1985: "Meteorologist"

[pogrmman]

Alt text: Hi, I'm your new meteorologist and a former software developer. Hey, when we say 12pm, does that mean the hour from 12pm to 1pm, or the hour centered on 12pm? Or is it a snapshot at 12:00 exactly? Because our 24-hour forecast has midnight at both ends, and I'm worried we have an off-by-one error.

I like this one! The questions the math guy poses are good especially given how forecasting is done today. It’s still amazing to me that we are able to predict the weather with any degree of accuracy — sure, the models aren’t all that good beyond a few days out, but it’s such a complicated system with so many variables that it’s amazing that we can get anything at all. And yet people still get fussy when they say there’s a 20% chance of rain, and it ends up raining :p

I really like watching when “forecasting” switches to “nowcasting” — it’s really neat. Like how the SPC changes stuff as a system is in progress. It seems to me like that’s where a lot of the difficulty in meteorology lies — sure, it’s not easy to figure out how to put together a coherent forecast from models, but it seems like accurately forecasting an event that’s basically in progress without necessarily having up to date model runs to rely on is hard.

[owenneil]

I have always wondered the answers to exactly the questions the mathematician raises.

[jovialbard]

I've always assumed that a 20% chance of rain means that they believe it will be raining for 20% of the hour/day with some degree certainty. So 5 hours in a row with 20% chance of rain means that there's maybe a 90% chance that 20% of each of those hours will involve rain. Basically, there's close to a 100% chance of rain if there's any rain in the forecast at all, the percentage just tells you how much, how hard, and whether there will be breaks in the rain. I also think this approach is consistent with human perception of what those percentages mean. If you see 50% chance of rain you'll tend to prepare for rain, we don't tend to assume that it means it's a coin toss between sunny and dry or raining cats and dogs.

[pogrmman]

It depends on the specific setup and specific day. It’s a mixture of the forecast confidence and the forecast coverage, iirc. I tend to just read the forecast discussions instead — sure, they’re a bit technical, but the usually give a good overview of the forecast for the area and their reasoning behind it.

[madaco]

I like this one!

[Heimhenge]

This one is funny. Reminded me of when I learned the difference between "partly sunny" and "mostly cloudy" (not synonyms as I thought). The new NWS website actually has a glossary to help with translating this jargon: http://w1.weather.gov/glossary/

Spoiler: They do not define "20% chance of rain".

[petercooperjr]

For what it's worth, the National Weather Service does have a page on the meaning of "Probability of Precipitation": https://www.weather.gov/ffc/pop

What does this "40 percent" mean? ...will it rain 40 percent of of the time? ...will it rain over 40 percent of the area?

The "Probability of Precipitation" (PoP) describes the chance of precipitation occurring at any point you select in the area.

How do forecasters arrive at this value?

Mathematically, PoP is defined as follows:
PoP = C x A where "C" = the confidence that precipitation will occur somewhere in the forecast area, and where "A" = the percent of the area that will receive measureable precipitation, if it occurs at all.

So... in the case of the forecast above, if the forecaster knows precipitation is sure to occur ( confidence is 100% ), he/she is expressing how much of the area will receive measurable rain. ( PoP = "C" x "A" or "1" times ".4" which equals .4 or 40%.)

But, most of the time, the forecaster is expressing a combination of degree of confidence and areal coverage. If the forecaster is only 50% sure that precipitation will occur, and expects that, if it does occur, it will produce measurable rain over about 80 percent of the area, the PoP (chance of rain) is 40%. ( PoP = .5 x .8 which equals .4 or 40%. )

In either event, the correct way to interpret the forecast is: there is a 40 percent chance that rain will occur at any given point in the area.

But that still doesn't completely address the length of the time-box within which that forecast holds, nor how the "forecast area" gets defined.

[pkcommando]

Reminded me of when I learned the difference between "partly sunny" and "mostly cloudy" (not synonyms as I thought). The new NWS website actually has a glossary to help with translating this jargon: http://w1.weather.gov/glossary/

When I was in my mid-teens and one of my brothers worked at a radio station, we did a little 'free talk' session where we goofed off and improved a bit of radio banter and interviews and recorded it to cassette. I don't remember the numbers now, but I played an expert giving the percentages of sun/clouds needed for each term. I wish I still had the tape to see how close I got.

At the time, though, I thought they were synonyms like you (and probably so many other people) did and just did the percentages because I thought it was funny if there were hard numbers there.

[jabrown371]

This is amazing. My brother:
- Has an applied mathematics degree
- Is getting his Masters in Linguistics
- Spent 1 year in between doing software development
My brother could be all three people at once!

[Heimhenge]

This is amazing. My brother:
- Has an applied mathematics degree
- Is getting his Masters in Linguistics
- Spent 1 year in between doing software development
My brother could be all three people at once!

Geez, I hope he never takes a job as a meteorologist.

[pogrmman]

This is amazing. My brother:
- Has an applied mathematics degree
- Is getting his Masters in Linguistics
- Spent 1 year in between doing software development
My brother could be all three people at once!

Geez, I hope he never takes a job as a meteorologist.

I hope he does Those would be some interesting forecasts to read!

[Soupspoon]

The British institution that is The Shipping Forecast is a precision thing, yet artful (it has inspired parody, poetry and music galore - go searching if you wish) in its 350/380 word constraints.

The standard glossary doesn't cover percentage chances, especially not as much as the comic does (accumulation of certainty not technically being part of it), but seemingly vague timescales and rates of change of pressures are covered more explicitly than the layperson might appreciate.

Meanwhile, the regular UK weather might start to officially use a wider language-base, to reflect local linguistic terms…

[Mikeski]

Up next, the conspiracy theorist, who will scale the forecast according to his car-washing, lawn-watering, and/or picnic-hosting plans.

[rattusprat]

My (anecdotal only) experience informs me that what is meant by 20% chance of rain depends on what part of the world you are in.

I used to live in a city that is very dry with very low annual rainfall. There my experience of "20% chance of rain on Wednesday" means "it's not going to rain, we're just saying this to keep the farmers interested and cover our but if you get a few spots". I would ballpark this actually translates to me experiencing at least one drop of rain in my specific location about 5% of the days where that was forecast.

I now live in a city with a much higher annual rainfall. Here my experience of "20% chance of rain on Wednesday" means "It will almost certainly rain (maybe 90% chance) but most likely will be a light shower lasting maybe an hour or less".

Again, this is only anecdotal evidence - I have never recorded actual data.

[rmsgrey]

For what it's worth, the National Weather Service does have a page on the meaning of "Probability of Precipitation": https://www.weather.gov/ffc/pop [...]

So, the answer to the questions in the first panel are all "maybe" - sometimes the underlying prediction is that a band of rain will pass through at some point during that 5 hour period, with dry weather before and after. Sometimes the underlying prediction is that it will be cloudy all afternoon, but any of those clouds could rain at any point.

Moving on to the second panel, scattered showers suggests the latter - that there's an approximately independent 20% chance for any given point and any given hour covered by the forecast - so with two reasonably separate locations there's a roughly 36% chance that at least one of them will see rain in any given hour, but only a 4% chance that both will.

The linguistic questions, I'm not qualified to address.

[da Doctah]

I live in Phoenix. "x% chance of rain" where x<90 means "at some point during the day it's going to cloud up alarmingly, but no actual rain will fall".

I used to live in Seattle. "x% chance of rain" where x>10 means "it's raining right now, and the individual drops are the size of canned hams".

[Soupspoon]

I remember, when very young, trying to work out whether walking or running in the rain would get me wetter. Walk and (barring wind) it only falls down onto my head but it does it a lot. Run, and I'm under the rain less, but I'm also running headlong into more raindrops.

I didn't realise it at the time, but I was trying to do applied maths beyond my years. Including my deduction(/induction?) tbat standing perfectly still would bring me an unlimited amount of rain while I did not progress towards shelter, and if I went as fast as the fastest thing ever (or more!) I'd instantaneously sweep out a body-profiled volume through all thr rain effectively just hanging there. If I remained upright. I think I concluded that leaning over to correcrly match my speed (or choosing a speed to correctly match my possible forward lean) based upon the local windspeed so that all rain would incident upon my head upon the same diagonal axis as I was off-vertical would be the most optimal rain-slicing method (and best yet if I wore a hat, a practice I have taken up again to weatherproof myself against all kinds of meteorological phenomena over the last decade or more, with a nice wide-brimmed hat with an optionally deployed chin-strap).

I was rather pleased with myself and working that out, in purely theoretical terms.


But in a very light rain (or, rather, a very sparse rain of raindrops large enough to obviously land, that kind of "end of a hot spell" form of precipitation), I start musing at the odds that I can walk through the budding shower for any given distance and just never quite be wear the raindrops land. The "Cohen the Barbarian" method of keeping dry. Imagine, through the air-column, a diagonally-upwards (and twisty-turny) instantaneous path of a human-sized cross-section that does not intersect with any drop of water, rooted at my current location, then as you wind time forward you move along the ground to match where the moisture-free path continuously falls to intersect it. Like a hyper-prescient Nicholas Cage.

[zjxs]

It's getting hot in here.

[Pfhorrest]

*takes off all his clothes*

[orthogon]

This probably explains some bizarre features of the BBC weather forecast. The summary for the day often seems completely at odds with the hour-by-hour forecast. The day summary can be "overcast" when every individual hour is "sunny". And the minimum temperature for the day is sometimes 2°C lower than any of the hourly values.

My mum has a theory that the BBC always forecasts worst weather than is actually predicted for bank holiday weekends, in order to "reduce traffic on the roads". (She's not convinced about the moon landings either).

I've wondered, too, about the dummy pronoun thing. Pronoun-dropping languages like Spanish can avoid the issue by not specifying what is raining at all. And yet Japanese, which could employ the same trick, goes with the more prosaically accurate "rain is falling". As a schoolchild I always wondered whether pleuvoir had conjugations for all the different persons. (Je pleux?) It's irregular, so it's not obvious that they would even exist. (Though perhaps you can say figurative things like "you're raining on my parade").

[YellowYeti]

I remember reading an article on just this in the mathematical gazette some years back:
http://www.people.hbs.edu/dbell/walk%20or%20run%20in%20the%20rain.pdf

TL;DR
" keep pace with the wind if it is from behind; otherwise, run for it."

Yay for maths!

I remember, when very young, trying to work out whether walking or running in the rain would get me wetter. Walk and (barring wind) it only falls down onto my head but it does it a lot. Run, and I'm under the rain less, but I'm also running headlong into more raindrops.

I didn't realise it at the time, but I was trying to do applied maths beyond my years. Including my deduction(/induction?) tbat standing perfectly still would bring me an unlimited amount of rain while I did not progress towards shelter, and if I went as fast as the fastest thing ever (or more!) I'd instantaneously sweep out a body-profiled volume through all thr rain effectively just hanging there. If I remained upright. I think I concluded that leaning over to correcrly match my speed (or choosing a speed to correctly match my possible forward lean) based upon the local windspeed so that all rain would incident upon my head upon the same diagonal axis as I was off-vertical would be the most optimal rain-slicing method (and best yet if I wore a hat, a practice I have taken up again to weatherproof myself against all kinds of meteorological phenomena over the last decade or more, with a nice wide-brimmed hat with an optionally deployed chin-strap).

I was rather pleased with myself and working that out, in purely theoretical terms.


But in a very light rain (or, rather, a very sparse rain of raindrops large enough to obviously land, that kind of "end of a hot spell" form of precipitation), I start musing at the odds that I can walk through the budding shower for any given distance and just never quite be wear the raindrops land. The "Cohen the Barbarian" method of keeping dry. Imagine, through the air-column, a diagonally-upwards (and twisty-turny) instantaneous path of a human-sized cross-section that does not intersect with any drop of water, rooted at my current location, then as you wind time forward you move along the ground to match where the moisture-free path continuously falls to intersect it. Like a hyper-prescient Nicholas Cage.

[Wee Red Bird]

It might also be a low probability of a heavy downpour but a high probability of light showers.
Throw in the likely hood of the heavy storm that may hit at 2pm is the same storm that will be at 3 and 4, but could be diverted away by a change in wind any any point before 2, then you'd need a 3d graph Time vs strength of rain vs likely hood. And you could augment with lines of probability, eg fanning out at 1:30 to either the light rain or the heavy rain.

Oh, snow and hailstones. Got to add in another axis for temperature.

[zanglebert]

Obligatory get out of me head...

That said:

Does anybody who works or worked in weather forecasting know how we're supposed to interpret those probabilities? 5 consecutive hours of "20% chance of rain" certainly isn't intended to say "certain rain in the afternoon". It could be the probability over the entire period then, but if that's the case, then what does it mean if there's a case like: "1pm to 3pm: 10%" and "3pm to 5pm: 20%"? Twice as much chance of rain in slot 2, compared to the average rain chance over the entire afternoon period?

Disclaimer: I'm neither a pure mathematicians nor a linguist, but I work in an applied math field. I can think of several ways to interpret these forecast probabilities, and know what they would mean, but I don't know how they're supposed to be understood. Maybe somebody can clarify what's the intended interpretation.

[khakipuce]

I've always assumed it's a forecast confidence measure. They run the forecast model with the range of possible inputs and if the various runs agree closely confidence and therefore the "% chance" is high, if they differ a lot then the chance is low. You can run the model for 1 hour ahead or 4 hours ahead get different confidence levels, presumably you can also run it for different sizes of land surface and get different confidence levels.

[Vroomfundel]

I remember reading an article on just this in the mathematical gazette some years back:
http://www.people.hbs.edu/dbell/walk%20or%20run%20in%20the%20rain.pdf

Poe's law FTW. I'm pretty sure these are genuine math problems but it sounds just like a joke about mathematicians out of touch with the real world.

To summarize, the quoted gazette presents the problem whether bicycle mudguards are necessary, solves it by means of radial forces required for dislodgement and concludes that mudguards are unnecessary as mud particles will inevitably fly off in rearward direction. Conspicuously absent is any discussion about the discrepancy between this result and the real world observation of mud particles ending up on one's back when riding an actual bicycle, leading a non-mathematician observer to conclude that the authors have never indulged in said endeavour.

[Soupspoon]

There's an additional reason to have mudguards (and full ones, not just those token 'back saver' ones) if you ride on roads in company with other cyclists, in that the water/road-mud that does handily fling itself generally rearwards, rather than hang on for just long enough to lose its increasingly tenuous adhesion as the radius starts moving in a direction back over the axle, will splatter the face of the rider currently directly behind you.

It may be an incentive for some people to continuously take the front of a chain of cyclists, taking the brunt of the force of the wind or even ground-stationary air just to not get a face full of tyre-spatter, but it's antisocial and relegating the two mudguardless individuals¹ perpetually to the back of the gang and never doing any of the hard work earns them few friends.

Mud guards and mud-flaps have only been removed from my various bikes when racing (which I was never that good at, but I certainly can't blame the additonal wind resistence from mudguards for that!) in the solitary way of my sport. Biking over peat bogs and muddy tracks did often necessitate the aquisition of s scraping stick to degunk the tyre-to-guard space of my general purpose road bike used on roads before and after the route over the barely used mountain passes concerned, but that's an age-old activity conducted for many decades before the newly-adventurous MAMIL drove up to a visitor's car-park and de-racked his dedicated mountain bike for a racing circuit round a specially constructed 'racing circuit', near enough (or a downhill route parallel to the ski-lift, operated outside of snow season as a year-spanning revenue stream).

Aye, this is the world where "dual carriageways" is a farm track with two wheel-rutts through bogland, infilled by stones in the worst places if you're lucky, giving you a choice of which side of the grass meridian you can ride. And still with full mudgaurds, because of the person riding along behind you.


¹ Yes, two. As recommended except where particular road conditions dictate that it's safer to be in a twice as long single-file.

[rmsgrey]

I remember reading an article on just this in the mathematical gazette some years back:
http://www.people.hbs.edu/dbell/walk%20or%20run%20in%20the%20rain.pdf

TL;DR
" keep pace with the wind if it is from behind; otherwise, run for it."

Yay for maths!

It's worth noting the assumptions that limit the applicability of the result: particularly that the distance to travel is short - so you can assume that the rain will be constant for the entire time you're exposed, and that you won't reach an equilibrium state.

As in the article, there are two sources of water to consider - rain landing on you as a result of not being in shelter, and rain you run into as a result of moving relative to the wind.

For speeds where it will stop raining before you reach shelter, the amount of rain landing on you only depends on the duration of the rain, while the amount of rain you run into increases the faster you're moving relative to the wind, so it's best to move with the wind (in the absence of other constraints). If you can reach shelter before the rain ends, then it's a trade-off between the amount of rain you run into covering the distance (relative to the wind, so the shelter is effectively moving) to that shelter and the amount of rain that would have landed on you during the time after you reach shelter.

In circumstances where you will reach an equilibrium state, things get more complicated - depending on the rate of rain, and your outfit, the rain landing on you may represent a significantly different equilibrium to that of the rain you run into - and you may prefer one to the other. An umbrella or a suitable hat can reach equilibrium rapidly.

The paper also assumes that people are adequately modeled by a right parallelepiped - assuming that your horizontal cross section doesn't change significantly with speed (reasonable for a range of speeds, but breaks down for stationary people - as people walk/run, their legs extend beyond their upper body's rain-shadow - and may also fail when comparing different gaits) and that you're always vertical.

If people lean over as they walk/run, then modeling them as an angled parallelepiped is more appropriate. Relative to the rain, both a person's intrinsic lean, and their relative horizontal motion manifest as an angle (rain you run into is then rain striking the sloping front/back surface; rain that lands on you is rain striking the horizontal top surface) and you can adjust them to cancel each other out within limits - either remaining stationary and leaning at the same angle the rain is falling, or moving, leaning at the angle the rain is falling relative to yourself.

[Rombobjörn]

Calculating the total amount of rain that will hit you is insufficient. You also need to consider how it will be distributed. Do you prefer to have your hair and shoulders soaked and the rest of your body mostly dry, or to have the same amount of water spread out over your entire front side so that it only makes your clothes slightly damp? Note that water evaporates faster from a larger surface.

Consider also that running fast for more than a short time is likely to make your clothes wet from the inside.

[pogrmman]

I've wondered, too, about the dummy pronoun thing. Pronoun-dropping languages like Spanish can avoid the issue by not specifying what is raining at all. And yet Japanese, which could employ the same trick, goes with the more prosaically accurate "rain is falling". As a schoolchild I always wondered whether pleuvoir had conjugations for all the different persons. (Je pleux?) It's irregular, so it's not obvious that they would even exist. (Though perhaps you can say figurative things like "you're raining on my parade").

Or you can have a language like Russian that sidesteps the issue entirely by making rain (or other percipitation) the subject — “идёт дождь” (it’s raining) literally means “the rain goes”

[thunk]

I think I did something sort of similar on my last mesoscale exam.
When asked to use an equation to identify where points of shearing instability might be in a squall line profile, I put in a throwaway line about assuming that the field I was working in was R^3.

Guess I'll leave broadcast to my friends.

[somitomi]

I've wondered, too, about the dummy pronoun thing. Pronoun-dropping languages like Spanish can avoid the issue by not specifying what is raining at all. And yet Japanese, which could employ the same trick, goes with the more prosaically accurate "rain is falling". As a schoolchild I always wondered whether pleuvoir had conjugations for all the different persons. (Je pleux?) It's irregular, so it's not obvious that they would even exist. (Though perhaps you can say figurative things like "you're raining on my parade").

Or you can have a language like Russian that sidesteps the issue entirely by making rain (or other percipitation) the subject — “идёт дождь” (it’s raining) literally means “the rain goes”

Hungarian does that too, saying "esik az eső" (the rain falls) except that in everyday speech it's usually shortened to "esik", because Hungarians can have sentences with no apparent subject

[rmsgrey]

Calculating the total amount of rain that will hit you is insufficient. You also need to consider how it will be distributed. Do you prefer to have your hair and shoulders soaked and the rest of your body mostly dry, or to have the same amount of water spread out over your entire front side so that it only makes your clothes slightly damp? Note that water evaporates faster from a larger surface.

Consider also that running fast for more than a short time is likely to make your clothes wet from the inside.

Most of the time, if you're out in the rain, you're not getting any significant evaporation - the air's already saturated with water vapour.

It's also worth bearing in mind that if your hair and shoulders are soaked, that water is then going to flow down - water tends to spread and/or flow anyway, so the difference between some amount of water landing on you from above, and running into the same amount of water from the side is going to be reduced.

[Reka]

Hungarian does that too, saying "esik az eső" (the rain falls) except that in everyday speech it's usually shortened to "esik", because Hungarians can have sentences with no apparent subject

The thing I can do in Hungarian but not really in English is indirectly complain about the weather by saying "világít a nap". Literally, it means "the sun is giving off light". The reason it works is that the usual way to say that the sun is shining is "süt a nap", where "süt" means both "shine" and "bake", i.e. it implies both heat and light. Specifying one implies a certain, um, dearth of the other.

(What can I say, I grew up in Southern California.)

[YTPrenewed]

"Probability of precip" is based on what's known as ensemble forecasts. Basically, you use a computer model to simulate the physics of weather, and you experiment with the effects of various likely inputs for weather variable (temperature, wind speed, etc...) to reflect the uncertainty in what those variables are between weather stations, and how they translate to uncertainty in predicted results.

If 50% of those inputs result in precip, for instance, they base the 50-50 chance on that. Hence why there's not as much difference between probability per hour and probability for the whole afternoon as you'd expect from your usual introductory stats course.

See, this is why weather broadcasting should be required to hire people who've done actual courses in meteorology, instead of journalism majors pretending to be meteorologists.

[Soupspoon]

Specifying one implies a certain, um, dearth of the other.

Without quite the same sparseness of words, there are ways round here of indicating that it's a bright winter morning, i.e. the temperature is going to need gloves, your ears will feel nipped and the vapour of your breath will be visible. The brighter the morning, the more cloudless the night and the more hoarfrost on the meadow grass. (Or the crisper the snow, if there's some left over, but cold and bright tends to be both colder and brighter (skywards, anyway) without having had actual snow.)

See, this is why weather broadcasting should be required to hire people who've done actual courses in meteorology, instead of journalism majors pretending to be meteorologists.

UK weather presenters have tended to be dedicated Met Office employees, at least historically upon the BBC (TV and radio set pieces). I don't know how this has changed with the contract changing stuff, I've heard that some of the Big Name weather-people may have made themselves self-employed contractors so that they can continue to work as presenters and as forecasters exactly as much as everyone can agree, without being directly affected by the B2B contract changes.

The contract loss might have made this training course more necessary, or it might have already been in existence for commerical broadcasters who didn't always have such a direct-from-expert feed.

(We Brits tend to like the people who tell us the weather about as much as we like to talk about the weather itself. And if we can combine the two and use a clip of Michael Fish at the start of the 2012 Olympics - possibly to hedge the bets against them being rained on, which turned out not to happen, surprisingly - it makes us feel all happy inside, however wet outside.)

[qvxb]

Channel 4 just hired Stormy Daniels to be their new weatherperson. She edged out Jenny McCarthy for the position.

Be sure to check out the new "Big Blow" hurricane coverage.

[Rombobjörn]

Most of the time, if you're out in the rain, you're not getting any significant evaporation - the air's already saturated with water vapour.

If you're trying to decide whether you should run or walk through the rain, then you're probably aiming to get indoors, where the air is probably not saturated.

[rmsgrey]

Most of the time, if you're out in the rain, you're not getting any significant evaporation - the air's already saturated with water vapour.

If you're trying to decide whether you should run or walk through the rain, then you're probably aiming to get indoors, where the air is probably not saturated.

And most places which are indoors, you can take off (layers of) damp clothing, and use other means of getting dry than waiting for evaporation.

[Mikeski]

This probably explains some bizarre features of the BBC weather forecast. [...] And the minimum temperature for the day is sometimes 2°C lower than any of the hourly values.

That's similar to one of my favorites. When the weather is changing rapidly, you'll see Monday's forecast high temp at 20°F, and Tuesday's forecast low temp at 27°F. Apparently the models say it instantly jumps 7 degrees at midnight?

[MrT2]

This probably explains some bizarre features of the BBC weather forecast.

The other explanation is that it's been outsourced about 6-8 weeks ago to another company, no longer the Met Office, and they don't appear to have any appreciation for the multitude of microclimates the UK has (there are days when a thousand square miles gets the same weather, or days when you can take any line 100 miles long in the country, and find 20+ different weather conditions; you can also get rain, hail, snow/sleet and a suntan on the same day while staying in one location).