 Use a projected coordinate reference system for spatial analysis As a general rule your layers should use a projected CRS With multiple layers they should have the same CRS

3 Buffer vectors with st_buffer() The dist argument controls the radius and is in CRS units. # Buffer five trees, this is feet > trees_buf <- st_buffer(trees5, )

4 Compute centroids with st_centroid() > poly <- st_read("poly.shp") > poly_cent <- st_centroid(poly)

5 SPATIAL ANALYSIS IN R WITH SF AND RASTER Let's practice!

6 SPATIAL ANALYSIS IN R WITH SF AND RASTER Zev Ross President, ZevRoss Spatial Analysis Bounding boxes, dissolve features and create a convex hull

7 Defining a region Rectangular bounding box Tighter polygon with a convex hull

8 Compute the coordinates of the bounding box with st_bbox() # Read data in and get the bbox coords > poly <- st_read("poly.shp") # Compute the bounding box coordinates > st_bbox(poly) xmin ymin xmax ymax

9 Create a bounding box # Bounding box of counties in red > poly_grd <- st_make_grid( + poly, n = 1 + ) # Bounding box of centroids in green > cent_grd <- st_make_grid( + poly_cent, n = 1 + )

10 Dissolve features with st_union() For polygons you dissolve multiple features into a single feature For points you cluster individual points into a MULTIPOINT geometry This is required for some computations including convex hull

11 Two polygons in the data frame # Simple feature collection with 2 features and 1 field # geometry type: POLYGON # dimension: XY # bbox: xmin: ymin: xmax: ymax: # epsg (SRID): NA # proj4string: +proj=lcc +lat_1= lat_2= # BoroName geometry # 1 Manhattan POLYGON (( # 2 Brooklyn POLYGON ((

12 Two separate polygons

13 Dissolve into a single polygon with st_union() > grds_union <- st_union(polys) > grds_union # Geometry set for 1 feature # geometry type: POLYGON # dimension: XY # bbox: xmin: # epsg (SRID): NA # proj4string: +proj=lcc... # POLYGON ((

14 Our point data frame starts with five individual points Here we have a data frame of five points (the centroids from our counties). The geometry type is POINT. > select(boro_cent, BoroName, geometry) # Simple feature collection with 5 features and 1 field # geometry type: POINT # dimension: XY # bbox: xmin: ymin: xmax: ymax: # epsg (SRID): NA # proj4string: +proj=lcc +lat_1= # BoroName geometry # 1 Manhattan POINT ( # 2 The Bronx POINT ( # 3 Staten Island POINT ( # 4 Brooklyn POINT ( # 5 Queens POINT (

15 With st_union() points get bundled into a single grouped feature > boro_union <- st_union(boro_cent) > boro_union # Geometry set for 1 feature # geometry type: MULTIPOINT # dimension: XY # bbox: xmin: ymin: xmax: ymax: # epsg (SRID): NA # proj4string: +proj=lcc +lat_1= # MULTIPOINT ( , 9...

16 They look the same though Five points (5 features) vs 1 MULTIPOINT (1 feature with five pieces)

17 Compute a tight bounding box called a convex hull > chull <- st_convex_hull(boro_union)

18 SPATIAL ANALYSIS IN R WITH SF AND RASTER Let's practice!

19 SPATIAL ANALYSIS IN R WITH SF AND RASTER Multi-layer geoprocessing and Zev Ross President, ZevRoss Spatial Analysis relationships

20 Multi-layer spatial analysis Linking features from multiple layers (e.g., spatial join) Determine relationships between features from different layers (e.g., intersect, distance)

21 Spatial join

22 No tract ID in the roads data frame The only attributes are fullname and geometry. > head(roads) # Simple feature collection with 6 features and 2 fields # geometry type: MULTILINESTRING # dimension: XY # bbox: xmin: ymin: xmax: # epsg (SRID): # proj4string: +proj=utm +zone=12 +south +datum=wgs84 +units=m +no_defs # fullname linearid geom # N MULTILINESTRING (( # 4 W Thomas Rd N MULTILINESTRING (( # 5 W Thomas Rd S MULTILINESTRING (( # 6 W Lost Creek Dr E MULTILINESTRING (( # 7 W Lost Creek Dr W MULTILINESTRING (( # 8 I MULTILINESTRING ((

23 Use st_join() to conduct the spatial join > roads <- st_join(roads, tracts) > head(roads) # Simple feature collection with 6 features and 3 fields # geometry type: MULTILINESTRING # dimension: XY # bbox: xmin: ymin: xmax: # epsg (SRID): # proj4string: +proj=utm +zone=12 +south +datum=wgs84 +units=m +no_defs # fullname linearid geoid geom # N MULTILINESTRING (( # N MULTILINESTRING (( # 4 W Thomas Rd N MULTILINESTRING (( # 5 W Thomas Rd S MULTILINESTRING (( # 6 W Lost Creek Dr E MULTILINESTRING (( # 7 W Lost Creek Dr W MULTILINESTRING ((

24 Color-code the plot based on the new attribute > plot(roads["geoid"]) > plot(st_geometry(tracts), border = "black", add = TRUE)

25 Computing relationships between multiple layers Which roads intersect with the grey polygon -- st_intersects() Which roads are completely contained by the polygon -- st_contains()

26 Looking at intersection with st_intersects() # Result is a list > intersects <- st_intersects(tract, roads)[[1]] # The index number of roads that intersect > head(intersects) [1] > plot(st_geometry(roads[intersects,]), add = TRUE, col = "red")

27 Looking at contains with st_contains() > contains <- st_contains(tract, roads)[[1]] > plot(st_geometry(roads[contains,]), add = TRUE, col = "blue")

28 Clip features with st_intersection() # Clip the roads to the tract polygon > clipped <- st_intersection(tract, roads) > plot(st_geometry(tract), col = "grey", border = "white") > plot(st_geometry(clipped), add = TRUE)

29 Computing distance between features

30 Compute distance with st_distance() > dist1 <- st_distance(tract_cent, roads) > roads$dist <- dist1[1, ]

31 SPATIAL ANALYSIS IN R WITH SF AND RASTER Let's practice!

32 SPATIAL ANALYSIS IN R WITH SF AND RASTER Geoprocessing with rasters Zev Ross President, ZevRoss Spatial Analysis

33 Example: Elevation raster and polygons

34 Mask your raster with mask() > polys <- as(polys, "Spatial") > el_mask <- mask(elevation, mask = polys) > plot(el_mask)

35 Crop your raster with crop() > el_crop <- crop(elevation, polys) > plot(el_crop) > plot(polys, add = TRUE)

36 Mask and crop! > el_crop <- crop(elevation, polys) > el_mask <- mask(el_crop, mask = polys) > plot(el_mask)

37 Extract values with extract() With points the raster values under each point are returned With polygons you have two options controlled by the fun argument: Return all values in each polygon (fun = NULL) Supply a summary function (e.g., fun = mean)

38 Extract elevation values for the polygons # For polys fun = NULL will return all values > vals <- extract(elevation, polys, fun = mean) > vals # [,1] # [1,] # hill polygon # [2,] # lake polygon

39 Raster math with overlay() Here we have two rasters: 1. The elevation raster 2. The raster with multiplier values

40 Raster math > f <- function(rast1, rast2) rast1 * rast2 > alt_ele <- overlay(elevation, multiplier, fun = f)

41 Result of overlay Elevation values have been multiplied by either 1.25 or 0.75 depending on their location. Otherwise raster values are set to NA.

42 SPATIAL ANALYSIS IN R WITH SF AND RASTER Let's practice!