Z - R中多边形（shapefile）的值

Question

my goal is to create a 3D-Visualization in R. I have a shapefile of urban districts (Ortsteile) in Berlin and want to highlight the value (inhabitants/km²) as a z-value. I have implemented the shapefile into R and coloured the value for desnity ("Einwohnerd") as followed:

library(rgdal)
library(sp)

berlin=readOGR(dsn="C...etc.", layer="Ortsteile")

berlin@data

col <- rainbow(length(levels(berlin@data$Name)))
spplot(berlin, "Einwohnerd", col.regions=col, main="Ortsteil Berlins", sub="Datensatz der Stadt Berlin", lwd=.8, col="black")

How it is posible to refer a certain polygon (urban district) to a z-value (inhabitant/km²) and how can I highlight this z-value?

Hope that someone will have an answer! Best regars SB

Thanks for the answer, but I am still on my wy to find out the best to use the density as z-value so that I can create a 3D Model. I found out that it is not possible to use the polygons of the shape but that it is possible to rasterize the polygon and to use a matrix for a different perspective and rotation.

Here is the code but the final 3D visualization looks not sharp and good enough. Maybe it would be better to calculate the the z-value in anther way so that the first values did not start so high or to use the center of the polygon and than to draw a column in z-direction:

library(rgdal)
library(sp)

setwd("C:\\...")
berlin=readOGR(dsn="C:\\...\\Ortsteile", layer="Ortsteile") 

col <- rainbow(length(levels(berlin@data$Name)))  
spplot(berlin, "Einwohnerd", col.regions=col, main="Ortsteil Berlins",                 
sub="Datensatz    der Stadt Berlin", lwd=.8, col="black")

library(raster)

raster <- raster(nrows=100, ncols=200, extent(berlin)) 

test <- rasterize(berlin, raster, field="Einwohnerd")

persp(test, theta = 40, phi = 40, col = "gold", border = NA, shade = 0.5)  

for(i in seq(0,90,10)){     
persp(test, theta = 40, phi = i, col = "gold", border = NA, shade = 0.5)
}

library(rgl)         
library(colorRamps)
mat <- matrix(test[], nrow=test@nrows, byrow=TRUE)
image(mat)
persp3d(z = mat, clab = "m")
persp3d(z = mat, col = rainbow(10),border = "black")
persp3d(z = mat, facets = FALSE, curtain = TRUE)

Answer 1

Is this what you had in mind?

library(ggplot2)
library(rgdal)           # for readOGR(...) and spTransform(...)
library(RColorBrewer)    # for brewer.pal(...)

setwd("<directory with shapefile>")
map <- readOGR(dsn=".",layer="Ortsteile")
map <- spTransform(map,CRS=CRS("+init=epsg:4839"))
map.data <- data.frame(id=rownames(map@data), map@data)
map.df   <- fortify(map)
map.df   <- merge(map.df,map.data,by="id")
ggplot(map.df, aes(x=long, y=lat, group=group))+
  geom_polygon(aes(fill=Einwohnerd))+
  geom_path(colour="grey")+
  scale_fill_gradientn(colours=rev(brewer.pal(10,"Spectral")))+
  theme(axis.text=element_blank())+
  labs(title="Berlin Ortsteile", x="", y="")+
  coord_fixed()

Explanation

This is a great question, in that it provides an example of a very basic choropleth map using ggplot in R.

Shapefiles can be read into R using readOGR(...) , producing SpatialDataFrame objects. The latter have basically two sections: a polygons section containing the coordinates of the polygon boundaries, and a data section containing information from the attributes table in the shapefile. These can be referenced, respectively, as map@polygons and map@data .

The code above reads the shapefile and transforms the coordinates to epsg:4839 . Then we prepend the polygon ids (stored in the rownames) to the other information in map@data , creating map.data . Then we use the fortify(...) function in ggplot to convert the polygons to a dataframe suitable for plotting ( map.df ). This dataframe has a column id which corresponds to the id column in map.data . Then we merge the attribute information ( map.data ) into map.df based on the id column.

The ggplot calls create the map layers and render the map, as follows:

ggplot:       set the default dataset to map.df; identify x- and y-axis columns
geom_polygon: identify column for fill (color of polygon)
geom_path:    polygon boundaries
theme:        turn off axis text
labs:         title, turn off x- and y-axis labels
coord_fixed:  ensures that the map is not distorted

A note on scale_fill_gradientn(...) : this function assigns colors to the fill values by interpolating a color palette provided in the colours= parameter. Here we use the Spectral palette from www.colorbrewer.org . Unfotrunately, this palette has the colors revered (blue - red), so we use rev(...) to reverse the color order (high=red, low=blue). If you prefer the more highly saturated colors common in matlab, use library(colorRamps) and replace the call to scale_fill_gradientn(...) with:

  scale_fill_gradientn(colours=matlab.like(10))+