Sampling

Currently, there are 9 functions associated with the sample verb in the sgsR package:

Algorithm Description Reference
sample_srs() Simple random
sample_systematic() Systematic
sample_strat() Stratified Queinnec, White, & Coops (2021)
sample_sys_strat() Systematic Stratified
sample_nc() Nearest centroid Melville & Stone (2016)
sample_clhs() Conditioned Latin hypercube Minasny & McBratney (2006)
sample_balanced() Balanced sampling Grafström, A. Lisic, J (2018)
sample_ahels() Adapted hypercube evaluation of a legacy sample Malone, Minasny, & Brungard (2019)
sample_existing() Sub-sampling an existing sample

sample_srs

We have demonstrated a simple example of using the sample_srs() function in vignette("sgsR"). We will demonstrate additional examples below.

raster

The input required for sample_srs() is a raster. This means that sraster and mraster are supported for this function.

#--- perform simple random sampling ---#
sample_srs(
  raster = sraster, # input sraster
  nSamp = 200, # number of desired sample units
  plot = TRUE
) # plot

#> Simple feature collection with 200 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431250 ymin: 5337710 xmax: 438510 ymax: 5343230
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>                  geometry
#> 1  POINT (431770 5340890)
#> 2  POINT (434330 5341750)
#> 3  POINT (432750 5339510)
#> 4  POINT (431990 5339850)
#> 5  POINT (435050 5343030)
#> 6  POINT (433510 5341850)
#> 7  POINT (434630 5337830)
#> 8  POINT (435470 5343210)
#> 9  POINT (435390 5340170)
#> 10 POINT (433170 5340970)
sample_srs(
  raster = mraster, # input mraster
  nSamp = 200, # number of desired sample units
  access = access, # define access road network
  mindist = 200, # minimum distance sample units must be apart from one another
  buff_inner = 50, # inner buffer - no sample units within this distance from road
  buff_outer = 200, # outer buffer - no sample units further than this distance from road
  plot = TRUE
) # plot

#> Simple feature collection with 200 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431130 ymin: 5337710 xmax: 438550 ymax: 5343170
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>                  geometry
#> 1  POINT (431450 5342850)
#> 2  POINT (433370 5340870)
#> 3  POINT (438550 5338070)
#> 4  POINT (438350 5338590)
#> 5  POINT (434410 5339990)
#> 6  POINT (437530 5342030)
#> 7  POINT (438330 5339470)
#> 8  POINT (436850 5338450)
#> 9  POINT (431450 5342490)
#> 10 POINT (434890 5342530)

sample_systematic

The sample_systematic() function applies systematic sampling across an area with the cellsize parameter defining the resolution of the tessellation. The tessellation shape can be modified using the square parameter. Assigning TRUE (default) to the square parameter results in a regular grid and assigning FALSE results in a hexagonal grid.

The location of sample units can also be adjusted using the locations parameter, where centers takes the center, corners takes all corners, and random takes a random location within each tessellation. Random start points and translations are applied when the function is called.

#--- perform grid sampling ---#
sample_systematic(
  raster = sraster, # input sraster
  cellsize = 1000, # grid distance
  plot = TRUE
) # plot

#> Simple feature collection with 38 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431122.5 ymin: 5337766 xmax: 438445.6 ymax: 5343227
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>                    geometry
#> 1  POINT (436297.3 5343227)
#> 2  POINT (435313.9 5343046)
#> 3  POINT (434330.5 5342864)
#> 4  POINT (433347.1 5342683)
#> 5  POINT (432363.7 5342501)
#> 6  POINT (431380.3 5342320)
#> 7  POINT (438445.6 5342606)
#> 8  POINT (437462.2 5342425)
#> 9  POINT (436478.7 5342244)
#> 10 POINT (435495.3 5342062)
#--- perform grid sampling ---#
sample_systematic(
  raster = sraster, # input sraster
  cellsize = 500, # grid distance
  square = FALSE, # hexagonal tessellation
  location = "random", # randomly sample within tessellation
  plot = TRUE
) # plot

#> Simple feature collection with 176 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431132.4 ymin: 5337742 xmax: 438535.9 ymax: 5343231
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>                    geometry
#> 1  POINT (431379.6 5342970)
#> 2  POINT (431235.5 5342688)
#> 3  POINT (431575.5 5343041)
#> 4  POINT (431831.4 5342717)
#> 5    POINT (431331 5341504)
#> 6  POINT (431791.2 5342100)
#> 7  POINT (432072.2 5342952)
#> 8  POINT (431253.6 5340850)
#> 9  POINT (431865.8 5341872)
#> 10 POINT (432084.9 5342740)
sample_systematic(
  raster = sraster, # input sraster
  cellsize = 500, # grid distance
  access = access, # define access road network
  buff_outer = 200, # outer buffer - no sample units further than this distance from road
  square = FALSE, # hexagonal tessellation
  location = "corners", # take corners instead of centers
  plot = TRUE
)

#> Simple feature collection with 618 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431316.9 ymin: 5337742 xmax: 438393.9 ymax: 5343239
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>                    geometry
#> 1  POINT (431380.7 5342697)
#> 2  POINT (431380.7 5342697)
#> 3    POINT (431517 5342443)
#> 4  POINT (431973.7 5343178)
#> 5  POINT (431821.5 5342933)
#> 6    POINT (431517 5342443)
#> 7  POINT (431380.7 5342697)
#> 8  POINT (431821.5 5342933)
#> 9  POINT (431957.8 5342679)
#> 10 POINT (431805.5 5342433)

sample_strat

The sample_strat() contains two methods to perform sampling:

method = "Queinnec"

Queinnec, M., White, J. C., & Coops, N. C. (2021). Comparing airborne and spaceborne photon-counting LiDAR canopy structural estimates across different boreal forest types. Remote Sensing of Environment, 262(August 2020), 112510.

This algorithm uses moving window (wrow and wcol parameters) to filter the input sraster to prioritize sample unit allocation to where stratum pixels are spatially grouped, rather than dispersed individuals across the landscape.

Sampling is performed using 2 rules:

The rule applied to a select each sample unit is defined in the rule attribute of output samples. We give a few examples below:

#--- perform stratified sampling random sampling ---#
sample_strat(
  sraster = sraster, # input sraster
  nSamp = 200
) # desired sample size # plot
#> Simple feature collection with 200 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431150 ymin: 5337750 xmax: 438530 ymax: 5343170
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata type  rule               geometry
#> x       1  new rule1 POINT (436050 5342330)
#> x1      1  new rule1 POINT (434670 5341170)
#> x2      1  new rule1 POINT (434390 5342610)
#> x3      1  new rule1 POINT (436950 5338070)
#> x4      1  new rule1 POINT (433910 5342950)
#> x5      1  new rule1 POINT (437790 5338290)
#> x6      1  new rule1 POINT (434690 5341670)
#> x7      1  new rule1 POINT (435690 5342470)
#> x8      1  new rule1 POINT (436990 5338230)
#> x9      1  new rule1 POINT (433790 5342990)

In some cases, users might want to include an existing sample within the algorithm. In order to adjust the total number of sample units needed per stratum to reflect those already present in existing, we can use the intermediate function extract_strata().

This function uses the sraster and existing sample units and extracts the stratum for each. These sample units can be included within sample_strat(), which adjusts total sample units required per class based on representation in existing.

#--- extract strata values to existing samples ---#
e.sr <- extract_strata(
  sraster = sraster, # input sraster
  existing = existing
) # existing samples to add strata value to

TIP!

sample_strat() requires the sraster input to have an attribute named strata and will give an error if it doesn’t.

sample_strat(
  sraster = sraster, # input sraster
  nSamp = 200, # desired sample size
  access = access, # define access road network
  existing = e.sr, # existing sample with strata values
  mindist = 200, # minimum distance sample units must be apart from one another
  buff_inner = 50, # inner buffer - no sample units within this distance from road
  buff_outer = 200, # outer buffer - no sample units further than this distance from road
  plot = TRUE
) # plot

#> Simple feature collection with 400 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431190 ymin: 5337730 xmax: 438530 ymax: 5343210
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata     type     rule               geometry
#> 1       1 existing existing POINT (437770 5337930)
#> 2       1 existing existing POINT (437830 5341230)
#> 3       1 existing existing POINT (435710 5341910)
#> 4       1 existing existing POINT (432990 5341830)
#> 5       1 existing existing POINT (435050 5339450)
#> 6       1 existing existing POINT (438070 5341070)
#> 7       1 existing existing POINT (434350 5341950)
#> 8       1 existing existing POINT (433010 5340570)
#> 9       1 existing existing POINT (436970 5338190)
#> 10      1 existing existing POINT (434730 5342250)

The code in the example above defined the mindist parameter, which specifies the minimum euclidean distance that new sample units must be apart from one another.

Notice that the sample units have type and rule attributes which outline whether they are existing or new, and whether rule1 or rule2 were used to select them. If type is existing (a user provided existing sample), rule will be existing as well as seen above.

sample_strat(
  sraster = sraster, # input
  nSamp = 200, # desired sample size
  access = access, # define access road network
  existing = e.sr, # existing samples with strata values
  include = TRUE, # include existing sample in nSamp total
  buff_outer = 200, # outer buffer - no samples further than this distance from road
  plot = TRUE
) # plot

#> Simple feature collection with 200 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431230 ymin: 5337750 xmax: 438530 ymax: 5343170
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata     type     rule               geometry
#> 1       1 existing existing POINT (437770 5337930)
#> 2       1 existing existing POINT (437830 5341230)
#> 3       1 existing existing POINT (435710 5341910)
#> 4       1 existing existing POINT (432990 5341830)
#> 5       1 existing existing POINT (435050 5339450)
#> 6       1 existing existing POINT (438070 5341070)
#> 7       1 existing existing POINT (434350 5341950)
#> 8       1 existing existing POINT (433010 5340570)
#> 9       1 existing existing POINT (436970 5338190)
#> 10      1 existing existing POINT (434730 5342250)

The include parameter determines whether existing sample units should be included in the total sample size defined by nSamp. By default, the include parameter is set as FALSE.

method = "random

Stratified random sampling with equal probability for all cells (using default algorithm values for mindist and no use of access functionality). In essence this method perform the sample_srs algorithm for each stratum separately to meet the specified sample size.

#--- perform stratified sampling random sampling ---#
sample_strat(
  sraster = sraster, # input sraster
  method = "random", # stratified random sampling
  nSamp = 200, # desired sample size
  plot = TRUE
) # plot

#> Simple feature collection with 200 features and 2 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431110 ymin: 5337710 xmax: 438550 ymax: 5343210
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata type               geometry
#> x       1  new POINT (432750 5343210)
#> x1      1  new POINT (437850 5338650)
#> x2      1  new POINT (437690 5339490)
#> x3      1  new POINT (431190 5340570)
#> x4      1  new POINT (438210 5342410)
#> x5      1  new POINT (434510 5341570)
#> x6      1  new POINT (438270 5340950)
#> x7      1  new POINT (433150 5341710)
#> x8      1  new POINT (437530 5339230)
#> x9      1  new POINT (436590 5339570)

sample_sys_strat

sample_sys_strat() function implements systematic stratified sampling on an sraster. This function uses the same functionality as sample_systematic() but takes an sraster as input and performs sampling on each stratum iteratively.

#--- perform grid sampling on each stratum separately ---#
sample_sys_strat(
  sraster = sraster, # input sraster with 4 strata
  cellsize = 1000, # grid size
  plot = TRUE # plot output
)
#> Warning: [readStart] source already open for reading
#> Processing strata : 1
#> Warning: [extract] source already open for reading
#> Processing strata : 2
#> Warning: [extract] source already open for reading
#> Processing strata : 3
#> Warning: [extract] source already open for reading
#> Processing strata : 4
#> Warning: [extract] source already open for reading

#> Simple feature collection with 40 features and 1 field
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431270.7 ymin: 5337911 xmax: 438370.5 ymax: 5343035
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata                 geometry
#> 1       1 POINT (437747.6 5342834)
#> 2       1 POINT (438370.5 5338406)
#> 3       1 POINT (437803.1 5339229)
#> 4       1 POINT (436668.2 5340876)
#> 5       1 POINT (435533.3 5342523)
#> 6       1 POINT (434709.9 5341955)
#> 7       1 POINT (435588.8 5338918)
#> 8       1 POINT (435021.4 5339741)
#> 9       1 POINT (433886.5 5341388)
#> 10      1 POINT (432751.6 5343035)

Just like with sample_systematic() we can specify where we want our samples to fall within our tessellations. We specify location = "corners" below. Note that the tesselations are all saved to a list file when details = TRUE should the user want to save them.

sample_sys_strat(
  sraster = sraster, # input sraster with 4 strata
  cellsize = 500, # grid size
  square = FALSE, # hexagon tessellation
  location = "corners", # samples on tessellation corners
  plot = TRUE # plot output
)
#> Processing strata : 1
#> Warning: [extract] source already open for reading
#> Processing strata : 2
#> Warning: [extract] source already open for reading
#> Processing strata : 3
#> Warning: [extract] source already open for reading
#> Processing strata : 4
#> Warning: [extract] source already open for reading

#> Simple feature collection with 1187 features and 1 field
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431109.3 ymin: 5337702 xmax: 438510.8 ymax: 5343231
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata                 geometry
#> 1       1 POINT (438138.6 5337796)
#> 2       1   POINT (438222 5338072)
#> 3       1 POINT (438274.3 5339112)
#> 4       1 POINT (438357.7 5339388)
#> 5       1   POINT (438222 5338072)
#> 6       1 POINT (438024.3 5338283)
#> 7       1 POINT (438107.6 5338559)
#> 8       1 POINT (438138.6 5337796)
#> 9       1 POINT (438138.6 5337796)
#> 10      1 POINT (438357.7 5339388)

This sampling approach could be especially useful incombination with strat_poly() to ensure consistency of sampling accross specific management units.

#--- read polygon coverage ---#
poly <- system.file("extdata", "inventory_polygons.shp", package = "sgsR")
fri <- sf::st_read(poly)
#> Reading layer `inventory_polygons' from data source 
#>   `C:\Users\goodb\AppData\Local\Temp\RtmpKwkxG7\Rinst2ec85f71703f\sgsR\extdata\inventory_polygons.shp' 
#>   using driver `ESRI Shapefile'
#> Simple feature collection with 632 features and 3 fields
#> Geometry type: MULTIPOLYGON
#> Dimension:     XY
#> Bounding box:  xmin: 431100 ymin: 5337700 xmax: 438560 ymax: 5343240
#> Projected CRS: UTM_Zone_17_Northern_Hemisphere

#--- stratify polygon coverage ---#
#--- specify polygon attribute to stratify ---#
attribute <- "NUTRIENTS"

#--- specify features within attribute & how they should be grouped ---#
#--- as a single vector ---#
features <- c("poor", "rich", "medium")

#--- get polygon stratification ---#
srasterpoly <- strat_poly(
  poly = fri,
  attribute = attribute,
  features = features,
  raster = sraster
)

#--- systematatic stratified sampling for each stratum ---#
sample_sys_strat(
  sraster = srasterpoly, # input sraster from strat_poly() with 3 strata
  cellsize = 500, # grid size
  square = FALSE, # hexagon tessellation
  location = "random", # randomize plot location
  plot = TRUE # plot output
)
#> Processing strata : 1
#> Warning: [extract] source already open for reading
#> Processing strata : 2
#> Warning: [extract] source already open for reading
#> Processing strata : 3
#> Warning: [extract] source already open for reading
#> Simple feature collection with 179 features and 1 field
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431115.5 ymin: 5337705 xmax: 438554.6 ymax: 5343240
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata                 geometry
#> 1       1 POINT (432177.2 5343114)
#> 2       1 POINT (432546.4 5343207)
#> 3       1 POINT (432304.6 5342684)
#> 4       1 POINT (433067.1 5342912)
#> 5       1 POINT (432821.6 5342656)
#> 6       1 POINT (433560.7 5342583)
#> 7       1 POINT (435971.9 5343195)
#> 8       1   POINT (434004 5342693)
#> 9       1 POINT (434772.7 5342938)
#> 10      1 POINT (435483.1 5342823)

sample_nc

sample_nc() function implements the Nearest Centroid sampling algorithm described in Melville & Stone (2016). The algorithm uses kmeans clustering where the number of clusters (centroids) is equal to the desired sample size (nSamp).

Cluster centers are located, which then prompts the nearest neighbour mraster pixel for each cluster to be selected (assuming default k parameter). These nearest neighbours are the output sample units.

#--- perform simple random sampling ---#
sample_nc(
  mraster = mraster, # input
  nSamp = 25, # desired sample size
  plot = TRUE
)
#> K-means being performed on 3 layers with 25 centers.

#> Simple feature collection with 25 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431990 ymin: 5338010 xmax: 438450 ymax: 5343050
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>        zq90 pzabove2  zsd kcenter               geometry
#> 15509  5.86     55.9 1.38       1 POINT (435410 5342410)
#> 27274 12.90     75.0 3.30       2 POINT (431990 5341770)
#> 66746  9.86     64.4 2.57       3 POINT (438130 5339670)
#> 10797 17.50     88.4 4.35       4 POINT (438150 5342670)
#> 36036 20.40     92.7 4.62       5 POINT (435650 5341310)
#> 3725  20.10     71.7 6.12       6 POINT (438450 5343050)
#> 32617  4.58     29.5 1.06       7 POINT (434410 5341490)
#> 65471  7.54     13.9 2.02       8 POINT (435010 5339730)
#> 47146  3.07      6.7 0.60       9 POINT (434050 5340710)
#> 27558 15.90     73.2 4.14      10 POINT (437670 5341770)

Altering the k parameter leads to a multiplicative increase in output sample units where total output samples = \(nSamp * k\).

#--- perform simple random sampling ---#
samples <- sample_nc(
  mraster = mraster, # input
  k = 2, # number of nearest neighbours to take for each kmeans center
  nSamp = 25, # desired sample size
  plot = TRUE
)
#> K-means being performed on 3 layers with 25 centers.


#--- total samples = nSamp * k (25 * 2) = 50 ---#
nrow(samples)
#> [1] 50

Visualizing what the kmeans centers and sample units looks like is possible when using details = TRUE. The $kplot output provides a quick visualization of where the centers are based on a scatter plot of the first 2 layers in mraster. Notice that the centers are well distributed in covariate space and chosen sample units are the closest pixels to each center (nearest neighbours).

#--- perform simple random sampling with details ---#
details <- sample_nc(
  mraster = mraster, # input
  nSamp = 25, # desired sample number
  details = TRUE
)
#> K-means being performed on 3 layers with 25 centers.

#--- plot ggplot output ---#

details$kplot

sample_clhs

sample_clhs() function implements conditioned Latin hypercube (clhs) sampling methodology from the clhs package.

TIP!

A number of other functions in the sgsR package help to provide guidance on clhs sampling including calculate_pop() and calculate_lhsOpt(). Check out these functions to better understand how sample numbers could be optimized.

The syntax for this function is similar to others shown above, although parameters like iter, which define the number of iterations within the Metropolis-Hastings process are important to consider. In these examples we use a low iter value for efficiency. Default values for iter within the clhs package are 10,000.

sample_clhs(
  mraster = mraster, # input
  nSamp = 200, # desired sample size
  plot = TRUE, # plot
  iter = 100
) # number of iterations

The cost parameter defines the mraster covariate, which is used to constrain the clhs sampling. An example could be the distance a pixel is from road access (e.g. from calculate_distance() see example below), terrain slope, the output from calculate_coobs(), or many others.

#--- cost constrained examples ---#
#--- calculate distance to access layer for each pixel in mr ---#
mr.c <- calculate_distance(
  raster = mraster, # input
  access = access, # define access road network
  plot = TRUE
) # plot
#> |---------|---------|---------|---------|=========================================                                          

sample_clhs(
  mraster = mr.c, # input
  nSamp = 250, # desired sample size
  iter = 100, # number of iterations
  cost = "dist2access", # cost parameter - name defined in calculate_distance()
  plot = TRUE
) # plot

sample_balanced

The sample_balanced() algorithm performs a balanced sampling methodology from the stratifyR / SamplingBigData packages.

sample_balanced(
  mraster = mraster, # input
  nSamp = 200, # desired sample size
  plot = TRUE
) # plot

#> Simple feature collection with 200 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431110 ymin: 5337730 xmax: 438550 ymax: 5343230
#> Projected CRS: +proj=utm +zone=17 +ellps=GRS80 +units=m +no_defs
#> First 10 features:
#>                  geometry
#> 1  POINT (433050 5343230)
#> 2  POINT (433610 5343210)
#> 3  POINT (434450 5343210)
#> 4  POINT (435790 5343210)
#> 5  POINT (437410 5343150)
#> 6  POINT (431210 5343130)
#> 7  POINT (436190 5343130)
#> 8  POINT (437290 5343110)
#> 9  POINT (438310 5343110)
#> 10 POINT (438010 5343090)
sample_balanced(
  mraster = mraster, # input
  nSamp = 100, # desired sample size
  algorithm = "lcube", # algorithm type
  access = access, # define access road network
  buff_inner = 50, # inner buffer - no sample units within this distance from road
  buff_outer = 200
) # outer buffer - no sample units further than this distance from road
#> Simple feature collection with 100 features and 0 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431190 ymin: 5337750 xmax: 438510 ymax: 5343230
#> Projected CRS: +proj=utm +zone=17 +ellps=GRS80 +units=m +no_defs
#> First 10 features:
#>                  geometry
#> 1  POINT (434870 5343230)
#> 2  POINT (435650 5343210)
#> 3  POINT (435030 5343110)
#> 4  POINT (437370 5343030)
#> 5  POINT (432950 5343010)
#> 6  POINT (437870 5342950)
#> 7  POINT (434230 5342890)
#> 8  POINT (435950 5342870)
#> 9  POINT (433770 5342850)
#> 10 POINT (435050 5342850)

sample_ahels

The sample_ahels() function performs the adapted Hypercube Evaluation of a Legacy Sample (ahels) algorithm usingexisting sample data and an mraster. New sample units are allocated based on quantile ratios between the existing sample and mraster covariate dataset.

This algorithm was adapted from that presented in the paper below, which we highly recommend.

Malone BP, Minansy B, Brungard C. 2019. Some methods to improve the utility of conditioned Latin hypercube sampling. PeerJ 7:e6451 DOI 10.7717/peerj.6451

This algorithm:

  1. Determines the quantile distributions of existing sample units and mraster covariates.

  2. Determines quantiles where there is a disparity between sample units and covariates.

  3. Prioritizes sampling within those quantile to improve representation.

To use this function, user must first specify the number of quantiles (nQuant) followed by either the nSamp (total number of desired sample units to be added) or the threshold (sampling ratio vs. covariate coverage ratio for quantiles - default is 0.9) parameters.

#--- remove `type` variable from existing  - causes plotting issues ---#

existing <- existing %>% select(-type)

sample_ahels(
  mraster = mraster,
  existing = existing, # existing sample
  plot = TRUE
) # plot
#> Simple feature collection with 339 features and 7 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431150 ymin: 5337750 xmax: 438530 ymax: 5343210
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>      type.x zq90 pzabove2  zsd strata type.y  rule               geometry
#> 1  existing 5.34     63.9 1.20      1    new rule1 POINT (437770 5337930)
#> 2  existing 7.23     69.0 1.64      1    new rule1 POINT (437830 5341230)
#> 3  existing 9.57     78.1 2.51      1    new rule1 POINT (435710 5341910)
#> 4  existing 5.07      2.8 1.18      1    new rule1 POINT (432990 5341830)
#> 5  existing 3.94      7.6 0.79      1    new rule1 POINT (435050 5339450)
#> 6  existing 6.67     77.2 1.41      1    new rule1 POINT (438070 5341070)
#> 7  existing 3.22     17.1 0.59      1    new rule1 POINT (434350 5341950)
#> 8  existing 7.69     41.1 1.99      1    new rule1 POINT (433010 5340570)
#> 9  existing 8.22     89.4 1.68      1    new rule1 POINT (436970 5338190)
#> 10 existing 8.61     51.0 2.16      1    new rule1 POINT (434730 5342250)

TIP!

Notice that no threshold, nSamp, or nQuant were defined. That is because the default setting for threshold = 0.9 and nQuant = 10.

The first matrix output shows the quantile ratios between the sample and the covariates. A value of 1.0 indicates that the sample is representative of quantile coverage. Values > 1.0 indicate over representation of sample units, while < 1.0 indicate under representation.

sample_ahels(
  mraster = mraster,
  existing = existing, # existing sample
  nQuant = 20, # define 20 quantiles
  nSamp = 300
) # desired sample size
#> Simple feature collection with 500 features and 7 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431110 ymin: 5337710 xmax: 438550 ymax: 5343210
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>      type.x zq90 pzabove2  zsd strata type.y  rule               geometry
#> 1  existing 5.34     63.9 1.20      1    new rule1 POINT (437770 5337930)
#> 2  existing 7.23     69.0 1.64      1    new rule1 POINT (437830 5341230)
#> 3  existing 9.57     78.1 2.51      1    new rule1 POINT (435710 5341910)
#> 4  existing 5.07      2.8 1.18      1    new rule1 POINT (432990 5341830)
#> 5  existing 3.94      7.6 0.79      1    new rule1 POINT (435050 5339450)
#> 6  existing 6.67     77.2 1.41      1    new rule1 POINT (438070 5341070)
#> 7  existing 3.22     17.1 0.59      1    new rule1 POINT (434350 5341950)
#> 8  existing 7.69     41.1 1.99      1    new rule1 POINT (433010 5340570)
#> 9  existing 8.22     89.4 1.68      1    new rule1 POINT (436970 5338190)
#> 10 existing 8.61     51.0 2.16      1    new rule1 POINT (434730 5342250)

Notice that the total number of samples is 500. This value is the sum of existing units (200) and number of sample units defined by nSamp = 300.

sample_existing

Acknowledging that existing sample networks are common is important. There is significant investment into these samples, and in order to keep inventories up-to-date, we often need to collect new data for sample units. The sample_existing algorithm provides the user with methods for sub-sampling an existing sample network should the financial / logistical resources not be available to collect data at all sample units. The functions allows users to choose between algorithm types using (type = "clhs" - default, type = "balanced", type = "srs", type = "strat"). Differences in type result in calling internal sample_existing_*() functions (sample_existing_clhs() (default), sample_existing_balanced(), sample_existing_srs(), sample_existing_strat()). These functions are not exported to be used stand-alone, however they employ the same functionality as their sample_clhs() etc counterparts.

While using sample_existing(), should the user wish to specify algorithm specific parameters (e.g. algorithm = "lcube" in sample_balanced() or allocation = "equal" in sample_strat()), they can specify within sample_existing() as if calling the function directly.

I give applied examples for all methods below that are based on the following scenario:

See our existing sample for the scenario below.

#--- generate existing samples and extract metrics ---#
existing <- sample_systematic(raster = mraster, cellsize = 200, plot = TRUE)


#--- sub sample using ---#
e <- existing %>%
  extract_metrics(mraster = mraster, existing = .)

sample_existing(type = "clhs")

The algorithm is unique in that it has two fundamental approaches:

  1. Sample exclusively using existing and the attributes it contains.
#--- sub sample using ---#
sample_existing(existing = e, nSamp = 300, type = "clhs")
#> Simple feature collection with 300 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431166.7 ymin: 5337700 xmax: 438559.5 ymax: 5343229
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>      zq90 pzabove2  zsd                 geometry
#> 858 21.20     94.6 4.15 POINT (432611.9 5338202)
#> 733 21.10     95.4 3.72 POINT (432970.8 5339860)
#> 62  19.60     94.2 4.73 POINT (437390.7 5342232)
#> 354 22.20     92.2 5.65 POINT (435806.4 5340260)
#> 703 11.30     84.4 2.73 POINT (433572.2 5339296)
#> 463  7.19      6.0 1.94 POINT (434917.3 5340163)
#> 490  5.51      1.3 2.62 POINT (435074.1 5339550)
#> 646 26.30     95.8 6.59 POINT (433583.5 5340017)
#> 810 17.20     50.5 5.78 POINT (432358.1 5339704)
#> 9   13.40     90.1 2.96 POINT (438123.1 5342942)
  1. Sub-sampling using raster distributions

Our systematic sample of ~900 plots is fairly comprehensive, however we can generate a true population distribution through the inclusion of the ALS metrics in the sampling process. The metrics will be included in internal latin hypercube sampling to help guide sub-sampling of existing.

#--- sub sample using ---#
sample_existing(
  existing = existing, # our existing sample
  nSamp = 300, # desired sample size
  raster = mraster, # include mraster metrics to guide sampling of existing
  plot = TRUE
) # plot
#> Simple feature collection with 300 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431129.5 ymin: 5337700 xmax: 438556.3 ymax: 5343237
#> Projected CRS: +proj=utm +zone=17 +ellps=GRS80 +units=m +no_defs
#> First 10 features:
#>        zq90 pzabove2  zsd                 geometry
#> 91950  9.96     85.1 2.42 POINT (433669.2 5338407)
#> 91986 21.00     82.6 5.75 POINT (431984.7 5340653)
#> 91877 21.70     94.7 3.43 POINT (432765.5 5340918)
#> 91779 14.50     95.3 2.99 POINT (433811.5 5340402)
#> 91835  2.76      7.5 0.45 POINT (434233.4 5339008)
#> 91515  9.92     41.4 2.63 POINT (436732.7 5339191)
#> 91686  8.22     16.5 2.20 POINT (434965.7 5339718)
#> 91861 12.20     89.7 2.59 POINT (434823.4 5337723)
#> 91624 22.80     92.2 5.78 POINT (435843.6 5339094)
#> 91477 15.00     51.7 4.21 POINT (435227.7 5342266)

The sample distribution again mimics the population distribution quite well! Now lets try using a cost variable to constrain the sub-sample.

#--- create distance from roads metric ---#
dist <- calculate_distance(raster = mraster, access = access)
#> |---------|---------|---------|---------|=========================================                                          
#--- sub sample using ---#
sample_existing(
  existing = existing, # our existing sample
  nSamp = 300, # desired sample size
  raster = dist, # include mraster metrics to guide sampling of existing
  cost = 4, # either provide the index (band number) or the name of the cost layer
  plot = TRUE
) # plot
#> Simple feature collection with 300 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431155.3 ymin: 5337723 xmax: 438488.4 ymax: 5343237
#> Projected CRS: +proj=utm +zone=17 +ellps=GRS80 +units=m +no_defs
#> First 10 features:
#>        zq90 pzabove2  zsd dist2access                 geometry
#> 91929 20.20     95.4 3.29    171.4450 POINT (432862.5 5340028)
#> 91631 14.80     95.9 3.22    243.9676 POINT (435085.4 5340271)
#> 91485 12.20     80.1 3.13    187.7314   POINT (437659 5338122)
#> 91255 17.30     92.7 4.24    358.8087 POINT (437607.4 5341896)
#> 92002 18.40     91.6 4.38    195.5259   POINT (432683 5339199)
#> 91680 17.00     91.7 3.47    198.0760 POINT (435723.9 5338541)
#> 91629 22.00     92.6 4.92    101.3995   POINT (435302 5339935)
#> 91748 16.90     93.5 3.66    368.9830 POINT (434087.9 5340342)
#> 91975 17.00     92.1 3.97    579.6924 POINT (433284.4 5338635)
#> 91690  3.74     37.2 0.75    156.6058 POINT (434315.9 5340727)

Finally, should the user wish to further constrain the sample based on access like other sampling approaches in sgsR that is also possible.

#--- ensure access and existing are in the same CRS ---#

sf::st_crs(existing) <- sf::st_crs(access)

#--- sub sample using ---#
sample_existing(
  existing = existing, # our existing sample
  nSamp = 300, # desired sample size
  raster = dist, # include mraster metrics to guide sampling of existing
  cost = 4, # either provide the index (band number) or the name of the cost layer
  access = access, # roads layer
  buff_inner = 50, # inner buffer - no sample units within this distance from road
  buff_outer = 300, # outer buffer - no sample units further than this distance from road
  plot = TRUE
) # plot
#> Simple feature collection with 300 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431155.3 ymin: 5337712 xmax: 438559.5 ymax: 5343237
#> Projected CRS: +proj=utm +zone=17 +ellps=GRS80 +units=m +no_defs
#> First 10 features:
#>        zq90 pzabove2  zsd dist2access                 geometry
#> 91333  7.84     31.4 1.92   143.46549 POINT (435179.2 5342710)
#> 91607 13.40     79.9 3.69   282.74638 POINT (433187.4 5339524)
#> 91294  6.88     14.2 1.77    71.61624 POINT (438006.6 5339060)
#> 91647 12.30     94.9 3.43    63.13478 POINT (432623.2 5338923)
#> 91449  8.83     77.4 2.19   110.67972 POINT (434375.7 5341003)
#> 91355  4.33     10.1 1.03    67.79961 POINT (435167.9 5341989)
#> 91210  2.98      3.6 0.54    97.84857 POINT (437835.3 5342280)
#> 91340 17.40     88.7 3.56   115.09550 POINT (436419.1 5340416)
#> 91217 14.80     91.6 3.65   140.96187 POINT (438317.1 5341163)
#> 91228 10.60     96.0 1.95    80.54679 POINT (438040.6 5341223)

TIP!

The greater constraints we add to sampling, the less likely we will have strong correlations between the population and sample, so its always important to understand these limitations and plan accordingly.

sample_existing(type = "balanced")

When type = "balanced" users can define all parameters that are found within sample_balanced(). This means that one can change the algorithm, p etc.

sample_existing(existing = e, nSamp = 300, type = "balanced")
#> Simple feature collection with 300 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431106.9 ymin: 5337700 xmax: 438559.5 ymax: 5343237
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>     zq90 pzabove2      zsd                 geometry
#> 4  17.10     71.3 3.680000 POINT (438182.9 5343218)
#> 5  19.50     93.6 5.190000 POINT (438556.3 5342269)
#> 14 16.10     83.8 3.190000 POINT (438171.6 5342497)
#> 17  5.05      5.3 1.710000 POINT (437846.7 5343001)
#> 20 22.10     91.7 4.770000 POINT (438436.7 5341716)
#> 27  2.39      2.3 0.350000 POINT (437678.5 5342893)
#> 28  4.73     64.5 1.050000 POINT (437570.2 5343061)
#> 29 23.40     83.9 8.429999 POINT (437461.9 5343229)
#> 34  2.98      3.6 0.540000 POINT (437835.3 5342280)
#> 38 17.40     98.8 2.260000 POINT (437402.1 5342953)
sample_existing(existing = e, nSamp = 300, type = "balanced", algorithm = "lcube")
#> Simple feature collection with 300 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431110 ymin: 5337709 xmax: 438556.3 ymax: 5343226
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>     zq90 pzabove2  zsd                 geometry
#> 4  17.10     71.3 3.68 POINT (438182.9 5343218)
#> 5  19.50     93.6 5.19 POINT (438556.3 5342269)
#> 8  19.20     89.8 5.78 POINT (438231.4 5342773)
#> 19 21.30     81.5 5.96   POINT (438545 5341548)
#> 32 15.80     82.1 4.04 POINT (438051.9 5341944)
#> 33 11.00     55.0 2.80 POINT (437943.6 5342112)
#> 34  2.98      3.6 0.54 POINT (437835.3 5342280)
#> 39 24.40     96.8 7.55 POINT (437293.8 5343121)
#> 40 25.70     97.2 6.75 POINT (438533.7 5340827)
#> 43 11.30     76.3 3.01 POINT (438208.8 5341331)

sample_existing(type = "srs")

The simplest, type = srs, randomly selects sample units.

sample_existing(existing = e, nSamp = 300, type = "srs")
#> Simple feature collection with 300 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431166.7 ymin: 5337723 xmax: 438522.4 ymax: 5343237
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    zq90 pzabove2   zsd                 geometry
#> 1  20.4     65.6  7.06 POINT (432246.6 5343200)
#> 2  10.7     74.7  2.96 POINT (437610.5 5338567)
#> 3  13.5     23.1  4.44 POINT (435133.9 5339826)
#> 4  19.6     89.9  5.15 POINT (435817.7 5340981)
#> 5  19.8     94.7  5.40 POINT (433452.6 5338743)
#> 6  20.9     72.3  6.66 POINT (431465.8 5342935)
#> 7  12.0     77.0  3.18 POINT (437536.2 5340898)
#> 8  10.2     50.5  2.71 POINT (435339.2 5338769)
#> 9  17.4     84.3  3.74 POINT (433449.4 5342072)
#> 10 29.0     86.9 10.67 POINT (433344.2 5338911)

sample_existing(type = "strat")

When type = "strat", existing must have an attribute named strata (just like how sample_strat() requires a strata layer). If it doesnt exist you will get an error. Lets define an sraster so that we are compliant.

sraster <- strat_kmeans(mraster = mraster, nStrata = 4)

e_strata <- extract_strata(sraster = sraster, existing = e)

When we do have a strata attribute, the function works very much the same as sample_strat() in that is allows the user to define the allocation method ("prop" - defaults, "optim", "manual", "equal").

#--- proportional stratified sampling of existing ---#
sample_existing(existing = e_strata, nSamp = 300, type = "strat", allocation = "prop")
#> Simple feature collection with 299 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431110 ymin: 5337709 xmax: 438522.4 ymax: 5343226
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata zq90 pzabove2  zsd                 geometry
#> 1       4 17.6     85.0 4.09 POINT (433400.9 5342516)
#> 2       4 19.0     93.0 4.09 POINT (436273.6 5341750)
#> 3       4 15.0     51.7 4.21 POINT (435227.7 5342266)
#> 4       4 18.9     82.6 4.03 POINT (432175.5 5342203)
#> 5       4 17.1     96.3 3.60 POINT (434162.2 5338011)
#> 6       4 19.2     87.3 4.31 POINT (437368.1 5340790)
#> 7       4 14.5     98.6 3.14 POINT (436068.3 5342807)
#> 8       4 21.5     93.8 4.03 POINT (435866.2 5340536)
#> 9       4 17.0     92.1 3.97 POINT (433284.4 5338635)
#> 10      4 21.6     97.3 4.28   POINT (435926 5340812)

TIP!

Remember that when allocation = "equal", the nSamp value will be allocated for each strata.

We get 400 sample units in our output below because we have 4 strata and nSamp = 100.

#--- equal stratified sampling of existing ---#
sample_existing(existing = e_strata, nSamp = 100, type = "strat", allocation = "equal")
#> Simple feature collection with 400 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431129.5 ymin: 5337709 xmax: 438548.2 ymax: 5343237
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata zq90 pzabove2  zsd                 geometry
#> 1       4 15.9     93.6 4.28 POINT (435407.1 5343095)
#> 2       4 15.2     76.7 3.66 POINT (431671.1 5341878)
#> 3       4 12.9     91.1 3.19 POINT (433235.9 5339080)
#> 4       4 15.7     73.9 3.94 POINT (431443.1 5341493)
#> 5       4 14.7     93.5 3.11 POINT (431614.4 5338273)
#> 6       4 18.9     74.5 3.87 POINT (436826.5 5341630)
#> 7       4 20.3     95.8 4.19 POINT (436490.3 5341414)
#> 8       4 17.0     91.7 3.47 POINT (435723.9 5338541)
#> 9       4 17.9     89.7 4.04 POINT (431383.3 5341217)
#> 10      4 17.1     90.8 3.65 POINT (433680.5 5339128)
#--- manual stratified sampling of existing with user defined weights ---#
s <- sample_existing(existing = e_strata, nSamp = 100, type = "strat", allocation = "manual", weights = c(0.2, 0.6, 0.1, 0.1))

We can check the proportion of samples from each strata with:

#--- check proportions match weights ---#
table(s$strata) / 100
#> 
#>   1   2   3   4 
#> 0.2 0.6 0.1 0.1

Finally, type = "optim allows for the user to define a raster metric to be used to optimize within strata variances.

#--- manual stratified sampling of existing with user defined weights ---#
sample_existing(existing = e_strata, nSamp = 100, type = "strat", allocation = "optim", raster = mraster, metric = "zq90")
#> Simple feature collection with 99 features and 4 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: 431192.5 ymin: 5337780 xmax: 438559.5 ymax: 5343237
#> Projected CRS: UTM Zone 17, Northern Hemisphere
#> First 10 features:
#>    strata zq90 pzabove2  zsd                 geometry
#> 1       4 16.8     91.9 4.61   POINT (437069 5339408)
#> 2       4 15.6     70.6 3.92   POINT (435171 5338661)
#> 3       4 17.5     89.3 4.35 POINT (436202.5 5340753)
#> 4       4 15.3     69.1 3.64 POINT (432634.5 5339644)
#> 5       4 15.5     82.8 3.31   POINT (437103 5341571)
#> 6       4 19.5     93.7 3.19 POINT (436034.4 5340644)
#> 7       4 18.8     95.9 3.20 POINT (431696.9 5339991)
#> 8       4 21.3     92.8 3.33 POINT (432443.8 5338093)
#> 9       4 16.5     97.2 2.07 POINT (431192.5 5339666)
#> 10      4 16.2     85.8 3.54 POINT (433016.1 5342744)

We see from the output that we get 300 sample units that are a sub-sample of existing. The plotted output shows cumulative frequency distributions of the population (all existing samples) and the sub-sample (the 300 samples we requested).