Jun 30 2008
Color and Temperature, Thoughts on Psycology and Information Transfer
Red is hot, and blue is cold. Seemingly everyone knows this. Even though blue light has a higher energy, and thus temperature, than red light. Watts has a recent post about Color and Temperature: Perception is everything. He questions, “Is it just me or does there appear to be a warm bias in the color temperature presentation of the majority of providers shown here?”
I have previously posted a couple thoughts on the use of color when presenting figures to the public; for example Using Color Contours to Improve Public Interpretation of the Temperature Record and Using Color to Visualize Decreases in Sea Ice Extent. Color is not just a way to make a boring figure pretty. The color has to add information. For instance, in his Color is Everything post, Watts shows current (at time of posting) national temperature figures from several sources.
In my opinion, there is one source that stands out. A lot of comments said they thought that the figure from NOAA image was best. I agree that it is better than average, but I disagree that it’s the best. The reason is that the NOAA map uses both red and green in its map. The image he displayed does not have any red on it, but there will definitely be instances where temperatures in one part of the country will be in the 70s (green) and others will be in the 100s (red).
The choice of using the “rainbow” colors has been presented in EOS in 2004. They show how using the rainbow color scheme can make it difficult or impossible for a not insignificant portion of its readers to correctly interpret its colors.
The map from the Weather Channel has the best use of color. It uses a wide dynamic range of colors, from blue to magenta), and it has good spatial resolution. I wish they had a color bar, but because of the many numbers on the map, that isn’t needed.
On the other hand, if we were to be using color to map a map look good, then the choice of colors on the Weather Channel are not the best. Again, I think there is a clear winner in the pretty category, and that is the map from IntelliWeather. It’s got a rainbow color scheme, where all the colors are visible, which is visually appealing. But in my opinion the colors do not add information to the plot. If I was located in Las Vegas, the color is yellow. The color bar represents this as ‘warm’. However, the Weather Channel map shows that the temperature in Las Vegas to be near 100.
A couple commenters pointed to a blog post written in 2002, which was reproduced as Coloring the Models: Climate Change through Color Change, which asserts that figures in a then recent government report ‘were “doctored” to distort public perceptions of climate change.’ I’ve taken the liberty of reproducing the key figure here.
Which of the two color schemes is the best? Clearly, both of them fail the pretty test. Although, I would consider the rainbow to look slightly better. However, because we have been conditioned to think that red equals hot and blue equals cold, this set of colors gives a vastly different impression than the figures that are nearly all red. Both of the set of figures show how the choice of color influences casual interpretation of figures.
The figures with the rainbow color scheme give a false sense of cooling. As was mentioned by several commenters in the perception thread, red is hot, blue is cold, but green is just right. When we look at the color bar for the rainbow figures, green is not “just right”; it represents a change of 5 degrees Fahrenheit. A much better scheme would have been to have the top red color to be +15F and the low blue color to be -15F. This would have put the “just right” green at 0F. This would have shifted the colors so that there would be a little green, but mostly the colors would have been yellows, oranges, and reds.
Instead, they decided to go with what appears to be a two color scheme, where the two different colors, blue and red, are used to differentiate negative and positive anomalies respectively. To indicate levels of anomaly, they changed the level of saturation. That is, “dark” reds indicate high levels of warming and “light” reds low levels of warming. This color scheme results in no noticeable blue regions on the map; there are no regions of projected cooling.
So does this change “distort” perception? I would argue that it does not. Red means warm right? So when there are temperature projections that indicate that there will be warmer temperature in the future, it is reasonable to use red.
One commenter was kind enough to point to this article about color choices in data presentation: Why Should Engineers and Scientists Be Worried About Color? And a related blog post. They show quite dramatically how the use of the rainbow color scheme can distort what known data looks like, and that using alternative color maps increases the amount of information conveyed.
An example of misleading color maps is when there is a known distinction in the data. They use elevation above sea level and depth below to illustrate how poor the rainbow scheme shows changes.
In the top image, it’s hard to tell exactly what is being displayed. The bottom image gives a clear contrast between areas above sea level and below sea level. This is because there was a clear color change that occurred when the altitude was equal to zero. The ocean is shades of blue and the land is shades of green. For the rainbow scheme, the distinction between above sea level and below sea level is masked with the color yellow.
This is exactly what was done with the temperature projections figure. There is a natural distinction between regions of projected warming and regions of projected cooling. The choice of only using two colors may also have been good. They write:
Stevens found that hue was not a good dimension for encoding magnitude information. Thus, we have followed Stevens, and have used luminance-based and saturation-based colormaps for representing continuous variations in data magnitude. In the case of ratio data or interval data with a threshold, we have used colormaps which capture the transition explicitly via a perceived discontinuity.
We are much better at distinguishing differences in luminance (brightness) and saturation than changes in hue (color). Thus the choice of using only 2 colors with changes in saturation as used in the 2000 National Assessment was good for the information that it attempted to show; temperature projections for the United States show large areas of warming and very little areas of cooling.
(No Ratings Yet)
Loading ...Related Posts:
6 Responses to “Color and Temperature, Thoughts on Psycology and Information Transfer”
To reduce spam, comments are automatically closed 30 days after the last comment. If you would like to comment on any closed thread, please use the contact form at the top of this page.
To me, the weather channel map seems inaccurate. It looks like the entire US is in a heat wave. There were worse maps at Anthony’s blog, but to me 66F in the UP shouldn’t look like a color I would most often find in licks of fire dancing on a burning log. Is 54F yellow too?
I don’t watch the Weather Channel. What color is 0F?
I don’t watch TWC either, so I have no idea. I was running with what I had available. They may switch colors throughout the year.
personally I prefer to download a netcdf and use panoply on it
has lot’s of different color schemes.
This is yet another recycled argument from denialists - I remember some years ago on Tech Central Station an essay that offered proof that AGW is a conspiracy, by presenting the fact that someone changed the color on a map.
There you have it, proof we all needed: the colors are too warm looking, therefore there is no globul warmin.
BTW, I map with ArcGIS and choosing colors that will make sense, be differentiable, and print on paper is hard. Some folk think about this, some don’t.
Best,
D
Over 50% of the male population has a serious RED/GREEN deficiency. In my case it approaches color blind. If you have relatively normal color vision, your interpretation of what the maps look like will differ radically from mine for physical reasons as opposed to mentally subjective reasons.
The maps being shown by Anthony, to look hot, have an OMINOUS look to them in my perception. It is only by looking at the numbers or a key that this OMINOUS look translates to HOT! In other words, dark colors are more connected for me with brooding, overcast, impending doom, calamity…, a SUBJECTIVE interpretation. The fact that it is red has a lesser impact. Often I will not be able to identify the actual colors and definitely not shades.
I definitely think they are UGLY. I see no advantage to those charts from my lack of color perspective.
Incidentally, all those spaghetti charts are MADDENING. It is extremely difficult for me to tell which one is what! Many times I am simply at a loss as to what I am observing.
Just remember the audience when crafting those charts. The ones with high contrast will give the widest audience the best experience.