Wikipedia:Reference desk/Archives/Miscellaneous/2012 August 22

Welcome to the Wikipedia Miscellaneous Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

August 22

Map colours

A question I was looking at yesterday had a link to File:Map of world by intentional homicide rate.png. This started me wondering why it is necessary to have variations of one colour throughout a map? In this case the colours are not to bad but I have seen maps with 10 to 12 shades of one colour making it very hard to decide what colour the area you are looking at is. File:Euskal Herria municipalities population.svg or File:Urbanized population 2006.png gives a better idea of what I'm talking about. I will say though that a lot of the maps I looked at on the Commons are like File:Prisoner population rate UN HDR 2007 2008.PNG and are very easy to read. CambridgeBayWeather (talk) 03:40, 22 August 2012 (UTC)

Wikipedia:WikiProject Maps may be a better place to raise this issue. People who work on that Wikiproject have established the standards and likewise by which maps are formatted. --Jayron32 03:42, 22 August 2012 (UTC)

That map also violates another principle of mine: "water is blue !" (or maybe blue-green, in the case of saltwater, to distinguish from pure blue fresh water). It's not too bad in this case, but the map at the Port Arthur massacre (Australia) article was completely incomprehensible to me due to not having blue water. StuRat (talk) 03:53, 22 August 2012 (UTC)

I'm happy with water is blue or "absense of land" is white. The Port Arthur map suffers by mixing two conventions in one diagram. The main has white water and the inset has blue water and white land. Chnaging it like this leads to confusion. -- SGBailey (talk) 17:43, 22 August 2012 (UTC)

And then they also use blue for some of the land on the main map, what a mess. StuRat (talk) 22:17, 22 August 2012 (UTC)

Also, when you want to show a range like this, the colors shouldn't be random at each step, but follow a logical progression, so you can tell which is more without reading the key. You could do the rainbow, though, so violet, blue-violet, blue, blue-cyan, cyan, cyan-green, green, yellow-green, yellow, yellow-orange, orange, red-orange, red. More steps can be added as needed. StuRat (talk) 03:59, 22 August 2012 (UTC)

It's not really about the maps on Wikipedia but about maps in general. StuRat has an answer but why the logical progression when it is hard to differentiate between areas of the map? CambridgeBayWeather (talk) 04:05, 22 August 2012 (UTC)

There are some competing concepts:

1) Being able to tell the difference between each adjacent shade in the key. So, just changing the amount of one color, like blue, limits how many shades you can differentiate. That's why I suggested the rainbow approach.

2) Being able to get the gist of the map without reading the key. Thus, "more blue" means "more X" and "less blue" means "less X" (or, in my rainbow example, "closer to red" means "more X" and "closer to violet" means "less X").

3) Certain things shown on maps have real life colors, so those should be preserved on the maps. Fresh water should be blue, saltwater blue-green, snow and glaciers should be white, desert should be yellow, etc. And for some reason weather maps seem to use green for rain, I suppose to distinguish from oceans, seas, lakes and rivers.

4) There are also some colored maps where the color doesn't mean anything, it's just used to differentiate regions. For example, a map of the US with different colors for each state so they stand out. There you want to use a large enough number of different colors so that you never have 2 adjacent states the same color. You don't have to worry about having a color continuum either, you can add in whites, browns, greys, etc., as needed (of course, you still need to ensure that the text color stands out, if written on those color blocks, but you can always invert the color, where needed). StuRat (talk) 05:50, 22 August 2012 (UTC)

"Large enough number of colo[u]rs" I do believe is precisely four. Franamax (talk) 05:59, 22 August 2012 (UTC)

Sort of. That means no edges will have the same color on both sides, but you could get vertices with the same color on more than one region sharing that vertex. This isn't always desirable. For example, in the case of US voting districts, Gerrymandering can actually lead to regions that reduce down to a virtual point and then open back up on the other side. Then there's also the issue of enclaves. StuRat (talk) 06:08, 22 August 2012 (UTC)

Well, since you had already mentioned labelling of the regions I'm not worried about the vertex problem, but you got me on the enclaves - we even have an article on how that could need more than 4 colours. Thanks for getting me to read the whole thing. ;) Franamax (talk) 06:34, 22 August 2012 (UTC)

You guys are not talking about choropleth maps, as far as I can tell. Using a rainbow spectrum for a single statistical variable is not a good idea. All the maps the OP posted about are choropleths. Pfly (talk) 08:08, 22 August 2012 (UTC)

I was with my points 1 and 2. Here's an example of using rainbow colors: [1] (this map is animated, but the same applies to static maps). I don't like the way they put violet next to red, though, I'd have put it at the other end. StuRat (talk) 08:32, 22 August 2012 (UTC)

Oh, you're right--rainbow spectrums are regularly used for temperature maps. They aren't choropleths in that they don't color by areas like states or counties but rather by contour line, or isolines, or whatever they are called. That is, the areas colored are defined by the data rather than something more arbitrary and predefined, like a political unit or a census tract. The rainbow thing works nicely in those cases because the spectrum is usually nicely ordered on the map—I mean you can get pockets and complexity, but orange areas are always between red and yellow areas, never next to greens or blues, and so on. But I do take back my statement about rainbow spectrums not being a good idea...or at least amend it by adding "for choropleth maps" (ie, maps with predefined area units), and adding "usually" to "not a good idea". I'm sure there are exceptions! Also, it's possible I'm misusing the term "choropleth" and perhaps it includes contour type maps, although I think I have it right. Temperature maps are, if I'm not mistaken, are isopleth maps. Pfly (talk) 21:33, 22 August 2012 (UTC)

Oop, one final point! That link I just used, to isopleth, is a little misleading as it says "...lines that depict a variable which cannot be measured at a point, but which instead must be calculated from data collected over an area." Obviously temperature data is measured at points. I think what that quote is trying to say is that values between the points are interpolated. Here is a somewhat clearer definition. Pfly (talk) 22:04, 22 August 2012 (UTC)

Also, the four color theorem is applicable to nominal maps, not cloropleths. Pfly (talk) 08:22, 22 August 2012 (UTC) (oops, I was totally wrong there—with nominal data you need as many colors as you have nominal groups!) Pfly (talk) 21:38, 22 August 2012 (UTC)

Well, it is a question about maps at Wikipedia, because the OP is noting that a map at Wikipedia sucks, and wants to know what it is the way it is. I was directing them to the place on Wikipedia where a map like the one he found may have been created. Asking the creator of the map why they made the choices they did would be more productive than asking us. StuRat's answer is good for offer options for desuckifying that map, but the OP still seems to be asking about the differences between maps at Wikipedia, rather than general principles of cartography. If a general answer is sought, it really comes down to principles of design itself: why certain colors or combinations of colors are chosen for a certain purpose. I agree with the premise that more contrast needs to be use when trying to use color to convey information, but what else of it? --Jayron32 04:11, 22 August 2012 (UTC)

That map type is called a choropleth map—but I have to say the first map on the "choropleth map" page seems to be a great example of bad color choices. In my experience most cartography textbooks suggest 5-7 classes/color shades as being ideal for most situations. If you go much beyond that it becomes increasingly hard to tell which class a shade represents. One of your examples, File:Euskal Herria municipalities population.svg, uses eight classes and I agree it is hard to read. I bet it would be much better with 5 or 6 classes, without much loss of information. Usually when mapping a single variable using shades of a single color is sensible and least likely to cause confusion. Using slightly different colors and shades can work too, and sometimes be even clearer if done well. Changing the hue of each end of the spectrum slightly can often make it easier to read the map without causing confusion. All too often color and shade choices are poorly made, resulting in confusing maps. Use of too many classes can also cause confusion, unless the map is quite simple. That one example, File:Urbanized population 2006.png, is "unclassed" (ed. at least I think it is—the more I look at it the more I suspect the map is classed by the legend isn't, which would be even worse!). The color shades are not grouped into classes, rather directly reflect the data value. At first this approach might seem like the best way to go, but usually it just makes map reading harder. Plus there are a number of perceptual issues—a particular shade surrounded by darker areas will appear lighter than the same shade surrounded by lighter areas. Unclassed maps are a bad idea, I think, unless you are *trying* to show how the underlying data is muddy. Sometimes you see so-called purple maps showing US election results. They are useful for showing how things are not as binary as simple blue and red maps might suggest, but less good for actually interpreting specific voting patterns. All that said, I'm not quite clear on the question. Single color schemes are useful for showing single statistic variations. Using multiple colors often makes things confused, as File:2004US election map.svg shows—without looking at the legend I would be hard-pressed to guess the meaning of the orange and brown, for example. The yellow leaps out as very bright, yet apparently is meant to show a middling rather than an extreme value. A two color shade "bipolar" scheme would have been much clearer, I think. Pfly (talk) 04:50, 22 August 2012 (UTC)

I have to say, that's a great, well informed answer Pfly. Thanks! --Jayron32 04:59, 22 August 2012 (UTC)

Heh thanks, I'm a cartographer wannabe... Pfly (talk) 05:16, 22 August 2012 (UTC)

I'm a lifelong map geek myself, but more as a consumer than a producer. That is, I look at maps like a restaurant customer looks at his food, not like a chef does. I know what I like, and I know what is good and what isn't, but I'll be damned if I know how they make it. --Jayron32 05:41, 22 August 2012 (UTC)

Also, for what it's worth, I only found File:Prisoner population rate UN HDR 2007 2008.PNG easy to read after looking at the key. Even then I wouldn't call it "easy". I had to look at the key again a few times to remind myself how the scheme went. Also, are there any countries colored in the 500-700 class? The color is very close to the 700+ red. Is Russia in the 500-700 class? I can't quite tell. Pfly (talk) 05:24, 22 August 2012 (UTC)

Thanks for all the replies, I know knoe more than I did yesterday. While I used a Wikipedia map as an example it could have been this map or or this. I find that when two shades are close to each other, both in colour and on the map, it is usually OK. However if they are far apart, on the map, it's hard to tell exactly which is which. CambridgeBayWeather (talk) 22:14, 22 August 2012 (UTC)

I think the example linked in Pfly´s post is "easy to read" because the key uses the colours of the visible spectrum, red > orange > yellow > green > blue (> purple) in the "correct sequence". As such it is almost intuitive to understand that red and blue are the extremes and (orange >) yellow (> green) represents some average population. Such a "normal" rainbow order seems much easier to interpret than having n shades of a single base colour to represent the various population ranges. --Cookatoo.ergo.ZooM (talk) 13:01, 23 August 2012 (UTC)

Disagree. Shades of a single color show a progression — if the grade is chosen correctly then it's easy to tell "light" from "intense." ROYGBIV may be "natural" in the sense that it's in a rainbow, but I don't find it nearly as visually intuitive. It's not at all clear, either, that the shades chosen are equally distant from from one another in terms of the spectrum. Three shades of red/orange, one yellow, three shades of green, then a totally different shade of blue. It isn't a good ROYGBIV spectrum at all, really, because it's changing saturation as well as hue in some cases (the dark green to a lighter blue is particularly egregious). --Mr.98 (talk) 14:07, 23 August 2012 (UTC)

I think both have their place. If you only want to show a few gradiations, let's say a half dozen or so, then shades of one color work well. But, if you need to show more than that, then the adjacent shades get too close together to easily distinguish. This is when you need to go to the rainbow. At first, you might just use a part of the spectrum, let's say varying from yellow to orange to red, then add in more colors as needed. StuRat (talk) 02:45, 24 August 2012 (UTC)

Here is a fairly good use of a rainbow spectrum for a choropleth map with many classes (10, plus a gray "no data" or "out of range"): [2] (plus it satisfies StuRat's "the sea must be blue!" criterion). I can tell shade from shade pretty well—with some slight difficulty with the light blues and greens. It's at the edge of being reasonably readable, I would argue. I think these kind of schemes can work, but one needs to take extra care when trying it. Mapmakers should ask themselves whether they really need more than 5-7 classes and make a few maps with different numbers of classes and classification schemes (there are many ways to split data into classes—what works best depends on your data). Often, I think, people uses too many classes for no good reason. Obviously one needs to think about the purpose of a map and what one is trying to communicate. Pfly (talk) 07:08, 24 August 2012 (UTC)

They seem to have only used orange through blue. If they added red to the one end and violet to the other, they could drop out a couple of those close intermediate colors. (However, if concerned that red and violet appear too close, they could just add red.) StuRat (talk) 07:20, 24 August 2012 (UTC)

I thought the same thing—that by using a redder red they might have been able to make the whole thing more readable. Pfly (talk) 07:28, 24 August 2012 (UTC)

Yes, and all the colors are rather dark. Brighter colors would be easier to distinguish. StuRat (talk) 07:30, 24 August 2012 (UTC)

Not dark so much as desaturated. There's something to that—overly saturated colors can seem to vie for attention. Perhaps this can aid in distinguishing colors, but it can be "eye-watering" too. There's a balance to be struck, which differs for every map. Perhaps this map went too far with desaturation, I'm not sure. Also, it's obvious that the map is meant to be viewed much larger—you can't even read the text in this low-res version. Perhaps—probably when seen printed as a poster it would be better overall. Pfly (talk) 07:37, 24 August 2012 (UTC)

Amelia Earhart's plane

Amelia Earhart's plane may have been found. This expedition cost $2,200,000 and there were other expeditions. Who pays for this? Is it worth the cost? Bubba73 ^{You talkin' to me?} 23:55, 22 August 2012 (UTC)

Not sure in this case, but that might be a reasonable cost to incur while making a documentary about the event (which would hopefully pay for itself). StuRat (talk) 23:59, 22 August 2012 (UTC)

How are we supposed to objectively answer your final question about whether it's worth it or not? Dismas|^(talk) 00:12, 23 August 2012 (UTC)

Isn't speculative investing what Wall Street is built on? 67.163.109.173 (talk) 01:14, 23 August 2012 (UTC)

You have a point. I'm interested to know, but I can think of better ways to spend millions of dollars. Bubba73 ^{You talkin' to me?} 00:28, 23 August 2012 (UTC)

Again, there isn't a lot of objectivity to that. For example, I've seen a lot of comments after images and articles about the Curiosity rover that say the money would be better spent here on Earth to house the homeless, feed the hungry, etc. And even then, if you were to take the money and feed needy people in Ethiopia or some other foreign (however the observer defines foreign) land then people would still say that there are needy people right here at home (again, however the observer defines home). If those people keep refining what the money should be spent on, it eventually narrows down to whatever cause is most important to them... Which may not be that important to someone else. Dismas|^(talk) 03:06, 23 August 2012 (UTC)

We could all think of "better" ways to spend the money. (For example: Give it to me, I'll soon give it a good home.) But this is not the place to chew the fat about this part of your question, because there is zero possibility of locating a suitable reference. And this is, after all, a reference desk. -- ♬ Jack of Oz ♬ ^{[your turn]} 04:13, 23 August 2012 (UTC)

Thank you. That is what I was getting at although you were more succinct about it. Dismas|^(talk) 04:35, 23 August 2012 (UTC)

I still don't know who is paying for it. ((spoiler follows!)) And after all of these millions spent (plus all the work) we may find out that she crashed into the Pacific Ocean. Bubba73 ^{You talkin' to me?} 04:57, 23 August 2012 (UTC)

From what I've read and heard on the radio as of the last six months or so, a big supporter of the search is TIGHAR. (I see that's a redirect, here is their official site]) Their funding, it looks like, comes mainly from the public. A search for "Amelia Earhart search funding" brings up quite a bit of info. A great deal of it referring to TIGHAR. Dismas|^(talk) 05:39, 23 August 2012 (UTC)

About the comparison to Curiosity on Mars - Curiosity is is learning new things and is trying to answer the big question about life on Mars. The Earhart thing will just verify what we have known for decades. Bubba73 ^{You talkin' to me?} 16:30, 23 August 2012 (UTC)

What have we known for decades, Bubba? That she's dead? That seems very likely anyway, as she'd be 115 now if she managed to survive all these years. But exactly where she died, and when, and under what precise circumstances - these have always been the big unknowns. A zillion tons of speculation and hypothesis are as nothing compared to a single ounce of actual evidence-based knowledge. -- ♬ Jack of Oz ♬ ^{[your turn]} 19:37, 23 August 2012 (UTC)

Is that knowledge worth millions of dollars (OK...it is to Hollywood...) ? At this point, it's trivia to finish off the story. At least space travel helped lead to technological advancements in hundreds of other areas. --Onorem♠Dil 19:45, 23 August 2012 (UTC)

Once again, what anyone subjectively believes about the ultimate value of this search, and whether or not it's worth the $2million+ they've spent on it, is entirely their own affair. Just why anyone would want to convince others about their point of view, here on a reference desk where we deal with things that are on the record and not merely opinion, escapes me.

Now, if the search were being paid for by public funds, one might have a legitimate cause to object. Even in that case, this reference desk would not be the place to voice such objections. But there's no evidence I'm aware of that the whole thing is anything but privately funded. -- ♬ Jack of Oz ♬ ^{[your turn]} 21:54, 23 August 2012 (UTC)

There had better not be one red cent of public money (at least American public money) going for this waste-of-time search for Earhart's plane. ←Baseball Bugs ^{What's up, Doc?} carrots→ 22:46, 23 August 2012 (UTC)

The Christian Science Monitor says: "Gillespie's group raised enough funds to embark on the nearly month-long voyage through individual and corporate donors, including funds from Discovery, which plans to document the trip and air it on cable TV in August, and $750,000 worth of free shipping from FedEx of the underwater science gear", but "the trip is nearly a half-million dollars short". Clarityfiend (talk) 23:41, 23 August 2012 (UTC)

As to why a government might take on such a search, about halfway into this article (subheading "Absolutely an Agenda) is speculation as to why one nation might do so. Mingmingla (talk) 00:19, 24 August 2012 (UTC)

As predictable as clockwork. I say that here would not be the place to raise any objections to the use of public funds for this search. And what does the very next editor do? Raise an objection. Great. -- ♬ Jack of Oz ♬ ^{[your turn]} 00:30, 24 August 2012 (UTC)

Indeed :-/ SemanticMantis (talk) 01:34, 24 August 2012 (UTC)

Or else what, Bugs? I'm sure you're familiar with WP:SOAPBOX. Please do us all a favor and give it a rest. SemanticMantis (talk) 01:34, 24 August 2012 (UTC)

And as predictable as clockwork, an editor comes from out of left field to do nothing but take shots at another user. Maybe you could do something useful, such as saving your lectures for other editors who made snippy comments to the OP. ←Baseball Bugs ^{What's up, Doc?} carrots→ 02:26, 24 August 2012 (UTC)

Bugs, your only contribution to this post is a personal political opinion and nothing else. SemanticMantis is correct. Keep such things to yourself. Nobody cares about your political opinions (or mine, either). That's not what we're here for. --Mr.98 (talk) 01:39, 26 August 2012 (UTC)

The nanny speaketh. ←Baseball Bugs ^{What's up, Doc?} carrots→ 06:59, 26 August 2012 (UTC)