I'm working on using smartphone sensors to detect anything they can relating to weather, including space weather - magnetic storms. I've been fascinated with crowdsourcing mobile sensor data for like 10 years now and it's only going to get more interesting: air quality sensors should be making their way to phones soon, others as well.
My open source library with limited sensor features is here: https://github.com/JacobSheehy/AllClearSensorLibrary for Android 'background' access to sensors (using foreground library). I don't think that I've got magnetic data there yet but it's mostly a matter of adding another field.
I typically track hurricanes with obvious low-pressure curves but there's a lot more that can be done.
Crowdsourcing sensor data from phones:
Accelerometers can detect earthquakes, Microphones can detect noise pollution, Magnetometers can detect solar activity, Barometers can detect weather changes
Most phones have those sensors now, and I'm looking out for UV sensors, air quality sensors, and more. The applications of crowdsourcing sensor data are immense and constantly improving as new sensors get added.
--
The main issues honestly are privacy and noise filtering. Ensemble kalman filters running on device can reduce noise dramatically, and you can use anonymizing features like geographic bounding boxes (where sensors can go in and out and only their bounding box is used as location data)
You might be interested in Luftdaten.info. I could even see a collaboration between your projects, as I'm fairly certain they are all about open-data and opensource. I personally just discovered them and ordered an EnviroPI+ and a Raspberry Pi Zero W (with an L-shaped GPIO extender and case with a big hole to avoid heat discrepancies from the Pi) so that I can also contribute. Would also love to run your app but I'm in Europe :(
I can see benefits for local phenomenon like air quality and even temperature. But what’s the advantage for using cell phone arrays for cosmic scale events like solar storms? Are the effects of solar storms that different over small geographic areas?
The article headline has been simplified a bit[1]; the storms in question are geomagnetic storms[2]. Geomagnetic storms are perturbations of Earth's magnetic field by changes in the solar wind (i.e. solar storms).
Having a high-spatial-resolution record of the way the Earth's field changes when interacting with the plasma presumably yields extra insights into how space weather behaves.
I don't know, but there's only one way to find out!
Lots of people told me that there was no use at all in measuring air pressure on scales smaller than the US already does with official measurements. But years later, there are multiple papers published and I think IBM now uses their mobile apps to crowdsource barometer data as it improves their weather forecasts.
I'm sure there will be naysayers who even say crowdsourcing magnetometer data is useless, but I don't really care or think they're right.
AFAIK there isn't a large-scale effort yet to study crowdsourced magnetometer data, so we really just don't know yet.
> Are the effects of solar storms that different over small geographic areas?
I don't know but now I'm extremely curious how small the measurable variations are in solar storm effects on Earth. And amazingly, there are billions of deployed sensors all around Earth right now that are simply not contributing to a central network. A simple code library is all that's needed to know this answer, which to me, is amazing, and a reason enough to try it out.
Yeah after considering this for the last few minutes it makes more sense as just capturing all available sensor data in advance of any specific question.
I’m not sure that solar storm data specifically is useful, maybe interesting. But widespread magnetometer data could have all kinds of unexpected applications.
Detecting solar storms with such an array just seems like a proof of concept.
if you want to kill your battery, yes, it does make a very inefficient (small area) ionizing radiation detector. Keep in mind that the size of the sensor is O(1 cm^2). It would be easier if the phone had a shutter.
The Academic Times post contains multiple inaccuracies and incorrect statements, as if the author only has a very approximate idea about what they are writing about (source: am a space scientist).
The conclusion is they are not, so I guess it's good someone did the study then [7]
I'm not sure there's any real purpose to doing this since proper machines can record it. Unlike earthquakes for instance were local data might be useful.
Using the magnetometers for geo-location is a more interesting use I've heard about. Given a GPS location using local interference with the magnetometer as beacons to further pinpoint where you stand. Are you near the metal door. Or if you could create a beacon to mess with it.
[7] "The conclusion from this study is that some smartphone models are indeed capable of detecting and measuring geomagnetic storms with a high enough fidelity to be of some value to studies of actual major storm events spanning a wide range of latiudes. When taken under environmental conditions where the temperature is stable, where the smartphone is situated on a level, vibration-free, non-magnetic tabletop, and where no external electromagnetic interference (motors, air conditioners etc) are present, sensitivity levels can closely approach the ±0.2 µT noise limit of the electronics and allow the detection of the stronger geomagnetic storms.
[8] "When you go out for a walk, run, or other outdoor activities, use CrowdMag to measure the magnetic data along your path and save the data as a “magtivity”. " - https://www.ngdc.noaa.gov/geomag/crowdmag.shtml
I only skimmed but I didn’t see any suggestion of leveraging this capability for any benefit. Is there some advantage to making an array of solar storm detectors from cell phones instead of pre-existing and more capable specialized equipment?
I haven't studied the data that comes from crowdsourced phone magnetometers yet, but I expect it would follow some properties of other data I've studied, which is
- There are vastly more of these sensors and they are located in places where specialized equipment is not
- The data quality per sensor is very low. You typically need a lot of sensors in a nearby geography in order to get any kind of real data out of it
- Often those geographies where you have lots of smartphones aren't covered by the specialized equipment, so you get new data that you otherwise didn't have
What I don't know is how valuable that extra coverage is or how valuable solar storm data is in the first place.
Yeah, but don't you need just a handful of specialized sensors for 100% coverage? To my knowledge solar storms are massive phenomena that are effectively homogeneous over areas hundreds or even thousands of miles wide.
They are already used as earthquake detectors. Each one is not as sensitive as a professional seismograph. But thousands of times as many make up for that.
> Developed by the nation’s leading seismologists, the ShakeAlert system uses signals from more than 700 seismometers installed across the state by USGS, Cal OES, University of California Berkeley, and the California Institute of Technology.
They just send notifications to the phones if there is an earthquake. They don’t use the phones themselves as a sensor.
>All smartphones come with tiny accelerometers that can sense signals that indicate an earthquake might be happening. If the phone detects something that it thinks may be an earthquake, it sends a signal to our earthquake detection server, along with a coarse location of where the shaking occurred
Here in the PNW the one(s) we tend to worry about most will be close in but offshore where the sea floor dives under the continental plate. Buoys can be close to the source and transmit fast enough to give some of us a minute or two warning thanks to the wave propagation speed difference between radio in the air and P & S waves through the earth.
Had lunch years ago with a really interesting guy and friends in SF around this topic. He was writing the software in Java, back when you could run that on android devices. Wish I could reconnect, but sadly gmail is failing with searches now.
My open source library with limited sensor features is here: https://github.com/JacobSheehy/AllClearSensorLibrary for Android 'background' access to sensors (using foreground library). I don't think that I've got magnetic data there yet but it's mostly a matter of adding another field.
I've got this running live in an Android app if you want to check it out: https://play.google.com/store/apps/details?id=com.allclearwe... (US-only for now, sorry about that, international weather data is pricey)
I typically track hurricanes with obvious low-pressure curves but there's a lot more that can be done.
Crowdsourcing sensor data from phones:
Accelerometers can detect earthquakes, Microphones can detect noise pollution, Magnetometers can detect solar activity, Barometers can detect weather changes
Most phones have those sensors now, and I'm looking out for UV sensors, air quality sensors, and more. The applications of crowdsourcing sensor data are immense and constantly improving as new sensors get added.
--
The main issues honestly are privacy and noise filtering. Ensemble kalman filters running on device can reduce noise dramatically, and you can use anonymizing features like geographic bounding boxes (where sensors can go in and out and only their bounding box is used as location data)