Just how important are model layer colors? Very.

Which of the maps below appeal to you most?

  • HazWx

    Votes: 3 12.0%
  • NOAA NCEP

    Votes: 3 12.0%
  • AllisonHouse

    Votes: 4 16.0%
  • Wunderground

    Votes: 0 0.0%
  • WeatherBell

    Votes: 1 4.0%
  • StormVista

    Votes: 0 0.0%
  • TwisterData

    Votes: 14 56.0%

  • Total voters
    25
  • Poll closed .

Steve Miller

Owner Emeritus
Staff member
Supporter
Joined
Jun 14, 2004
Messages
1,793
Location
Moore, OK
Neat features like interactive mapping, layers, special views and model run completion status are included in many model sites but color tables and smoothing is what ultimately sells the goods; above and beyond the featureset.

It's clear that some maps are more appealing than others even though they display nearly the exact same information. Until higher resolution layers can be rendered more quickly, we are stuck with smoothing to make up for wider gradients. Smoothing is nice and all, but an eye for color and contrast is where the rubber truly meets the road.

There are plenty of guides available for color usage on maps. None of them make great model color tables and it's those who have thought outside the box that have the most stunning layer outputs. No one wants to stare at ugly maps and they sure as heck don't want to share them.

The maps below all depict the same information: GFS Total Precip at 18z on Tuesday. Which appeals most to you?
(COD is missing due to their site not working this morning)

HazWx


NOAA NCEP


AllisonHouse


Wunderground


WeatherBell


StormVista


TwisterData
 
Looking at all of the above, I voted for AllisonHouse. I like that it shows the amounts on the map (as opposed to having to match the color to the amount). I also like that the amounts aren't as cluttered as with HazWx (although maybe if I had the actual map available to me I could zoom in and it wouldn't feel so cluttered...I'm not sure). And finally, I like that AllisonHouse has the counties shown. I may feel differently if other plots were shown, so this vote only applies for this plot :)
 
I didn't choose any, because of the many choices, my preferred map (College of DuPage) isn't among them.

Colors are important, but for me the biggest selling point (and I've been saying this for years) is regional maps. Trying to figure out key placements of distinct features on a conus map is damn near imposible for me. With regional maps, I can actually see details within an area of an individual state, not just a generalized region.

Colors are important, but I can get used to pretty much any legend if forced to (such as the annoying blue hues CoD switched to for the upper end of their moisture scale a few years ago). Also, and this is just common sense IMO, I never pay for something I can get for free somewhere else.

EDIT: Of the maps displayed, I really don't have a favorite. However I don't like HazWx or AllisonHouse, because I don't like topo maps at all. IMO they blend too much with the graphics and just make everything look cloudy. I'd take any of the other choices (of these), simply because the background map is just state borders over solid color.
 
I think clarity is the most important aspect of mapping meteorological fields. You should be able to figure out quickly the value of the contour your are looking at each moment and what each color means without having to interpolate between other colors. For that reason, I have problems with HazWx, AllisonHouse, and WxBell because they have too many contours with very little difference in shade between adjacent contour values. I can't tell for sure exactly what the value of the field is at some locations because of how fine the color differences are. That's not always a big deal, but us old time mets like to know what range we are in without having to guess.

There has been research on how best to use color tables in meteorological products (http://member.nwas.org/sites/default/files/nwa_pubs/2014-JOM14.pdf, but you'll have to be an NWA member to read this particular one). The results are somewhat mixed, but indicate that a combination of just one or two base colors (with shades of value) is better than just one color going from light to dark shades and a total rainbow of colors. I tend to agree that unless you are plotting a field that is categorical in nature rather than numerical (a good example would be land use index or your favorite football team), too many colors is bad. Some of Earl Barker's maps do that, and I dislike his maps for that reason. None of the maps above do poorly with that, although WxBell has quite the range of colors spanned by the values. HaxWx and AllisonMaps blend colors together very well in that case.

The last aspect to clarity is resolution and technology. It's 2015, and most desktop monitors can now support HD sized graphics. Too bad for the mobile people...get a bigger screen. I don't tend to think highly of 800x600 or 1024x768 maps (like StormVista). I know browsers distort images that don't properly fit an image size (it will look fuzzy if the original image is larger than the browser window). It also depends on the browser settings Steve used when obtaining these graphics to make this post. The TwisterData and NCEP graphics look the same to me as I see on my desktops, and I think they're appropriately sized for my window. I like the capability to zoom in greatly using HazWx. In fact I think they could've gone further by adding zoom tiles. If you zoom in far enough you'll see what I mean (the pixelated nature of the images becomes apparent when you get in close).

So which one of these appeals to me the most? There is no clear winner, as there are at least two or three that I would prefer for one reason or another (discussed above). The ones I like the most, however, are TwisterData and HazWx.
 
Jeff,

JOM articles are only restricted to members for the year in which they are published. Thus the paper you linked is available to all. You can find it here: http://www.nwas.org/jom/articles/2014/2014-JOM14/2014-JOM14.pdf

(And if anyone is having problems getting the paper let me know. I'm the publications chair for the NWA and will work with you to make sure you have access.)
 
I think this is always going to be highly subjective depending more on which one you've been using the longest. You really do get used to a color scheme after a while to the point that transitioning to something else is remarkably cumbersome. Case in point - I've been a COD radar user for as long as they've had it, and I'm *still* getting used to their change in reflectivity color scale they did a couple of years ago. Storm cores of similar reflectivity look more "intense" (more oranges and reds) on the new scale than they did on the old one, even though the new scale has a finer gradient of colors than before.

As for models, I've been using the UCAR data and like them the best, and have resisted changing simply because I'm stuck in my ways. For pure aesthetics, I'd probably go with the NCEP and Twisterdata maps.
 
HazWx is the purtiest. But TwisterData communicates the most clearly and quickly of the bunch, and for that reason it gets my vote. State borders are clearly outlined, and so are contours and colors for rainfall amounts. Moreover, not including isobars, pressure centers, and so forth (e.g., NOAA NCEP) makes for a cleaner presentation.

I'm another who finds the sector views to be a strong feature of CoD. Similarly, I like F5 Data for its custom zooming and custom overlaying.
 
I voted for the TwisterData map for many of the same reasons others already have; clarity. The color scales are good, well defined, and are delineated well on the map (both by the contours and by the contour labels). No need to second-guess how much QPF an area might get, and there is minimal referencing to the legend thanks to the on-map labels, which makes reading the maps quick and easy.
 
Back
Top