High-volume weather blog Cliff: Pressure data from smartphones can dramatically improve weather forecasting: Will big tech help?

Meteorologists need Big Tech’s help to improve weather prediction and save lives.

How? By collecting pressure from smartphone.

This blog asks for their support.

There is no more useful weather observation than surface pressure. With surface pressure alone, one can determine the three-dimensional structure of the atmosphere, sense precursors of severe thunderstorms, and much more.

Many smartphones include excellent pressure sensors, mainly to help determine altitude.

With billions of smartphones with pressure sensors around the world, has the potential to significantly improve weather forecasting, especially in third world countries with sparse weather observations and severe thunderstorms in the US

There are research groups, including mine, that have solved technical problems with smartphone pressure calibration using machine learning and anonymizing observations.

But there is one big problem: Collect pressure from smartphone. To solve the problem, we need the help of technology
Community: Apple, Google, Amazon, Samsung, Facebook, Twitter, Instagram, Microsoft or others.

Help that you can revolutionize weather prediction and save many lives.

The proven potential of smartphones to dramatically improve weather forecasting

The current generation of smartphones has pressure sensors as good as those used in many weather stations.

A decade ago, some of us realized the potential and found one source: IBM/The Weather Channel app. (IBM bought The Weather Company, which includes the Weather Channel smartphone app.) When people checked the forecast using the Weather Channel app, their barometric pressure was retrieved and stored.

With a special collaboration from IBM, IBM has allowed UW access to the pressure of smartphones, about 60 million observations per day mainly in the US

Here is an example of the pressure distribution of smartphones at an hour in 2018. There is an unusual density of observations in the eastern half of the United States, but very much in the West. Remember that this sample is from a single app reporting from one out of a thousand smartphones.

To show the value of smartphones, consider the landfall of Hurricane Michael on the Gulf Coast on October 10, 2018 (see below). The top right panel shows typical data collection (each dot is a pressure observation, color coded by pressure – red is high, green/blue is low. When power is out, multiple locations. Traditional observation does not have a network connection, but the number of observations on a smartphone is very large and clearly shows the position of the eyes.

For the recent devastating Marshall Fire near Boulder, Colorado, smartphone pressure identified low-pressure features that cause strong winds (dark red on the map below) – something the weather network transmits system omitted.

I could show you more examples like this, such as for a strong thunderstorm, where the pressure of a smartphone outlines important details.

I can also show you forecasts where smartphone pressure has improved the forecast. To provide a local example, the figure below shows you a radar image (left panel) showing the Puget Acoustic Convergence Zone (inside the red circle), a frequent weather feature of western Washington. On the right are two forecasts, one that uses smartphone pressure and the other that doesn’t. Without smartphone observation, the forecasting system does not predict the convergence zone (center panel), but with smartphone observation in the area it does!

Overcoming technological barriers

There are a number of technical hurdles that need to be overcome and this was made possible by my recent PhD. student, Callie McNicholas.

The first day, each phone has a different tolerance/pressure bias and the heights of the phones are also different. I won’t go into details, but using machine learning and the GPS altitude/location information given to each phone, we were able to successfully calibrate the phone’s pressure.

Second, because people care enough about privacy, we had to anonymize pressure observations. We did this by adding random noise to the position and collecting pressure in small boxes with some observations. Even with our steps to ensure privacy, smartphone pressure (smart-anon pictured below) has less error than current National Weather observations (MADIS). An extraordinary achievement by Dr. McNicholas!

Help the second and third world

We were able to demonstrate that even moderate smartphone pressure densities provided by IBM/WeatherChannel can help improve our analysis of surface pressure and improve forecasts for the future US. data rich. smartphone users but poor meteorological observation network. The surface observation map used by the European Center for Mid-Range Weather Forecasts for December 2020 is shown below. A LOT observed in Europe, but large holes in Africa and South America.

Why we need Big Tech’s help

We will be able to dramatically improve weather forecasting if meteorologists can secure more pressure data from smartphones.

And only big technology can provide this data. Only they have apps on multiple phones or control the operating system on phones.

For example, Google, which controls the Android operating system and key apps – like google maps, can solve the problem.

Can anyone in Google help? Or contact someone who can help in Google?

Apple, which controls the iPhone operating system and many apps, can make that happen.

Facebook or Twitter, with large numbers of users of their apps on smartphones, can provide data.

Amazon has apps on many smartphones.

You get the picture. We can dramatically improve weather forecasting,
but need the support of one or several large technology companies.

Are any readers of this blog able to help?

If so, give me a call, email or leave a comment.