sample.grid.cell: Function to spatially sample points

In eliotmiller/ebirdr: Manipulate eBird data

Description

Spatially thin eBird records based on a user-defined grid.

Usage

sample.grid.cell(xxx, yyy, xlim = c(NA, NA), ylim = c(NA, NA), nx, ny,
  jitter = FALSE, size, replace = FALSE)

Arguments

xxx	Vector of longitude or E-W coordinates.
yyy	Vector of latitide or N-S coordinates.
xlim	Together xlim and ylim define a bounding box within which lookup occurs. I.e. all (xxx, yyy) pairs outside of this box are ignored.
ylim	Together xlim and ylim define a bounding box within which lookup occurs. I.e. all (xxx, yyy) pairs outside of this box are ignored.
nx	Number of grid cells in x direction.
ny	Number of grid cells in y direction.
jitter	Defaults to FALSE. Set to TRUE to randomize grid location.
size	Maximum number of points per cell to sample
replace	Defaults to FALSE. Set to TRUE to sample with replacement.

Details

This function generates a stratified sample over a regular grid of strata, i.e. grid cells. It is relatively efficient computationally speaking. The sample() function cannot take a sample larger than the population (in a cell) when 'replace = FALSE' If 'replace = FALSE' and the size parameter is larger than the cell populatuion size this function passes back a vector of length size, but it will contain only as many unique points in the cell and the rest of the entries will be NA's. Note that the arguments in this function are currently named things that already mean something in base R. We may want to change that.

Value

Index vector of selected locations.

References

Team eBird. Daniel Fink.

Examples

#Generate Random Points over 2D field 
nnn <- 1000
xxx <- runif(nnn, 0, 10)
yyy <- runif(nnn, 0, 10)
par(cex = 0.5)
plot(xxx, yyy, 
 xlim = c(-1,11), 
 ylim = c(-1,11), 
 pch=20, 
 col="red", 
 cex=0.5)
sgc <- sample.grid.cell(
 xxx, 
 yyy, 
 xlim = c(-1,3), 
 ylim = c(3,6),
 nx = 5, 
 ny = 2, 
 jitter = TRUE, 
 size = 1, 
 replace = FALSE ) 
length(sgc$sample.index)
sum(!is.na(sgc$sample.index))
points(
 xxx[!is.na(sgc$cell.number)], 
 yyy[!is.na(sgc$cell.number)], 
 col = "blue", 
 pch = 20, 
 cex = 0.5)  
points(xxx[sgc$sample.index], yyy[sgc$sample.index], pch=20, cex=1, col="green")

# Reconstruct the Grid that was used
for (iii in 1:(sgc$nx+1))
 lines( rep(sgc$bb[1,1] + (iii-1)*sgc$xwidth, 2), range(sgc$bb[,"yyy"]), col="grey")   
for (iii in 1:(sgc$ny+1))
 lines( range(sgc$bb[,"xxx"]), rep(sgc$bb[1,2] + (iii-1)*sgc$ywidth, 2), , col="grey")   

table(sgc$cell.number)
length(table(sgc$cell.number))
sort(unique(sgc$sample.index))
sort(unique(c(1:length(xxx))[!is.na(sgc$cell.number)]))

eliotmiller/ebirdr documentation built on May 14, 2019, 10:33 a.m.