使用 R 和 ggplot2 绘制具有自然颜色的自然地球 RasterLayer

Question

I am trying to create a map of some parts of South America using naturalearthdata. It is important for me that I create a ggplot2 object, so I can further modify and combine the plot with geom_sf and geom_points. THe last part, which I thought would be most challenging, is already done.

My main problem right now is that the raster-object is colored as if the integer value were an intensity. However, I would like the color to be as in the source: https://www.naturalearthdata.com/downloads/10m-natural-earth-1/10m-natural-earth-1-with-shaded-relief-water-and-drainages/

I use the following code:

library(tidyverse)
library(raster)

hills <- raster("./map_data/NE1_LR_LC_SR_W_DR.tif")

# for the larger dataframe, my RStudio keeps crashing
hills_df <- as.data.frame(hills, xy = T) %>% 
            filter(x >= -90) %>% filter(x <= -55) %>% filter(y <= 5) %>% filter(y >= -30)

ggplot(data = hills_df) +
  geom_raster(aes(x=x, y=y,fill=NE1_LR_LC_SR_W_DR))

The problem is that the data frame only has another column which is an integer value. By setting the aesthetics to fill=value , this integer gets mapped as magnitude. How can I implement the natural coloring as in the source?

Best, Tarotis

Answer 1

One way to achieve this is making a data.frame with coordinates of each cell of the rasterStack and values of the three layers combined via rgb , refer to this blog post .

However, geom_tile and even geom_raster appear to be very slow when as few as tens of thousands of pixels are involved. I use the following workaround:

1. Create a 3D matrix from rasterStack
2. Collapse it to two dimensions with rgb function (like explained in this SO answer )
3. Introduce matrix of RGB values to ggplot with annotate_raster

The code can be as follows:

library(tidyverse)
library(raster)

natearth_map <- 
 raster::stack('./map_data/NE1_LR_LC_SR_W_DR.tif') %>%      # import tiff as rasterStack
  crop(extent(-90, -55, -30, 5)) %>%                        # subset to desired extent
  as.array %>%                                              # convert to 3D array
  `/`(255) %>%                                              # switch to proportions to meet rgb() requirements
  apply(c(1, 2), function(x) rgb(matrix(x, ncol = 3))) %>%  # collapse layers to RGB colors
  annotation_raster(-90, -55, -30, 5)                       # make it a ggplot object

Then you can add it to your ggplot like this.

ggplot(data, aes(x, y)) + natearth_map + geom_point()