Smartphones can alert us to wildfires and extreme weather conditions

A new study from Tel Aviv University has found that the smartphones we all carry in our pockets could help collect weather data from the public to provide early warnings for extreme weather, such as wildfires.

The researchers explain that all smartphones are equipped with multiple micro-sensors capable of collecting important environmental data, such as temperature, barometric pressure, humidity, magnetic field, light, sound, location, acceleration, gravity, and more. These data help us find our way or define our location, they warn us when the battery overheats or the device absorbs moisture—all in real time, without saving the data.

The researchers demonstrated that smart use of such data could support early warnings for severe wildfire events, especially since millions or even billions of data points are collected worldwide every day by our smartphones.

Today, early warning systems in remote forested areas typically lack data due to the absence of weather stations in remote locations. The public, however, take their smartphones everywhere, with each phone containing sophisticated micro-sensors that operate continuously in the background, but these data are normally lost and not saved. However, many companies have started to collect smartphone data to use for various purposes, with user consent.

The researchers believe that this huge data source (as of 2024, there are over 7 billion smartphones worldwide) could aid in forecasting extreme weather and natural disasters.

Prof. Colin Price and Ph.D. student Hofit Shachaf from the Department of Geophysics in the Porter School of the Environment and Earth Sciences at Tel Aviv University used data collected from the global public via the WeatherSignal app (OpenSignal) to develop a methodology for assessing wildfire risk based solely on smartphone data collected by the public. The results were recently published in the journal Natural Hazards and Earth System Sciences .

One key parameter determining the likelihood of a wildfire is the moisture content in vegetation (essentially the fuel available for the fire)—which, in turn, is determined by the temperature and relative humidity of the surrounding air. Both the temperature and relative humidity can be easily obtained from the public’s smartphones.

Hofit Shachaf explains, “We developed an index based on VPD (vapor pressure deficit), which reflects the dryness of the vegetation based on environmental parameters (temperature and humidity) . In hot and dry atmospheric conditions more moisture is drawn from the plants, due to enhanced evaporation and transpiration (evapotranspiration) that essentially facilitates ignition of fires. In cooler, more humid air, forests do not usually catch fire since their moisture level is too high.

“Previous studies have also used VPD to predict wildfire risk levels—though the data for these calculations is mostly obtained from local weather stations. The novelty of our study is the utilization of data collected from smartphones, without their users’ active participation, to calculate VPD over large areas and at high spatial and temporal resolutions—providing important insights into wildfire risk evaluation.”

However, smartphone data do contain errors. The temperature reading might reflect the air conditioning in your office, while the humidity sensor might identify moisture when the user is taking a shower. But the huge amount of data collected from smartphones allows us to remove outliers in the data set.

Furthermore, since the micro-sensors are not calibrated before they are put in our phones, it was necessary to first calibrate the local smartphone data against commercial meteorological stations. This procedure turned out to be relatively straightforward, with just a single calibration needed to correct a smartphone’s readings.

After calibrating or “training” the device, the researchers analyzed two major wildfire events: fires in Israel in November 2016 and the massive fire in Portugal in July 2013. The results were surprising, with smartphone data collected from the public showing significant VPD anomalies before and during these major fires.

Hofit Shachaf adds, “It’s surprising, but even though each smartphone has its own errors and biases, with large amounts of data from many smartphones, we can average out the errors and still retain useful data. The large volume of data helps overcome issues associated with individual smartphones.”

Global smartphone coverage has increased by about 30% in the last five years. As their global density continues to grow, data collected from smartphones could eventually offer better spatial resolution than traditional meteorological networks—especially in urban areas where natural disasters like fires and floods, can have a significant impact. Moreover, in low-income countries lacking meteorological infrastructures, smartphones could provide useful data for monitoring extreme weather conditions.

Prof. Price concludes, “Given the rapid increase in the number of smartphones worldwide, we propose utilizing this data source to provide better early warnings to the public and disaster managers about impending natural disasters. Better early warnings could prevent natural hazards from becoming natural disasters.”