You're right to point that out. I agree that a divergent scheme is not the best fit for the data. A simple gradient would have worked best, though it would have been less colourful.