In bmcgaughey1/HexSampleR: Create Sample Points For Polygon(s)
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
Introduction
This vignette provides examples for the HexSampleR package developed by Robert J. McGaughey with the
USDA Forest Service, Pacific Northwest Research Station. The HexSampleR package is designed to facilitate the
creation of sample point locations that represent a specific land area such as the area sampled by a single
plot within the Forest Inventory and Analysis program of the USDA Forest Service.
The examples presented in this document use a single polygon. However, the generatePolygonSamplePoints()
function accepts Spatial* or sf objects containing multiple features, each of which may contain multiple
polygons. With multiple features, the sampling logic demonstrated in this document is applied to each feature.
If the feature contains multiple polygons, sample points are spread over all polygons in the feature.
The HexSampleR package, specifically the generatePolygonSamplePoints() function in the package, makes use
of the spsample() function in the sp package. However, the generatePolygonSamplePoints() function expands
on the sample generation capabilities in spsample(). In addition, the computeCellSize() function greatly
simplifies the calculation of the cellsize parameter needed to call spsample() to generate points representing
a specific sample area.
library(sf)
library(raster)
library(ggplot2)
library(HexSampleR)
Load a polygon representing Snohomish County in Washington State
We first load a polygon representing Snohomish County, Washington from the HexSampleR package. This
polygon was download from the US Census Tiger data using the tigris package. The original polygon has some
fields with values too large to be written to a shapefile and some unnecessary fields. All of the problematic
and unnecessary field were dropped.
The original polygon is in NAD83 longlat. However, the generatePolygonSamplePoints() function will not work
with data using units of degrees so we need to project to a coordinate system that uses consistent units. UTM
zone 10N (EPSG:26910) is appropriate for this county.
Because of the way the data is included with the HexSampleR package, we can access it directly.
WASnoCo <- WASnoCo
# reproject into UTM zone 10...generatePolygonSamplePoints() won't work with data in LAT-LON
WASnoCo <- st_transform(WASnoCo, crs = 26910)
Generate sample points for the county
These examples each produce a set of sample points for Snohomish County. We will use the computeCellSize()
helper function to compute the cell size needed to generate the sample points given a desired sample area
and sampling pattern. Note that the units for cellsize will be feet since we are providing a sample area in
acres so we need to explicitly convert to meters to match our UTM projection by dividing by 3.2808.
For all examples, we use IDColumnLabel = "NAME" in the call to generatePolygonSamplePoints() to constuct
an identifier for each sample point using the value in the NAME field of WASnoCo ("Snohomish"). Sample point
identifiers will be "Snohomish_1", "Snohomish_2", "Snohomish_3", etc.
Example 1 creates a simple grid of sample points such that each point represents approximately 10 square
miles (6400 acres). The spacing of the points will be such that interior points will represent exactly 10
square miles but points near the exterior of the county polygon will represent more or less area due to
irregularities in the county polygon. The polyBuffer parameter in the call to generatePolygonSamplePoints()
buffers the county polygon inward to avoid placing sample points near the edge of the polygon.
Example 2 builds on example 1 but adds a random offset to each sample point location. The offset is up to
1/2 the sample cell size so points will always be located within the cell but can fall very close to a
point in an adjacent cell. Because the sample points are shifted, the area represented by each sample
point will vary.
Example 3 creates sample points on a regular hexagonal grid with each point representing 6000 acres. This is the
area that corresponds to the area sampled by a single FIA plot.
Example 4 builds on example 3 by introducing a random offset to the sample point location. In this case,
the offset is smaller than the distance to the edge of the hexagon so points will stay near the center
of the hexagons. Because the sample points are shifted, the area represented by each sample
point will vary.
The final example, example 5, uses the sample developed in example 4 but then subsamples the sample points
to produce a 10% sample for the county. This is accomplished by adding the count = 0.1 parameter. Values
of count between 0 and 1 are interpreted as the proportion of the sample points to be returned. Note that
this subsampling capability cannot be used to generate multiple subsamples without replacement. This type
of operation would be used to generate a 10% sample for a fixed set of plots but it is beyond the scope
of the HexSampleR package.
# Example 1
pattern <- "regular"
cellsize <- computeCellSize(6400, aunits = "acre", pattern = pattern) / 3.2808
sample_pts <- generatePolygonSamplePoints(WASnoCo
, polyBuffer = cellsize / 2
, pattern = pattern
, gridSpacing = cellsize
, IDColumnLabel = "NAME"
)
#st_write(sample_pts, "WA_SNOCO_points_Ex1.shp", delete_layer = TRUE, quiet = TRUE)
ggplot(WASnoCo) +
ggtitle("Example 1 -- Simple Grid of Sample Points") +
geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) +
geom_sf(data = sample_pts, colour = "black", size = 2)
#Example 2
pattern <- "regular"
cellsize <- computeCellSize(6400, aunits = "acre", pattern = pattern) / 3.2808
sample_pts <- generatePolygonSamplePoints(WASnoCo
, polyBuffer = cellsize / 2
, pattern = pattern
, gridSpacing = cellsize
, IDColumnLabel = "NAME"
, perturb = TRUE
, perturbMultiplier = 0.5
)
#st_write(sample_pts, "WA_SNOCO_points_Ex2.shp", delete_layer = TRUE, quiet = TRUE)
ggplot(WASnoCo) +
ggtitle("Example 2 -- Simple Grid of Sample Points With Random Offset") +
geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) +
geom_sf(data = sample_pts, colour = "black", size = 2)
#Example 3
pattern <- "hexagonal"
cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808
sample_pts <- generatePolygonSamplePoints(WASnoCo
, polyBuffer = cellsize / 2
, pattern = pattern
, gridSpacing = cellsize
, IDColumnLabel = "NAME"
)
#st_write(sample_pts, "WA_SNOCO_points_Ex3.shp", delete_layer = TRUE, quiet = TRUE)
ggplot(WASnoCo) +
ggtitle("Example 3 -- Hexagonal Grid of Sample Points") +
geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) +
geom_sf(data = sample_pts, colour = "black", size = 2)
#Example 4
pattern <- "hexagonal"
cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808
sample_pts <- generatePolygonSamplePoints(WASnoCo
, polyBuffer = cellsize / 2
, pattern = pattern
, gridSpacing = cellsize
, IDColumnLabel = "NAME"
, perturb = TRUE
, perturbMultiplier = 0.25
)
#st_write(sample_pts, "WA_SNOCO_points_Ex4.shp", delete_layer = TRUE, quiet = TRUE)
ggplot(WASnoCo) +
ggtitle("Example 4 -- Hexagonal Grid of Sample Points With Random Offset") +
geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) +
geom_sf(data = sample_pts, colour = "black", size = 2)
#Example 5
pattern <- "hexagonal"
cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808
sample_pts <- generatePolygonSamplePoints(WASnoCo
, polyBuffer = cellsize / 2
, pattern = pattern
, gridSpacing = cellsize
, IDColumnLabel = "NAME"
, perturb = TRUE
, perturbMultiplier = 0.25
, count = 0.1
)
#st_write(sample_pts, "WA_SNOCO_points_Ex5.shp", delete_layer = TRUE, quiet = TRUE)
ggplot(WASnoCo) +
ggtitle("Example 5 -- 10% Sample of Hexagonal Grid Points With Random Offset") +
geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) +
geom_sf(data = sample_pts, colour = "black", size = 2)
bmcgaughey1/HexSampleR documentation built on July 1, 2024, 10:02 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
knitr::opts_chunk$set( collapse = TRUE, comment = "#>" )
Introduction
This vignette provides examples for the HexSampleR package developed by Robert J. McGaughey with the USDA Forest Service, Pacific Northwest Research Station. The HexSampleR package is designed to facilitate the creation of sample point locations that represent a specific land area such as the area sampled by a single plot within the Forest Inventory and Analysis program of the USDA Forest Service.
The examples presented in this document use a single polygon. However, the generatePolygonSamplePoints() function accepts Spatial* or sf objects containing multiple features, each of which may contain multiple polygons. With multiple features, the sampling logic demonstrated in this document is applied to each feature. If the feature contains multiple polygons, sample points are spread over all polygons in the feature.
The HexSampleR package, specifically the generatePolygonSamplePoints() function in the package, makes use of the spsample() function in the sp package. However, the generatePolygonSamplePoints() function expands on the sample generation capabilities in spsample(). In addition, the computeCellSize() function greatly simplifies the calculation of the cellsize parameter needed to call spsample() to generate points representing a specific sample area.
library(sf) library(raster) library(ggplot2) library(HexSampleR)
Load a polygon representing Snohomish County in Washington State
We first load a polygon representing Snohomish County, Washington from the HexSampleR package. This polygon was download from the US Census Tiger data using the tigris package. The original polygon has some fields with values too large to be written to a shapefile and some unnecessary fields. All of the problematic and unnecessary field were dropped.
The original polygon is in NAD83 longlat. However, the generatePolygonSamplePoints() function will not work with data using units of degrees so we need to project to a coordinate system that uses consistent units. UTM zone 10N (EPSG:26910) is appropriate for this county.
Because of the way the data is included with the HexSampleR package, we can access it directly.
WASnoCo <- WASnoCo # reproject into UTM zone 10...generatePolygonSamplePoints() won't work with data in LAT-LON WASnoCo <- st_transform(WASnoCo, crs = 26910)
Generate sample points for the county
These examples each produce a set of sample points for Snohomish County. We will use the computeCellSize() helper function to compute the cell size needed to generate the sample points given a desired sample area and sampling pattern. Note that the units for cellsize will be feet since we are providing a sample area in acres so we need to explicitly convert to meters to match our UTM projection by dividing by 3.2808.
For all examples, we use IDColumnLabel = "NAME" in the call to generatePolygonSamplePoints() to constuct an identifier for each sample point using the value in the NAME field of WASnoCo ("Snohomish"). Sample point identifiers will be "Snohomish_1", "Snohomish_2", "Snohomish_3", etc.
Example 1 creates a simple grid of sample points such that each point represents approximately 10 square miles (6400 acres). The spacing of the points will be such that interior points will represent exactly 10 square miles but points near the exterior of the county polygon will represent more or less area due to irregularities in the county polygon. The polyBuffer parameter in the call to generatePolygonSamplePoints() buffers the county polygon inward to avoid placing sample points near the edge of the polygon.
Example 2 builds on example 1 but adds a random offset to each sample point location. The offset is up to 1/2 the sample cell size so points will always be located within the cell but can fall very close to a point in an adjacent cell. Because the sample points are shifted, the area represented by each sample point will vary.
Example 3 creates sample points on a regular hexagonal grid with each point representing 6000 acres. This is the area that corresponds to the area sampled by a single FIA plot.
Example 4 builds on example 3 by introducing a random offset to the sample point location. In this case, the offset is smaller than the distance to the edge of the hexagon so points will stay near the center of the hexagons. Because the sample points are shifted, the area represented by each sample point will vary.
The final example, example 5, uses the sample developed in example 4 but then subsamples the sample points to produce a 10% sample for the county. This is accomplished by adding the count = 0.1 parameter. Values of count between 0 and 1 are interpreted as the proportion of the sample points to be returned. Note that this subsampling capability cannot be used to generate multiple subsamples without replacement. This type of operation would be used to generate a 10% sample for a fixed set of plots but it is beyond the scope of the HexSampleR package.
# Example 1 pattern <- "regular" cellsize <- computeCellSize(6400, aunits = "acre", pattern = pattern) / 3.2808 sample_pts <- generatePolygonSamplePoints(WASnoCo , polyBuffer = cellsize / 2 , pattern = pattern , gridSpacing = cellsize , IDColumnLabel = "NAME" ) #st_write(sample_pts, "WA_SNOCO_points_Ex1.shp", delete_layer = TRUE, quiet = TRUE) ggplot(WASnoCo) + ggtitle("Example 1 -- Simple Grid of Sample Points") + geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) + geom_sf(data = sample_pts, colour = "black", size = 2) #Example 2 pattern <- "regular" cellsize <- computeCellSize(6400, aunits = "acre", pattern = pattern) / 3.2808 sample_pts <- generatePolygonSamplePoints(WASnoCo , polyBuffer = cellsize / 2 , pattern = pattern , gridSpacing = cellsize , IDColumnLabel = "NAME" , perturb = TRUE , perturbMultiplier = 0.5 ) #st_write(sample_pts, "WA_SNOCO_points_Ex2.shp", delete_layer = TRUE, quiet = TRUE) ggplot(WASnoCo) + ggtitle("Example 2 -- Simple Grid of Sample Points With Random Offset") + geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) + geom_sf(data = sample_pts, colour = "black", size = 2) #Example 3 pattern <- "hexagonal" cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808 sample_pts <- generatePolygonSamplePoints(WASnoCo , polyBuffer = cellsize / 2 , pattern = pattern , gridSpacing = cellsize , IDColumnLabel = "NAME" ) #st_write(sample_pts, "WA_SNOCO_points_Ex3.shp", delete_layer = TRUE, quiet = TRUE) ggplot(WASnoCo) + ggtitle("Example 3 -- Hexagonal Grid of Sample Points") + geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) + geom_sf(data = sample_pts, colour = "black", size = 2) #Example 4 pattern <- "hexagonal" cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808 sample_pts <- generatePolygonSamplePoints(WASnoCo , polyBuffer = cellsize / 2 , pattern = pattern , gridSpacing = cellsize , IDColumnLabel = "NAME" , perturb = TRUE , perturbMultiplier = 0.25 ) #st_write(sample_pts, "WA_SNOCO_points_Ex4.shp", delete_layer = TRUE, quiet = TRUE) ggplot(WASnoCo) + ggtitle("Example 4 -- Hexagonal Grid of Sample Points With Random Offset") + geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) + geom_sf(data = sample_pts, colour = "black", size = 2) #Example 5 pattern <- "hexagonal" cellsize <- computeCellSize(6000, aunits = "acre", pattern = pattern) / 3.2808 sample_pts <- generatePolygonSamplePoints(WASnoCo , polyBuffer = cellsize / 2 , pattern = pattern , gridSpacing = cellsize , IDColumnLabel = "NAME" , perturb = TRUE , perturbMultiplier = 0.25 , count = 0.1 ) #st_write(sample_pts, "WA_SNOCO_points_Ex5.shp", delete_layer = TRUE, quiet = TRUE) ggplot(WASnoCo) + ggtitle("Example 5 -- 10% Sample of Hexagonal Grid Points With Random Offset") + geom_sf(aes(fill = COUNTYFP), show.legend = FALSE) + geom_sf(data = sample_pts, colour = "black", size = 2)
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.