One Week of WeatherSignal

It’s been just over a week since we launched WeatherSignal, we billed it as “the most ambitious Weather crowdsourcing project ever” – that’s a fine claim to make, let’s see if we’re managing to live up to it.

By the numbers:
9,908 users
37k app sessions
2.3 million data points

And some very kind articles on The Next Web, TechCrunch and New Scientist.

While that might sound fairly impressive, this is very early days. On a single good day OpenSignal is downloaded 8000 times, and on our best day ever it was over 15,000. We want WeatherSignal to be be that big, in fact we’d rather like it to be even bigger, as with OpenSignal: the more people using it the more useful it will become.

What’s the data like?

We’ve received
2,305,086 datapoints
contributed by
9,836 users

Numbers, boring numbers – show me some graphs
Here’s pressure from the weather station at Milan’s Malpensa Airport offering the ‘official’ view of pressure for the last week:

pressure_mxp

And here’s what we measured on phones with barometers surrounding that same airport:
presure_app_max
Note how similar the pattern is, but our data is offset – this is because Milan is 121m above sea level and the weather station data corrects for this, this is something we’ll need to start doing.

How the light gets in
Ever wondered what the daily pattern of light intensity incident on your phone looks like? Here’s a week’s worth of Milanese sunshine:
light_app_mean

Does your phone comply with Newtonian Physics
Here’s another fun thing to look at: at different latitudes your weight varies, it does, seriously. At the equator you are lighter. There are two reasons for this: at the equator you are further from the centre of the Earth so the gravitational force is weaker, the second reason is due to the spinning of the Earth which reduces the resultant acceleration. The acceleration should vary between about 9.78 metres per second squared at the equator and 9.83 at the poles.

Here’s what we’ve measured: the below graph shows the absolute latitude at a resolution of 1 degree – so each dot, except for the one at 0, represents all the readings from users in 2 latitude bands – one above and one below the equator. The size of the dot shows the number of readings. The acceleration is the average of all accelerations between 0 and 20 m/s2 in those latitude bands – we figured the phone gets moved in lots of directions and positive accelerations relative to gravity will cancel out negative ones and we’d end up with the “rest” acceleration – the pull the phone feels due to gravity and the Earth’s spinning. The blue line is the best fit, but because it’s late it’s not been properly weighted :)
gravity_by_lat
So it’s a little messy, not nearly as smooth as the lovely (theoretical) illustration you’ll find here (thanks Berthold for sending me that), but as more sensor data comes in we should get good empirical confirmation of the wonderfully named “international gravity formula” (as if gravity could fail to be international!)

We’ve got our work cut out. In particular on cleaning up temperature readings – which we primarily get from reading battery temperature sensors – this deserves a blog post on its own. In the meantime, keep using the app – it’s not just for weather, you can use it to quantify yourself, conduct physics experiments or just to see what Atmos will do next.


That's Atmos - the shoes mean he's walking (i.e. you're walking), the moon's in its first quarter & he's got a hat on because it's cold.

That’s Atmos – the shoes mean he’s walking (i.e. you’re walking), the moon’s in its first quarter & he’s got a hat on because it’s cold.



Get the app – to see what else Atmos can do.

This entry was posted in WeatherSignal. Bookmark the permalink.

Leave a Reply