Leaflet R - plotting raster with colour palette

Question

I am trying to plot a raster image using Leaflet, however I am having issues getting the colours right.

I've read in the raster like this and plot the the resulting raster variable.

library(raster)
library(leaflet)
ras<-raster("https://raw.githubusercontent.com/jodyphelan/jodyphelan.github.io/master/img/test.tif")
plot(ras)

But when I try to plot the raster using leaflet and the addRasterImage() function there are two things I notice. Firstly, it looks as its it is doing some kind of antialising between the colours. Secondly, it is putting some colours in that shouldn't be there.

pal<-colorNumeric(ras@legend@colortable,0:255,na.color = "transparent")
leaflet() %>% addRasterImage(ras,colors = pal)  

This is the raster produced using plot(): An this is the one produced using leaflet:

The questions are:

1. Is it possible to disable the antialising-like behaviour?
2. Why is it changing colours of some squares?

I'm also not entirely sure whether I am creating the pallete in the right way. Any tips on these questions?

Answer 1

It seems that leaflet is reprojecting the raster and that is causing some interpolation of cells.

Since your grid is a lat-long decimal degree grid, if you add project=FALSE you can get it to skip this step and the output has none of these interpolated pixels and all seems to be the right colour:

leaflet() %>% addRasterImage(ras,colors = pal) 

...yuck

leaflet() %>% addRasterImage(ras,colors = pal, project=FALSE)  

...yay!

If you have categorical data that isnt on a lat-long grid and you need to reproject it then you should use method="ngb" to select the nearest neighbour cell for reprojection otherwise values get interpolated between integers which are no longer valid categories (unless you round them first...).