Why is a GeoTIFF uploaded to MapBox as a tileset shown in black and white and not the original colors in the TIFF?

Question

When uploading a GeoTIFF file to Mapbox to create a tileset, it is being displayed in Black and White? I would expect it to have the same colors as the TIFF when viewed in any other viewer?

How do you create a tileset with a GeoTIFF to use it as a base map in MapBox?

My ultimate goal is to use this GeoTIFF and then dynamically lay lat/long points on it, but I need to get the map looking correct first.

Seems to work just fine in QGIS.

Answer 1

The following commands can be used to translate the GeoTIFF file. This information was obtained by MapBox support.

1. this will leave you with a raster tile with a black background.

gdal_translate -co TILED=TRUE -co COMPRESS=LZW -co BIGTIFF=YES -expand RGB INPUT.tif OUTPUT.tif

2. Make your raster tile transparent

rio calc "(asarray (take a 1) (take a 2) (take a 3))" --co compress=lzw --co tiled=true --co blockxsize=256 --co blockysize=256 --name a=INPUT.tif OUTPUT.tif

3. Make NoData values to zero.

rio edit-info --nodata 0 INPUT.tif

Before Conversion

After Conversion

Answer 2

If you're using R / raster package for this, you need to do two things:

1. Set colNA option of RGB to "black".
2. On raster write, set NAflag to 0.

For instance:

library(raster)

rgb_raster <- RGB(a_raster, colNA='black')

# Not sure if these are needed, but match the official example
options = c("COMPRESS=LZW", "BIGTIFF=YES", "TILED=TRUE", "BLOCKXSIZE=256",
            "BLOCKYSIZE=256")

writeRaster(rgb_raster, 
    filename='rgb_raster.tif', 
    datatype='INT1U',
    NAflag=0,
    options=options) 

The writeRaster call needs to be done separate from the RGB call because an error results if you set datatype there.