The core argument in this discussion of color schemes in maps of Hurricane Harvey rainfall makes sense to me – darkness and light have intuitive intensity meanings to us and it is a problem when a visualization violates those meanings and expects a key to do the work of remapping our understanding. But the suggestion to rework the map with an entirely different visualization technique based on a gradient of color (perhaps with a slight hue shift as well) rather than a rainbow scheme seems to miss what I, at least, find to be functional about the rainbow scheme. I’m accustomed enough to how the rainbow scheme maps to amounts of rainfall in the online weather maps I use that I now have a learned sense of what “dark green rainfall” is like as compared to “light orange rainfall”, etc. It goes back to what one thinks the purpose of the map is. In the linked article, the Washington Post version does a nicer job of showing historical rainfall data about the region. But if the main purpose of National Weather Service maps is to help people understand the weather conditions they are about to experience, with a historical map of accumulated rainfall in a case like this being an outlier use case, I feel like the rainbow scheme makes that easier for me as a user of the visualization than a version using only a light-to-dark shift in a single color.