Why is Boston weather prediction so poor
May 27, 2022 7:02 AM Subscribe
I moved to Boston 8 months ago and all weather reporting seems absolutely pointless. In Chicago, Minneapolis, Detroit, etc, they could regularly make accurate predictions about weather. In Boston, it's always way, way off.
I moved here 8 months ago and so far there have been 7 "severe weather" alerts. Snow, ice, hail, etc.
Exactly 0 of them occured how even close to how they were predicted.
Even in the day to day, weather.com and darksky both said it would be a high of 76 today and it's currently 80 at 7am - we'll break 85 for sure.
Tomorrow they call for thunderstorms. Seems unlikely.
I've basically taken to standing outside and feeling how warm it is because all of technology seems to have failed.
So... Weather.com and darksky are just big algorithms, right? Do the algorithms fit Boston much worse than they do in Chicago, Detroit, Minneapolis, etc? Is that because of the distance to the coast? (People inland Massachusetts say they have equally bad reporting) or is there a human component unique to Massachusetts?
Or... Has the entire nation gotten worse at weather reporting in the last year? Did big government weather get a funding cut? Is global warming throwing everything out of wack? What's going on?!
I moved here 8 months ago and so far there have been 7 "severe weather" alerts. Snow, ice, hail, etc.
Exactly 0 of them occured how even close to how they were predicted.
Even in the day to day, weather.com and darksky both said it would be a high of 76 today and it's currently 80 at 7am - we'll break 85 for sure.
Tomorrow they call for thunderstorms. Seems unlikely.
I've basically taken to standing outside and feeling how warm it is because all of technology seems to have failed.
So... Weather.com and darksky are just big algorithms, right? Do the algorithms fit Boston much worse than they do in Chicago, Detroit, Minneapolis, etc? Is that because of the distance to the coast? (People inland Massachusetts say they have equally bad reporting) or is there a human component unique to Massachusetts?
Or... Has the entire nation gotten worse at weather reporting in the last year? Did big government weather get a funding cut? Is global warming throwing everything out of wack? What's going on?!
I don't know my way around Boston in the least, but here in Seattle the ocean and topography play havoc with weather forecasting, especially the sorts of naïve models like Dark Sky uses. It's complicated. I've seen it bucket down rain at home while a couple miles away the region's official weather station doesn't get a trace of rain.
Meanwhile in the Midwest the topography is vastly simpler and the region is blanketed with high quality weather radar and stations due to the risk of incredibly severe weather.
posted by wotsac at 7:13 AM on May 27, 2022 [2 favorites]
Meanwhile in the Midwest the topography is vastly simpler and the region is blanketed with high quality weather radar and stations due to the risk of incredibly severe weather.
posted by wotsac at 7:13 AM on May 27, 2022 [2 favorites]
Hmm, some of this might be Weather.com and DarkSky being specifically wrong - for instance I was expecting highs in the low/mid 80s in suburban Boston today based on Wunderground (my preferred weather app) and whatever Alexa uses.
I do also think, though, as a longtime coastal MA resident, that this past year's forecasts have been worse than usual all over New England.
posted by mskyle at 7:20 AM on May 27, 2022 [1 favorite]
I do also think, though, as a longtime coastal MA resident, that this past year's forecasts have been worse than usual all over New England.
posted by mskyle at 7:20 AM on May 27, 2022 [1 favorite]
In Chicago, the weather is vastly different near the lake versus away from the lake. In the winter I assumed it was going to be friggin cold and in the summer it would be hot and humid.
As far as Boston, it is my experience, and it is a short experience, that the best way to predict the weather is to see what is happening in NYC and on the eastern end of LI and then add say 12-18 hours and there you have it.
I use the Google weather app. Not sure who their backend is, but it seems to be reasonably accurate for a day or two out. Not much more than that.
posted by JohnnyGunn at 7:23 AM on May 27, 2022
As far as Boston, it is my experience, and it is a short experience, that the best way to predict the weather is to see what is happening in NYC and on the eastern end of LI and then add say 12-18 hours and there you have it.
I use the Google weather app. Not sure who their backend is, but it seems to be reasonably accurate for a day or two out. Not much more than that.
posted by JohnnyGunn at 7:23 AM on May 27, 2022
Best answer: The weather for Boston from many services is usually the prediction for Logan Airport, which sticks out into the ocean from East Boston. I live on the other side of the city, about 8 miles inland and there can be drastically different temperatures.
To whit: Mr. chiefthe was in the North End last weekend where it was 90+ in our neighborhood and at our friends' place in Somerville. It was a breezy 75 when he picked cookies up at Mike's.
Like mskyle, I use Wunderground and find it to not be too far off, if I use my specific zip code for my neighborhood. And I find the general long-term weather to be pretty accurate -- like it will rain on these couple days, and the general temp range (within 10 degrees) will be x-y.
posted by chiefthe at 7:26 AM on May 27, 2022 [10 favorites]
To whit: Mr. chiefthe was in the North End last weekend where it was 90+ in our neighborhood and at our friends' place in Somerville. It was a breezy 75 when he picked cookies up at Mike's.
Like mskyle, I use Wunderground and find it to not be too far off, if I use my specific zip code for my neighborhood. And I find the general long-term weather to be pretty accurate -- like it will rain on these couple days, and the general temp range (within 10 degrees) will be x-y.
posted by chiefthe at 7:26 AM on May 27, 2022 [10 favorites]
I think it is the effect of the ocean. There can be a twelve degree difference between the fog belt and a mile inland, and of course Boston is big enough that if you report the weather as it is just on the shoreline you'll be wrong for most of the greater Boston area, and if you report the weather as it is somewhat inland you'll be wrong for a huge chunk of the area near the water. If you average it you get to be wrong everywhere.
Where I live in an ocean side city they often predict summer days where the temperature will go up to eighty or so by mid afternoon. But it the temperature climbs enough the hot air inland creates an updraft which hauls the cool and moist air in the bay inland. This means that a sunny day with a prediction of 84 degrees is likely to transition to an overcast day at 63 degrees if the temperature rises fast enough. The hotter it gets inland, the cooler it seems to get on the very coast.
Very often the weather they predict is the weather at the airport - and that may be quite accurate. You are glaring at the drizzle, while the airport, built in the location most likely to provide good visibility year round, is enjoying intermittent mild sun and cloud.
And finally as governments work to be more efficient and save expenses, they may have laid off a lot of meteorologists. Instead of having someone actually paying attention to adjusting the forecast they have someone who is using historic trends to calculate it more quickly. But the historic trends are off as the extremes are becoming a little more extreme, so we are increasingly getting weather reports that greet us on waking up with the information that the high today will be 86* and the current temperature is 90*.
posted by Jane the Brown at 7:38 AM on May 27, 2022
Where I live in an ocean side city they often predict summer days where the temperature will go up to eighty or so by mid afternoon. But it the temperature climbs enough the hot air inland creates an updraft which hauls the cool and moist air in the bay inland. This means that a sunny day with a prediction of 84 degrees is likely to transition to an overcast day at 63 degrees if the temperature rises fast enough. The hotter it gets inland, the cooler it seems to get on the very coast.
Very often the weather they predict is the weather at the airport - and that may be quite accurate. You are glaring at the drizzle, while the airport, built in the location most likely to provide good visibility year round, is enjoying intermittent mild sun and cloud.
And finally as governments work to be more efficient and save expenses, they may have laid off a lot of meteorologists. Instead of having someone actually paying attention to adjusting the forecast they have someone who is using historic trends to calculate it more quickly. But the historic trends are off as the extremes are becoming a little more extreme, so we are increasingly getting weather reports that greet us on waking up with the information that the high today will be 86* and the current temperature is 90*.
posted by Jane the Brown at 7:38 AM on May 27, 2022
For thunderstorm warnings I use Lightning Maps - a community project which detects and maps lightning strikes in real time. There's an app.
posted by hat_eater at 7:43 AM on May 27, 2022
posted by hat_eater at 7:43 AM on May 27, 2022
Haha, yeah, that's Boston. I'm gonna disagree about the locality. I live in southern New Hampshire, and the forecasts are similarly off here, sometimes by 10 degrees. I assume it's a combination of the ocean, the Appalachians, and Canada. Not in like a South Park sense. Just like, a different part of Canada than affects the Midwest. Mount Washington, in northern NH, is famous for its turbulent weather, and I think a lot of the factors affecting it are the same further south.
As a native Midwesterner, I could spin it the other way: that Boston's mediocrity is normal, and everyone in the Midwest is above average. (Is Garrison Keillor cancelled? I can still make that joke, though, right?)
posted by kevinbelt at 7:47 AM on May 27, 2022 [2 favorites]
As a native Midwesterner, I could spin it the other way: that Boston's mediocrity is normal, and everyone in the Midwest is above average. (Is Garrison Keillor cancelled? I can still make that joke, though, right?)
posted by kevinbelt at 7:47 AM on May 27, 2022 [2 favorites]
I live in Providence and don't necessarily find weather predictions to be out of whack compared with the Midwest except for hurricanes and nor'easters which seem to be inherently unpredictable. That ocean element.
I do feel like there were more- than- usual big storms predicted that turned out to be nothing this fall/ winter.
posted by geegollygosh at 7:47 AM on May 27, 2022
I do feel like there were more- than- usual big storms predicted that turned out to be nothing this fall/ winter.
posted by geegollygosh at 7:47 AM on May 27, 2022
New England has two main sources of weather: fronts coming from the west (mid-west Great Lakes region, typical in heavy-snowing winter storms, or the north-west Arctic, for long cold snaps, getting more common thanks to climate change) and fronts coming from the south and east (hurricanes, nor-easters). Anything coming up the coast from the Caribbean or from the Atlantic Ocean more generally can be hugely unpredictable. It's fairly common for a storm to, for example, cruise up the coast at a steady rate, really predictably, then hit the CN/RI/Cape Cod "shelf" and suddenly stall, or bump out to visit Nova Scotia, or drag inland and hit a eastern-moving front and then who knows what. Or it could not do any of that! Very manic pixie dream weather.
You also need to think about incentives: NOAA's satellites have a ton of (publicly available, taxpayer funded!) information which is used by NOAA itself primarily for public safety measures (road conditions, potential flooding, etc.). Local forecasters and commercial weather apps/stations want you to keep tuning in/logging on--so it's in their best interest to be a little conservative with their predictions. If they predict showers and it rains in Cambridge but not the Back Bay, everyone's fine. If they predict sun, otoh, all those damp people in Cambridge are now pretty cranky, even though on balance the "success" rate is the same.
posted by radiogreentea at 8:35 AM on May 27, 2022
You also need to think about incentives: NOAA's satellites have a ton of (publicly available, taxpayer funded!) information which is used by NOAA itself primarily for public safety measures (road conditions, potential flooding, etc.). Local forecasters and commercial weather apps/stations want you to keep tuning in/logging on--so it's in their best interest to be a little conservative with their predictions. If they predict showers and it rains in Cambridge but not the Back Bay, everyone's fine. If they predict sun, otoh, all those damp people in Cambridge are now pretty cranky, even though on balance the "success" rate is the same.
posted by radiogreentea at 8:35 AM on May 27, 2022
Somewhere like Chicago has relatively predictable weather, lake effects notwithstanding, because it’s in the middle of a continent — the standard prevailing westerlies blowing the normal weather a thousand miles from the Rockies, and any anomalies, such as warm, moist air from the Gulf of Mexico or cold air from Canada are more straightforward to see coming. I think that the predictability of mid-continental weather is the more unusual case.
Boston has the standard westerlies from the interior, but is also next to the cool Gulf of Maine (Noreasters) and in the potential path of warmer, moister air from the Caribbean. Hurricanes are not common in Massachusetts, but they do happen and we know how hard their track can be to predict, even if we don’t take a sharpie to the map. Because of the specific geography Boston seems to sit near the intersection of various patterns and is very sensitive to the exact boundries of the various effects; a few miles can make a huge difference between which sysytem dominates local conditions.
[On preview, what radiogreentea says.]
posted by Quinbus Flestrin at 8:38 AM on May 27, 2022 [1 favorite]
Boston has the standard westerlies from the interior, but is also next to the cool Gulf of Maine (Noreasters) and in the potential path of warmer, moister air from the Caribbean. Hurricanes are not common in Massachusetts, but they do happen and we know how hard their track can be to predict, even if we don’t take a sharpie to the map. Because of the specific geography Boston seems to sit near the intersection of various patterns and is very sensitive to the exact boundries of the various effects; a few miles can make a huge difference between which sysytem dominates local conditions.
[On preview, what radiogreentea says.]
posted by Quinbus Flestrin at 8:38 AM on May 27, 2022 [1 favorite]
If they predict showers and it rains in Cambridge but not the Back Bay, everyone's fine.
IMO, most weather predictions are getting worse because they cover large areas, which can be one hundred miles or more. So a 20% prediction of rain over 500 or 1000 sq miles isn't particularly all that useful individually.
California actually has coastal, inland, and mountain forecasts, so separation of those climactic areas is totally possible if the temperature differences are significant.
posted by The_Vegetables at 8:48 AM on May 27, 2022
IMO, most weather predictions are getting worse because they cover large areas, which can be one hundred miles or more. So a 20% prediction of rain over 500 or 1000 sq miles isn't particularly all that useful individually.
California actually has coastal, inland, and mountain forecasts, so separation of those climactic areas is totally possible if the temperature differences are significant.
posted by The_Vegetables at 8:48 AM on May 27, 2022
Weather moves west -> east, generally, and East Coast weather is also affected by ocean temps and winds, and it's just a lot more volatile. I've lived in San Francisco and Colorado where the weather is pretty predictable, Ohio where it's more volatile, and now Maine. Compared to 30 years ago, weather predictions are very accurate, thanks, NOAA. But variable by proximity to the coast, microclimates, and capricious weather god/esses. I often look at the animated weather maps; it's easier to see where currents are heading. and I check the weather by zip code, not town. I've been in a couple massive storms years ago when predictions were based on what was happening 100 miles west, and I am really thankful that we get storm warnings in a useful time frame.
posted by theora55 at 9:32 AM on May 27, 2022
posted by theora55 at 9:32 AM on May 27, 2022
In my experience, the commercial weather sites tend to be a little more aggressive and the official NWS forecast is a bit more conservative. That won't affect severe weather watches/warnings since those are all issued by the feds, but for something like snow total predictions it seems like NWS will consistently predict less snow than the local news stations.
I would recommend checking out the NWS forecast discussion. It's a little technical, but there's a built in glossary for any of the weather jargon, and the forecasters do a really good job explaining what they're confident on and what's got a lot of uncertainty around it.
Anecdotally, I have found that the US weather global forecasting model (GFS) tends to be less accurate around here for surface winds and winds aloft than the European model (ECMWF). The NWS forecasters do use both models (and a limited North American model - NAM) and will frequently blend solutions from them, but it's possible that when they lean towards the GFS solution it may be less accurate in this area.
posted by backseatpilot at 9:53 AM on May 27, 2022
I would recommend checking out the NWS forecast discussion. It's a little technical, but there's a built in glossary for any of the weather jargon, and the forecasters do a really good job explaining what they're confident on and what's got a lot of uncertainty around it.
Anecdotally, I have found that the US weather global forecasting model (GFS) tends to be less accurate around here for surface winds and winds aloft than the European model (ECMWF). The NWS forecasters do use both models (and a limited North American model - NAM) and will frequently blend solutions from them, but it's possible that when they lean towards the GFS solution it may be less accurate in this area.
posted by backseatpilot at 9:53 AM on May 27, 2022
This might be a junk response, but I’m in Austin, where fall/winter/spring weather can be variable. (summer’s easy: dry and hotter than the surface of the sun, all day, every day.)
I wanted to say that here, Dark Sky is usually off by a lot. Accuweather is generally not bad. It’s a shame it’s not the other way around. Dark Sky has a way better UI and Accuweather is morally questionable.
But my top weather forecasts come from meteorologists on Facebook. (sorry to admit getting any kind of news on Facebook, but here we are.) My favorite is a TV weatherman posting his own stuff on the side — not reposted TV clips — but there are non-professionals worth following too. They provide timely, nuanced, relevant forecasts with interesting discussions, and they’re more accurate than the apps. So I wonder if there are similar people in your area.
posted by liet at 10:48 AM on May 27, 2022
I wanted to say that here, Dark Sky is usually off by a lot. Accuweather is generally not bad. It’s a shame it’s not the other way around. Dark Sky has a way better UI and Accuweather is morally questionable.
But my top weather forecasts come from meteorologists on Facebook. (sorry to admit getting any kind of news on Facebook, but here we are.) My favorite is a TV weatherman posting his own stuff on the side — not reposted TV clips — but there are non-professionals worth following too. They provide timely, nuanced, relevant forecasts with interesting discussions, and they’re more accurate than the apps. So I wonder if there are similar people in your area.
posted by liet at 10:48 AM on May 27, 2022
I moved back here after living in LA for years, where the weather service would predict rain to within about five minutes. It has been a shock to deal with the unpredictability here.
posted by rednikki at 1:58 PM on May 27, 2022 [1 favorite]
posted by rednikki at 1:58 PM on May 27, 2022 [1 favorite]
The weather predictions have been... erratic for us here just north of Chicago these past few months. And it's not the usual warm/cold front taking an extra day or two to show up. I mean day of dramatically off target: Rain not showing up, or coming in our of no where. A couple weeks ago my husband checked the day's forecast, said sunny and all clear so he took the covers off the patio furniture. Within 10 minutes we had a downpour.
I don't doubt that prediction is worse in Boston for all the reasons mentioned, but maybe it's an unusually bad year all around?
posted by ghost phoneme at 3:51 PM on May 27, 2022 [1 favorite]
I don't doubt that prediction is worse in Boston for all the reasons mentioned, but maybe it's an unusually bad year all around?
posted by ghost phoneme at 3:51 PM on May 27, 2022 [1 favorite]
I live north of Boston, and for what it's worth, the only weather forecasting I ever trust is Dave Epstein. I think he's on Boston.com, but I just follow him on Twitter at @growingwisdom. He updates very frequently, especially when something big is on the horizon, and shows a lot of different maps and potential paths of incoming weather. Also basically talks Boston weather all the time, so you get a historical perspective as well if you're into that. Also lots of pics of his garden, and some gardening tips for the area.
For just day to day "will it rain" and "what's the temp" and "what's the humidity", I've found Alexa seems to be accurate enough to keep me happy. I'm not sure what service she uses though.
posted by invincible summer at 4:44 PM on May 27, 2022 [2 favorites]
For just day to day "will it rain" and "what's the temp" and "what's the humidity", I've found Alexa seems to be accurate enough to keep me happy. I'm not sure what service she uses though.
posted by invincible summer at 4:44 PM on May 27, 2022 [2 favorites]
I don't have data to support this, but my pet hypothesis is that this phenomenon is happening everywhere. Weather reports are an algorithm, yes, but for the most part that algorithm is literally just "find the last 10/20/100 times in recorded history that this set of circumstances occurred, and report on what happened immediately after that." For places like Boston, we have remarkably accurate weather data going back a long ways, so with the exception of some severe outliers (hello Blizzard of '78) there's usually SOMETHING in the archives that lets the apps make their predictions.
Enter climate change. As anthropogenic carbon continues to pile up in the atmosphere, so too do the incidents of wildly unpredictable weather. CLIMATE can still be predicted with decent accuracy, but looking at 100 years of weather patterns in service of saying "based on today's conditions and the shape of the incoming front, the high tomorrow will be 83" is increasingly a fool's errand. Last weekend's freak temperature spike was mostly predicted 3 or 4 days out, but there's nothing in the archives that tell us much about 90+ degree days in mid-May, so the accuracy of the predictions is necessarily limited.
As others upthread have pointed out, you'll have better luck with local humans who are able to extrapolate from past data + known patterns + a healthy dose of realism, but looking for accuracy within a few degrees is going to become harder and harder to do as outlying data points become the norm.
posted by Mayor West at 5:14 PM on May 27, 2022 [1 favorite]
Enter climate change. As anthropogenic carbon continues to pile up in the atmosphere, so too do the incidents of wildly unpredictable weather. CLIMATE can still be predicted with decent accuracy, but looking at 100 years of weather patterns in service of saying "based on today's conditions and the shape of the incoming front, the high tomorrow will be 83" is increasingly a fool's errand. Last weekend's freak temperature spike was mostly predicted 3 or 4 days out, but there's nothing in the archives that tell us much about 90+ degree days in mid-May, so the accuracy of the predictions is necessarily limited.
As others upthread have pointed out, you'll have better luck with local humans who are able to extrapolate from past data + known patterns + a healthy dose of realism, but looking for accuracy within a few degrees is going to become harder and harder to do as outlying data points become the norm.
posted by Mayor West at 5:14 PM on May 27, 2022 [1 favorite]
I want to point out a couple of things.
1. Weather.com and Wunderground are both now owned by IBM. Although it does _not_ own the television Weather Channel, it does have a license to use the branding.
2. That doesn't necessarily mean they use the same data to make their predictions. Wunderground assimilates data from personal weather stations. Anyone can buy one, set it up, and contribute their data. Assuming their algorithms are similar, I'd think it would give the edge to Wunderground because their input data is more granular. Weather prediction models are complex, but I believe the more information they have about the situation at the beginning of computation, the better their predictions will be.
posted by TimHare at 6:11 PM on May 27, 2022
1. Weather.com and Wunderground are both now owned by IBM. Although it does _not_ own the television Weather Channel, it does have a license to use the branding.
2. That doesn't necessarily mean they use the same data to make their predictions. Wunderground assimilates data from personal weather stations. Anyone can buy one, set it up, and contribute their data. Assuming their algorithms are similar, I'd think it would give the edge to Wunderground because their input data is more granular. Weather prediction models are complex, but I believe the more information they have about the situation at the beginning of computation, the better their predictions will be.
posted by TimHare at 6:11 PM on May 27, 2022
I do agree that default use of Logan as a proxy for Boston makes a difference, as does distance from the ocean. I saw a similar effect when I lived a block away from the lake in Chicago: we were often several degrees warmer (winter) or more humid but cooler (summer) than O'Hare, which was the local default weather location, and snowfall tended to vary as well. Having lived in both cities as a pedestrian and bike commuter, I wouldn't say I've found a noticeable difference in forecast quality?
I also agree that not all weather services are equally good. Wunderground - by ZIP code, not city name - said it was at ~72 °F (not 80 °F, which it really wasn't) in my corner of Cambridge when I saw this in the morning, and it was predicted to hit somewhere a little over 80 °F, and both of those seemed to track pretty well with the actual weather.
posted by ASF Tod und Schwerkraft at 6:22 PM on May 27, 2022
I also agree that not all weather services are equally good. Wunderground - by ZIP code, not city name - said it was at ~72 °F (not 80 °F, which it really wasn't) in my corner of Cambridge when I saw this in the morning, and it was predicted to hit somewhere a little over 80 °F, and both of those seemed to track pretty well with the actual weather.
posted by ASF Tod und Schwerkraft at 6:22 PM on May 27, 2022
I use the Accuweather phone app in Cambridge and I find it remarkably accurate. Even when it says "Rain starting in 17 minutes" it's typically pretty close.
posted by Jasper Fnorde at 6:08 AM on May 28, 2022
posted by Jasper Fnorde at 6:08 AM on May 28, 2022
This does not answer your question, but...I believe you will have thunderstorms, because it's been storming here (western CT) for the past 2 hours. I just spoke to my son who is in Hartford, and he reports that the storms started there about an hour ago. They seem to be moving in a northeasterly direction, so I think you will start to see activity soon!
I love the Lightningmaps website!
posted by sundrop at 12:03 PM on May 28, 2022
I love the Lightningmaps website!
posted by sundrop at 12:03 PM on May 28, 2022
Follow Dave Epstein on Twitter @growingwisdom and use Accuweather. Also zoom into those maps of the Boston area because sometimes it can be forecast to rain in Cambridge and not in JP.
posted by jdl at 1:26 AM on May 29, 2022
posted by jdl at 1:26 AM on May 29, 2022
This thread is closed to new comments.
posted by Melismata at 7:08 AM on May 27, 2022 [7 favorites]