In choisy/globcoverVN: Land cover data for Vietnam from the GlobCover project

knitr::knit_hooks$set(margin = function(before, options, envir) {
  if (before) par(mgp = c(1.5, .5, 0), bty = "n", plt = c(.105, .97, .13, .97))
  else NULL
})

knitr::opts_chunk$set(margin = TRUE, prompt = TRUE, comment = "",
                      collapse = TRUE, cache = FALSE,
                      dev.args = list(pointsize = 11), fig.height = 3.5,
                      fig.width = 4.24725, fig.retina = 2, fig.align = "center")

The globcoverVN package contains land cover data for Vietnam from the GlobCover project. It is made of a rasterLayer object that can be loaded with the getgcvn function and of the SpatialPolygonsDataFrame provinces object that contains, for each province, the percentage of land cover of each type. Below we describe in detail these data and their use.

Installation and loading

You can install globcoverVN from Bitbucket with:

# install.packages("devtools")
devtools::install_bitbucket("choisy/globcoverVN", build_vignettes = TRUE)

Once installed, you can load the package:

library(globcoverVN)

Usage examples

The raster file can be loaded by the getgcvn function:

landcover <- getgcvn()

The code of the different categories of landcover can be seen with the legend method:

show_legend(landcover)

The output can also be assigned to an data frame:

leg <- show_legend(landcover)

And then:

str(leg)

This raster file can be plotted as follow:

colors <- leg$color
plot(landcover, col = colors, legend = FALSE, axes = FALSE)
l <- length(colors)
x <- ceiling(l / 2)
sel1 <- 1:x
sel2 <- (x + 1):l
leg1 <- leg$code[sel1]
leg2 <- leg$code[sel2]
col1 <- colors[sel1]
col2 <- colors[sel2]
legend("left", legend = leg1, fill = col1, bty = "n")
legend("right", legend = leg2, fill = col2, bty = "n")
show_legend(landcover)

In addition to this RasterLayer resource, the package also provides SpatialPolygonsDataFrame resources that contains the polygons of the provinces of Vietnam at different points in time together with the proportions of the different categories of land cover for each province, in the form of attributes. These SpatialPolygonsDataFrame resources can be retrieved thanks to the getlandcover function:

prov2008 <- getlandcover(2008)
str(prov2008@data)

The variables whose names end with _pop correspond to percentages of land cover when correcting for the local human population density as available from the worldpopVN package. The user interested in seeing how this is computed can have a look at the data_creation script in the data-raw folder of the package sources. This script makes use of the lcsummary and lcpopsummary non-exported functions that are in the lcsummary.R and lcpopsummary.R files in the R folder of the package sources. One can verify that the percentages of land cover sum to 1:

rowSums(prov2008@data[, 2:21])
rowSums(prov2008@data[, 22:41])

You can also check that, for example, in Vietnam, there is no "Permanent snow and ice":

sum(prov2008@data$`220`)
sum(prov2008@data$`220_pop`)

This data can be visualize this way as for the example show below for the "Post-flooding or irrigated croplands (or aquatic)". Let's first make a palette of colors form RColorBrewer:

n <- 9
pal <- RColorBrewer::brewer.pal(n, "Blues")

Let's find a classes intervals definition:

library(classInt)
tmp <- classIntervals(prov2008@data$`11`, n = n, style = "quantile")
plot(tmp, pal = pal, main = NA)

Once we're satisfied with the class interval definition we can plot the map:

plot(prov2008, col = findColours(tmp, pal))

Or, with another definition of the classes intervals:

tmp <- classIntervals(prov2008@data$`11`, n = n, style = "jenks")
plot(tmp, pal = pal, main = NA)
plot(prov2008, col = findColours(tmp, pal))

The same thing, but with values weighted by the local human population density:

tmp <- classIntervals(prov2008@data$`11_pop`, n = n, style = "jenks")
plot(tmp, pal = pal, main = NA)
plot(prov2008, col = findColours(tmp, pal))

As another example, let's see the urban areas:

tmp <- classIntervals(prov2008@data$`190_pop`, n = n, style = "jenks")
plot(tmp, pal = pal, main = NA)
plot(prov2008, col = findColours(tmp, pal))

Which, again, makes quite some sense (apart maybe for Cao Bang and Hai Phong?).

choisy/globcoverVN documentation built on Aug. 22, 2019, 12:42 a.m.