I had a bit of fun with this one.
A few points though.
- It seems that this image also has some shadows applied to it, to make the topography stand out some more.
- There also seems to be some aliasing, the raw data most likely being rather larger than the image presented here
- It is a JPEG, so the pixel information is a bit lossy.
Still, what I did was take the line profile of the color scale bar and stored the values into an array.
I then calculated the euclidean distance of the RGB value of each pixel to the RGB values of the scale bar array and kept the index of the shortest distance.
However because of the shadows, there were a few pixels here and there with values that matched the background rather than the scalebar. For these I then ran a median filter to clean up what I could.
I am uploading the 16-bit tif so there might be no preview.
Original 16-bit tif: here
Here is also an example of the image using a Union Jack Lookup table, as suggested by @romainGuiet
If you want the macro code, it is available as a public GIST on GitHub https://gist.github.com/lacan/db89358ca5ba5d4308a52fc37cd05550