Smartphones, tablets help UW researchers improve storm forecasts
The next advance in weather forecasting may not come from a new satellite or supercomputer, but from a device in your pocket. University of Washington atmospheric scientists are using pressure sensors included in the newest smartphones to develop better weather forecasting techniques.
“With this approach we could potentially have tens or hundreds of thousands of additional surface pressure observations, which could significantly improve short-term weather forecasts,” said Cliff Mass, a UW professor of atmospheric sciences.
PressureNet is a free app for Android devices that contain pressure sensors.
Owners of certain new Android smartphones and tablet computers can now download the PressureNet app, which measures atmospheric pressure and provides the data to UW researchers.
When some smartphone manufacturers recently added pressure sensors, to estimate the phone’s elevation and help pinpoint its location, Mass saw an opportunity to enhance weather prediction. In the autumn he approached Cumulonimbus, a Canadian app company that developed a barometer application for smartphones that collects all the data and shares it back with users.
The PressureNet app this week collected about 4,000 observations per hour, with users clustered in the northeastern United States and around some major cities.
“We need more density,” Mass said. “Right now it’s a matter of getting more people to contribute.”
Android devices equipped with pressure sensors include Samsung’s Galaxy S3, Galaxy Nexus, Galaxy Note and Nexus 4 smartphones, and the Nexus 10 and Motorola Xoom tablet computers.
Atmospheric pressure is the weight of the air above, and includes information about what is happening as air masses collide. Precise tracking of pressure readings and pressure changes could help weather forecasters to pinpoint exactly where and when a major storm will strike.
Mass is particularly interested in the center of the country, which is prone to severe storms but includes fewer weather observation stations.
“Thunderstorms are one of the areas of weakest skill for forecasting,” Mass said. “I think thunderstorms in the middle part of the country could potentially be the biggest positive for this approach. They are relatively small-scale, they develop over a few hours, they can be severe and can affect people significantly.”
Tracking storms a few hours out could help people better protect themselves and their property. In the Seattle area, the tool could improve short-term forecasts for wind and rain.
“I think this could be one of the next major revolutions in weather forecasting, really enhancing our ability to forecast at zero to four hours,” Mass said.
UW researchers are the first scientists to have access to the smartphone pressure data. They are plotting the observations and preparing them for use in weather-prediction models.
Cumulonimbus updated the app’s privacy settings last week so users could allow access to the data by scientific researchers. Since then, the UW group has been uploading the pressure data each hour and preparing it for use in weather forecasting models. The data will soon be available to all researchers who want to incorporate it in weather-prediction tools.
A project begun in 2010 by Mass and Gregory Hakim, a UW professor of atmospheric sciences, has explored ways to improve weather forecasts by taking advantage of surface pressure measurements. The current network of U.S. weather stations offers about one thousand air-pressure readings. Adding observations collected by small-scale weather networks and hobbyists, the UW team found, improves the forecasts. A weather station in every pocket would offer an unprecedented wealth of data.
A recent blog post by Mass explains more about the UW group’s approach. Luke Madaus, a UW graduate student in atmospheric sciences, will load the smartphone data into a weather-forecasting system. At first the tool will use only stationary data points, but eventually it may include data from devices in motion.
Building the system will take a few months, Mass said. By this summer’s thunderstorm season he hopes the UW team will be using smartphone data to forecast storms and compare their results against traditional forecasts.
If the technique is successful, the researchers hope to supply it to the National Weather Service and the weather bureaus of other countries.
The technique could be particularly useful, Mass noted, in countries that have little weather-forecasting infrastructure but where smartphones are becoming more common.
The research has been funded by Microsoft Corp. and the National Weather Service.
###
For more information, contact Mass at 206-685-0910 or cliff@atmos.washington.edu.
============================================================
I’ve used this app, and it works reasonably well, though is beholden to the whim of the user, so sometimes you’ll see a location you are monitoring drop out if the user leaves the cloud or turns off the device. – Anthony
There’s one for precip –
https://itunes.apple.com/us/app/mping/id584383400?mt=8
Real time display
http://www.nssl.noaa.gov/projects/ping/display/
Or even opens a door/window, leaves the room. Steps into an elevator or aircraft. I am not quite sure if this app will deliver reliable data.
Nobody tell Lewandowsky, he might think it’s another conspiracy.
/sarc, or is it?
John, they’re not that sensitive and not being calibrated they can’t produce accurate results. But aggregated together they can still produce meaningful information as the pressure changes with storm fronts.
Well we need a Surface Stations project to ensure all of the Android pressure sensors are calibrated properly and the phones are not installed in locations of artificial, excessive pressure (like one’s back pocket while in a seated position)?
;->
Seriously, it looks like an imaginative idea and I’m looking forward to the results.
The pressure readings would have to be voluntary as it is fairly clear that they would have to know your location, and being tracked is something we do not want in general.
“I think this could be one of the next major revolutions in weather forecasting, really enhancing our ability to forecast at zero to four hours,” Mass said.
Hmmmm… I assume a “zero hour forecast” is the same as looking out the window, yes?
I like it. Looks like scientists attempting to collect real data.
GPS (or cell) positioning and data communication?
This will drain battery …
Mass is funny: as if he has little clues about atmospheric circulation systems and pressure distribution in these systems.
They will no doubt collect a lot of data, but will they achieve sufficient coverage? I didn’t have to see that map to imagine what it would be like. Enough to remember what cellular coverage maps looked like 20 years ago. Even though the coverage has increased since the, the devices are still concentrated in the same near-useless pattern. It would be much wiser if instead they figured out a way to add a pressure sensor to each cell tower.
The question, why do they need to be calibrated.?
For example, if you have a fixed known calibrated source, if some is near that station, you can calibrate their device remotely. Location give height, and then you get the pressure plus an error offset.
Calibrate on the fly.
higley7 says:
February 7, 2013 at 6:56 am
It is already known where you are, to the closest cell site that is.
The police is using this to check the validity of your alibi.
“You said you were on 5th Ave but you sent a text from Brooklyn at that time, please explain.”
Sounds like a fun idea and no doubt useful to the weather forecasting industry but unless one wants to check the weather via the internet constantly, or listen to the radio/tv, there is a very high probability that it is far quicker and more accurate for your locality to look at the sky on a regular basis in order to get the wash in dry, or to start the outboard and head for shore.
Perhaps they can modify the app in such a way that it will send you a weather warning when and where applicable. But that is useful only when you remain within cell cover, leave the phone on and have it with you.
I take it that there is no interest in other worldwide locations? i.e. UK?
Other than that, I suppose it is a good way of collecting a bucketload of data – though from ‘loosely’ calibrated sensors, it is probably only likely to be classed as qualitative data?
Don’t want to be a misery but other than perhaps ‘seeing’ the approach and passing of a pressure system in real time (via a load of ‘stations’) I can’t really imagine that this will provide a great deal of extra insight – does anyone see any significant scientific benefit?
John says:
February 7, 2013 at 6:30 am
—
I would guess that the app is returning GPS data as well as barometric data. How else would they know where the phone is located. From the GPS data you can determine altitude as well as location.
It might be an interesting side project to figure out how to cross calibrate these things. This could be done by tracking individual phones. Whenever a phone is in close proximity to a known good source of barometric data, you could assign a corrective factor to that phone in your data base. Then when that phone gets in close proximity to other phones, those phones could be calibrated to the first one.
Other possibilities are if a group of uncalibrated phones get into close proximity and one of them is producing a reading that is off from the others, it could be adjusted back to an average of the other phones.
Several good criticisms on this idea already, so I won’t touch on that, but it has got me thinking…
Remember in the movie Twister, they were trying to get a tank full of tiny sensors picked up to offer data from inside the storm? What about doing a similar project on a really large scale using a cellular network?
For example, make a few million of these tiny sensors that record gps, temperature, pressure, etc. They could record data every few minutes, and then once a day, broadcast their collected data?
Of course, the more I think about it, the more this sounds like a terrible idea… some would land in the sun, some in the shade, aggregate data may be hard to interpret, there’s the problem of essentially littering electronics over a vast area, how long would the battery life be…
You know what, better just stick to volunteers with smart phones. I’m in the crowd of “don’t track me, bro” or I would join in.
Crowdsourcing data collection seems to have so much potential, but also must include so many caveats, I wonder if it would ever truly be worthwhile.
Why not temps too that also are coded for shade or full sun. Re pressure sensors, letting a ton of tiny pressure sensors loose into a hurricane or a twister might be also a fun exercise, or seeding the atmosphere from 40k feet from airlines to rain down temp sensors through the atmosphere! Getting a bit crazy here.
outtheback says:
February 7, 2013 at 8:27 am
> It is already known where you are, to the closest cell site that is.
Not good enough, except maybe in the Great Plains, which is one of the areas they most want to study.
One flight of stairs is about 0.01″ hg, three flights is about 1 mb. In hilly terrain, one cell tower covers a lot of elevations.
Big D in TX says:
February 7, 2013 at 9:25 am
One thought I had a while back during one gov’t vs private data collection thoughts was to equip FedEx cargo planes with GPS based autopiloted gliders, about the size of a radisonde, and release them high over Fed Ex sites they pass by.
The glider could glide down to the roof of the FedEx facility collecting information about the air column on the way. The gliders would be collected, memory downloaded, batteries recharged, and sent off with the next load of packages for another flight. I think it would be a lot cheaper than releasing ballons as is currently done.
Cliff has an interesting blog. In the weather is not climate camp. This is his most recent post on this project.
http://cliffmass.blogspot.com/2013/02/smartphone-pressure-observations-take.html
Maybe Anthony could invite a guest post.
Idiots question:- Wouldn’t the accuracy of these figures be influenced by people riding up and down in lifts, especially in tall buildings?
” … to develop better weather forecasting techniques. ”
I’ll believe that when they launch more than just two radiosonde’s (‘weather balloons’) a day (FOR UPPER LEVEL WIND DIRECTION, PRESSURE, TEMPERATURE, AND HUMIDITY) from the limited number of sites now in use …
.
higley7:
At February 7, 2013 at 6:56 am you write:
You are being tracked if you have a mobile phone (i.e. cell phone) on your person and its switched on.
This information about where you are is not normally recorded, but the police have used it to determine where criminals are so they can be arrested.
Richard
SasjaL says:
February 7, 2013 at 7:45 am
GPS (or cell) positioning and data communication?
This will drain battery …
—
GPS is a reception operation and is on all the time.
The cell phones are in communication with nearby cell towers. That’s how the system knows how to route an incoming call to you.
The amount of data that will have to be sent is small. Just a single packet, probably every minute or so.
Wouldn’t the accuracy of these figures be influenced by people riding up and down in lifts, especially in tall buildings?
===========
Depends what altitude the GPS returns. That way you cancel out your height.
That leaves the question of inaccuracy in you sensor. Here you cross check against know accurate sources that are near, or other phones that are near.
I think this is really quite a good project. Only by doing it for real will you get to see if there are errors.
Even if there are calibration errors, a lot of what you are after is change of pressure over time, and for those, calibration errors get canceled out.
(t1+e) – (t2+e) = t1-t1. No such error in the result
I’m skeptical of this for many of the reasons already mentioned. Vertical displacement in buildings is problematic – if you’re inside, you probably won’t get a good GPS fix to determine altitude. GPS accuracy in the vertical is not as good as horizontal anyway, but it’s more critical to get the altitude correct as it will have the strongest effect on pressure.
Download an app to turn on and display your phone’s GPS and watch that altitude vary by 50 feet or more even when stationary.
Even with all the drawbacks, the system is a set in the right direction. I was surprised to read that there are only 1000 pressure sensing station in CONUS.
This is a wonderful idea.Repeatedly the short term forecasts in Sydney are wrong. In the old days you could look up a weather map with barometric pressures and lows marked and work out for yourself if there were a stable low off Sydney.This plus a look at the sky was as good as any forecast.
“Precise tracking of pressure readings and pressure changes could help weather forecasters to pinpoint exactly where and when a major storm will strike.”
Pinpoint ??? I don’t see how it would tell you any more than what radar and satellite data already tell us. And what good is knowing EXACTLY where a major storm will hit? Close only counts with hand grenades. Hurricanes pretty well pretty well screw everybody. If a hurricane is going to strike ten miles up the coast, instead of twenty, do you prepare differently?
This application should be quite useful for people (like myself) who suffer from barometric pressure headaches- being able to predict pressure drops will help one to prepare for them.
I fly gliders. There are already difficulties in using a mobile because of accessing far too many cells at one time. I am interested in the pressure sensor as it could be used in a flight-direction app, but I fear any pressure data contributed to this system may be misleading. I hope they have an anomaly-trap built in.
Gene Selkov says:
I totally agree with this. Each cell tower should have, as part of the main package, a complete weather station. They’re already connected for sending their data, every minute from every cell tower would be a WEALTH of information. They are fixed, with known altitude and geolocation. And most cell towers I’ve seen are more isolated than most USHCN stations, usually in a grassy isolated fenced area in residential suburbs. Heck, throw in a camera or four and you’re laughing!
Imagine the usefulness if weather models could start with thousands of real-time weather sensors.
Already we have highway cams all up and down the main highways with elementary weather stations, I check them every time I’m about to go for a long drive, especially in winter, and most cities have their traffic cams and intersection cams available online.
Those of us with Smartphones (I have a Galaxy Nexus, btw, but the PressureNet app says I don’t have a pressure sensor) have more computer power in our phones than a high end PC did just a few years ago. It makes sense to find ways to harness the potential of all that mobile, GPS’d computer power. It would also be trivial to compare pressure readings from mobile devices with the static reading from the cell tower when they’re close, to map out the accuracy of the device.
I think it’s a solution looking for the right problem… but I like seeing people thinking about this.
Most people live in cities.More DATA Corrupted by Urban Heat Island.