Help us Create a New Weather Index

At the Capital Weather Gang (CWG), we strive to make weather information fun, interesting, and above all, relatable to daily living.
To that end, we're exploring the idea of creating a new index, or scale, that would fuse numerous bits and pieces of weather forecast information together with other relevant news information to convey how your overall "weather day" is going to go. We'd like your help to make it happen.
Think of the proposed index as a companion to the often-used wind chill and heat indexes, except that instead of being just for extreme wind or cold, this index would be for everything weather-related. I'll call it the "All Weather (+) Index" for the purposes of this column. The "+" sign denotes the potential to integrate non-meteorological variables, such as the beginning of a holiday weekend, that could change how we feel about the day despite inclement weather. Ultimately the index would be given a name with less of a corporate ring to it, like "Harold," to borrow a term from the world of improv comedy.
The index would be highly relevant, fun, and easy to use. It would be the "George Foreman Grill" of weather forecasts. I don't normally endorse products in this column, but man, that grill is a ridiculously convenient invention for busy professionals on the go. (Can I have a free one now?)
The concept for such an index, which was initially raised by the CWG's editor at washingtonpost.com, is aimed at capitalizing on the current state of weather forecasting and communications technology.
Meteorologists now have a formidable array of tools to employ to make their predictions, from Doppler radar to satellites to computer models. Sometimes it seems as if meteorologists have so many toys to play with that information overload is occurring. Forecasts can suffer in part because both forecasters and forecast consumers don't understand all of the information being given to them.
This phenomenon is similar to the concerns of aviation safety advocates who worry about the potential for new "glass cockpits" in aircraft to cause accidents by presenting so much information to pilots that they forget to focus first on flying the plane.
For example, although a weather geek like myself appreciates being given as much weather information as possible as part of a forecast, I doubt that everyday weather consumers need to know what the dew point temperature is, or that there is an "upper level disturbance" nearby, or even that a "bow echo" is approaching.
Instead, they need to know what this information means for them. Will it be a bad hair day? Will it snow? Should they head to the basement?
The All Weather (+) Index would help answer some of these questions. But figuring out what goes into this index and how to implement it is tricky. All of us at the Capital Weather Gang (CWG) are working on this, and we'd like your advice.
We have yet to decide what form the index would take. That's a pretty basic issue, I realize. But our failure of imagination in this regard hopefully won't prove fatal to this project. I mean, Steve Jobs must have a few moments of "idea block" before coming up with another directive to create an even smaller iPod, right?
Here are some of the issues we're grappling with:
Should the All Weather (+) Index consist of a number, such as an overall score? Or should it be a graphic, like a whole new set of those typical weather symbols you see on the five-day-forecast? Or would it be something entirely new and TBD, like the weather itself?
The index would be subjective in many ways, and potentially simplistic. But it doesn't have to pass the peer review process of an American Meteorological Society journal; it just has to be understandable and relevant to our site's readership (that's you).
Perhaps a meta-weather index would initially be crafted from a list of more distinctive adjectives to describe the weather. There is certainly room to explore this option, and some weather forecast outlets are already being more creative with their language. For example, Chicagoist recently published the following headline: "Wind Chill to join Wintry Mix at "Things That Suck" Convention."
Or along a different line of reasoning, a new index or scale could use a numerical baseline. Each day's weather would be assigned an overall number score on a scale of 0 to 100. A high temperature that does not go above freezing, for example, could knock the day's score back 20 points. That overall number could then be reported raw, or be converted into an adjective or other metric based on a formula or subjective judgment.
The problem with this method is that there are so many variables to consider when determining the implications of a weather forecast. Calculating the day's index would involve so many numbers that it could very quickly become a matter of creating an entirely new computer model to decipher a forecast that was itself generated in part from the output of other computer models. Yikes!
Another idea is to utilize a symbol system, which might prove useful as long as it doesn't veer into using emoticons.
"Tomorrow's forecast: ;) " doesn't add much value to weather information, although it would indicate that the forecast is for a sarcastic day, which could spell trouble.
So, tell us: What are you currently getting from our weather forecasts that you aren't getting elsewhere? What is still missing? How do you see a meta-weather index or scale as adding value to your life? Do you think it would be too simplistic?
Send your ideas of an index, including what variables it would include, how it would be calculated, and what the symbols should be via our comments section and/or via email. If we use your idea, we'll send you a free copy of the beautifully illustrated Washington Weather book by CWG photographer Kevin Ambrose.
By Andrew Freedman |
February 25, 2008; 11:00 AM ET
Capital Weather Gang
,
Freedman
Previous: Forecast: Brief Warmth Before Chill Returns |
Next: CommuteCast: Partly Cloudy and Mild
Posted by: hobbes | February 25, 2008 11:27 AM
Thanks Hobbes for your suggestions. I agree that a numerical scale could be a good starting point. It might be fun to then take the number score and filter it through a set of adjectives that match up with the numbers (a day with a numerical score of "55" might be called "decent" for example). This would require a robust way of calculating the number scores and a list of creative and accurate adjectives. This would be similar to how a day is called "partly" or "mostly" cloudy, but would encompass more weather parameters than just cloudiness.
Posted by: Andrew Freedman, Capital Weather Gang | February 25, 2008 11:44 AM
I like the idea of an "infographic" similar to your "nice day stamp" and "Snow lover's crystal ball" but expanded to include a variety of types of weather days at-a-glance. Then one could continue to read the full report for the in-depth coverage that you all so reliably give.
Posted by: ShawnDC | February 25, 2008 11:52 AM
What about a somewhat-personalized scale?
I envision this something along the lines of the build-your-burrito calorie counter that Chipotle's website has.
Questions like "How much does wind bother you" or "Do you hate being cold?"
"Does snow make or break your day?"
Ooh -- here's a good one: "Are you planning on driving today?"
I would envision kind of a real-time updating number based on what you input.
You would have your baseline number that would adjust based on people's input. It could just be a little embedded applet on the site. Or, if you want to get fancy, have people register their weather likes and not-likes, and e-mail their personalized index.
Posted by: Jake in Reston | February 25, 2008 11:59 AM
I would go with a simple 1 to 5 scale representing how much impact the weather will have on your life - key would be assume that most folks know it is cold in winter, hot in summer.
1 - no need to check any weather sites, no coat or umbrella needed, all events go ahead as planned
2 - might need a jacket, or you may sweat a little walking to your car depending on the season - most events go ahead
3 - Check the forecast if you plan on being outdoors for more than a few minutes - events with young children may be impacted. Drive times may lengthen.
4 - Significant weather anticipated. Many events will be cancelled. Driving is going to be impacted. Plan accordingly
5 - Panic! Major hurricane or snow/ice. Expect to be without power, etc.
Posted by: MDScot | February 25, 2008 12:17 PM
This is a really cool topic. I'm not sure a number by it's self would be enough information. But maybe a color & number combination. Even if the color is an indication of the season. Blue /Winter, Yellow/Spring, Green/Summer, Orange/Autumn
I think you would need to incorporate a lot of rules for augmenting the numerics based on when it happens. For example a big bump up for snow at Christmas and a big decrease for snow at Easter. Where on any other particular day snow wouldn't have such a large effect.
Posted by: John - Burke | February 25, 2008 12:18 PM
Jake, that's an interesting idea. I'll forward your comments to our site's technical folks to see what they think about the feasibility of doing something interactive like that. Thanks!
Shawn DC, a graphic is definitely in the running, thanks for your suggestion.
Posted by: Andrew Freedman, Capital Weather Gang | February 25, 2008 12:19 PM
I like MDScot's simplicity but assuming I can at least remember yesterday's weather-feel (was it cold, hot, humid, etc.), how about something that tells me how much today is going to be like yesterday. Cause if yesterday is like today, I'll just act (in a weather sense) like I did yesterday. Maybe add a trend over several days - if each day is gettng warmer and today will be the same, I know it will be warmer than yesterday.
Posted by: Steve | February 25, 2008 12:24 PM
The index needs to be seasonally sensitive. That is, a day with a high of 30 would be ideal in many people's minds on Dec. 24 if folks want to get into the winter spirit and dream of a white Christmas. Even rather extreme low temps can be considered a bonus in the winter for those who might want to ice skate on the C&O canal, for example. In the summer, a day with a high of 70 might not be agreeable for those who want to lie on the beach or by the pool, but would be a welcome respite if it came after a solid week of 90-plus days.
Other variables depending on the season:
a string of 70-degree temps early in March, for example, might be appealing to many folks, but would that mess up the cherry blossom schedule down the line? A day of soaking rain in July might be a pain, but wouldn't it be welcome if it came in the middle of a drought?
Posted by: jmbethesda | February 25, 2008 12:49 PM
MDScot's method is probably the way to go. The simpler you keep it the better it will be in the long run. However, Jake in Reston's suggestion cannot be discounted.
The capitalweathergang readership may change (if it hasn't already) now that you are on WaPo, but I imagine most of the crowd who followed you at capitalweather.com were to some degree, weather hobbyists. And, by nature, people interested in the weather, for weather's sake, tend to enjoy extreme weather - whether that be thunderstorms, blizzard, hurricanes, etc.
"Nice" weather generally rates pretty low on a weather weenie's scale. :-)
As Jake pointed out,
Posted by: Mike @ Twinbrook | February 25, 2008 12:52 PM
As I was reading these I clued in on the "some of us are weather hobbyists". For those of us some of the things that would make a day a "negative" one for the general public would make it a "positive" for us. For example, a day when thunderstorms were forecast would have a negative impact on driving, outdoor activities, etc for almost everyone but I'd be in heaven.
Posted by: Ana B | February 25, 2008 1:01 PM
I think everyone has some great ideas. I'll be thinking of some myself. This site is already good.
Off topic, The GFS next 12 days sure does look cold and active...:).Especially 3/4.
Posted by: StormChaser | February 25, 2008 1:10 PM
I think you all need to build an artifical weather dome with a machine that produces winter weather. If we want a big snow, we pay to get snowed on in the dome...jk.:)
I think your site is most informative already. Maybe post some graphical weather maps easily accessible on homepage
Posted by: skinsfn | February 25, 2008 1:22 PM
Don't forget gardeners in your mix of users - we keep looking for rain in the forecast to know when to plant, and when we'll have to allow time to water.
Seriously, I like "simple" (and I'll happily plan to read the full report), but I'm wary of anything which presumes an emotional response to a given weather condition. Specifically, something like giving a high positive score to snow on 12/24 is a bad idea - 1) not everyone celebrates Christmas, and 2) of those that do, some will be trying to get out to services, to do last-minute shopping, or to travel to family. That subgroup would be seriously put out by heavy snow for the holiday! It's fine to code that people will need a raincoat or a jacket, but beyond that, let's keep it objective.
Posted by: Hymncat | February 25, 2008 1:39 PM
I suggest doing a different scale for each type of person: high scores for rainy days for a pessimist, low scores on a 20 degree snowy day for an optimist, etc.
Of course, I'm an optimist and think a 20 degree snowy day deserves a high score :)
Posted by: Kenny | February 25, 2008 1:40 PM
More great suggestions from those who have commented so far. Keep 'em coming!
I think that while many of our readers are weather geeks, a scale should be devised to have the broadest possible application. Although weather enthusiasts would love a good snowstorm pretty much no matter when it hits, most of the general readers who come to us (and who we want to serve) may not be so charitable to the white stuff.
So, it's a matter of balancing things out and not assuming that everyone will have a particular emotional response to every weather situation, as some of you have pointed out.
Posted by: Andrew Freedman, Capital Weather Gang | February 25, 2008 2:20 PM
I think that Steve's idea above has alot of merit. Using a scale like colder than yesterday, warmer than yesterday, rainier than yesterday would have alot of appeal to people who don't pay alot of attention to specifics of the weather. Perhaps, geekily, you could do yesterday++ or yesterday--.
Of course someone who just got of a plane from the other side of the country might have some difficulties with that.
Posted by: doug | February 25, 2008 2:39 PM
I like the idea of a simple scale type format. 1 - 10, with 10 being what most people would consider being a 'good' day, or something similar. I would also include a short descriptor of what the primary reason(s) the score is what it is - like "9 - Sun" or "3 - Rain, Cold".
Posted by: Tim in Kalorama | February 25, 2008 2:53 PM
Here is an idea:
http://www.crh.noaa.gov/pah/
Toward the middle of the page is a graphic that says: "TODAY'S QUAD STATE OUTLOOK." It lists things like "Thunderstorms" and "Lightning" and "Hail" and stuff.
I was thinking that you guys could do that, but expand it passed severe weather. This would be a good idea because there are a lot of readers out there that aren't as knowledgeable about weather as some of the other visitors here.
Here would be an example:
Temps (Green Light)
Winds (Yellow Light)
Precip:
Snow (Black - None)
Rain (Red Light)
Thunder (Yellow Light)
Severe (Yellow Light)
Something like that. It would be able to give everyone a quick overview of the weather "threats" (or lack thereof) for the day.
Posted by: weatherdudeVA (Lake Ridge) | February 25, 2008 3:05 PM
give grades, A+, B, C-, ect. also, for days where snow is in the forecast, make a snow haters index and a snow lovers index (Forecast: 2 feet of snow with a high of 23, Snowlovers index: A+++, Snowhaters index: F-
Posted by: sam | February 25, 2008 3:11 PM
It would be nice to do an index of how these continual dance night rain forecasts will impact the next dance!
Posted by: El Bombo | February 25, 2008 4:26 PM
I love this concept, Andrew. Different people like different weather. Maybe you could have a set of icons that react in different ways to make a picture. The earthworm can sing and dance on a rainy day and stay in his sleeping bag underground when it's dry. The icons could appear or disappear according to season. The sailboat can sail when it is windy and the wolf will howl at the full moon. You could pick and chose the icons and activities that best describe the forecast using meterological measures; and program them to appear automatically when conditions are right. Say, the sailboat appears when the wind is 10-25 knots. The wolf only appears when it is a full moon. The earthworm dances when the chance of rainis over 50% and so on.
Posted by: Nell | February 25, 2008 4:54 PM
Maybe what you need is to call it a "Weather News-worthyness" scale. Under other typical scales, the problem as noted above is that some of us would rate a blizzard as a "10" day while others would rate it as a "1" day. However under a News-worthy scale it would be a "10 in everyone's book.
This scale could relate to the number of minutes the weather gets in the evening news or the number of column-inches it would get in the Post! This scale would also reflect "personal inconvenience" as news coverage increases with the weather's effect on our daily lives. Maybe a 1-5 scale would suffice...where a '5' means regular programming will likely be suspended.
Posted by: Rob-Rockville | February 25, 2008 5:22 PM
I would so love a three dimensional plot with 2/3 dimensions being temp and precip, and the third could be time if you wanted to see how things progress throughout the day, or it could be less complex and have a third static variable like wind, precip type, departure from day before or from averages. Then we could just look at these 3D images to get a feel for the how days are different from each other.
But that is probably too geeky for the average person. =-) I even just drew a little model on my notepad.
Posted by: tjd | February 25, 2008 5:23 PM
Everyone is weather sensitive but each of us enjoy different weather so a "one size fits all" index is tough. Why not create a small number of "weather types" so users could pick and choose the type that they are. Thus "storm lovers" would be a type and a forecast of 6-12" of snow would get a grinning snowman or a score of 1. "Every day should be Sunny and 72" types would get a 10 or a supermarket icon on days when it would be 30 and a dusting. Many people have pollen allergies or have trouble on poor air quality days, both which are part of "weather". 5-10 "Weather Type" categories might cover most of us and then we could have our Weather Index for our "type"
Posted by: rtwx | February 25, 2008 6:09 PM
rtwx: The concept of differentiating between people of different weather tastes was raised by a number of comments today, and I think it's an issue to explore. It may be difficult to manage several types of weather personalities and match it up with data though.
Tjd: I think what you're suggesting is great, but TV stations with certain graphics programs may be more adept at pulling that off. It's a great idea though, maybe we could accomplish this using Google Earth or something?
Posted by: Andrew Freedman, Capital Weather Gang | February 25, 2008 6:26 PM
I would suggest (much like the nice day stamp)
A simple index from -10 (Bad day, bad weather, bad commute. Think, Monday Downpour W/ Dense Fog etc) to a +10 (A 3 day weekend with tempatures expected to be in the lower 80's with low humidity and cloud free skies) would be the best. A simple 0 could suffice if it was going to be a not too horrible, not too great kinda day.
Or for a smaller range scale it down from
-5 to a +5.
(-5 being the worst, +5 being the best)
You could call it the moodcast scale.
Posted by: Havoc | February 25, 2008 8:10 PM
For temperatures, you could do it based on departure from average. Examples: In December, avg. 40 degrees. A 10 degree day would get a low score, as would a 70 degree day. Again, it all depends on your opinion of a "nice day" or an "enjoyable day". I like the interactive idea, hopefully your technical guys know how to pull it off!
Posted by: Model Monkey | February 25, 2008 9:00 PM
Hey,can you make a scale that tells me how to dress the kids? Like, "Heavy coats today" or "T-shirt weather, but pack an umbrella." At least I could blame you guys when my children complain about wearing their jackets! :)
Posted by: BusyMom | February 25, 2008 9:19 PM
I'm with Nell on the sailboat depicting winds of 10-25kt! Nell, do you sail, by chance?
Posted by: ~sg | February 25, 2008 9:54 PM
Nell, I just caught up with your idea, and I love it! It could really BE a picture, rather than just a bunch of icons, and you look to see what the various bits of it are doing! (Of course, the downside of such a picture is that it would require some study initially, to recognize all the components, and then a bit more than a quick glance to analyze the data each time.)
It could, however, be easy to see jacket or raincoat weather from such a picture. Have a couple of cartoon kids dressed appropriately, have a lake in the background for the sailboat, plus the usual sun/clouds/lightning icons.
The other downside is the complexity of creating and running the code... :-(
Posted by: hymncat | February 26, 2008 9:58 AM
The comments to this entry are closed.











I would suggest an overall numerical scale. For example, you take the temperature information for a day and give it a number from 1-10. 75 degrees or so would be a 10 (at least IMO). Next you take the overall cloudiness conditions of the day and rate them from 1-10. Cloudy, gray days get a 1, perfectly clear sunshiney days get a 10. Weather conditions get a numerical score, clear days a 10, rainy/snowy days (depending upon who you talk to) get a 1. Modifiers for precipitation could be that a day that has a 50% chance of precipitation would get a 5, or it could also depend more upon how much it's supposed to rain. Humidity should get it's own scale, because it can get pretty damn humid around here.
There could be other bonus modifiers like holidays, which could give an extra 5 points or whatever. So theoretically the above scale would give you a maximum of 30 points (or you could make humidity a 1-5 scale because it's only a part of the year issue here). Higher scores yield nicer days.
Certainly not perfect but it could be used as a starting idea?