“Predicting Canada’s Wildfire Future: How Phone Data is Leading the Charge”

Klamath National Forest, Yreka, United States, 2017

A recent investigation has discovered that the smartphones we carry could assist in gathering weather data from the public to issue early alerts for wildfires and other severe weather phenomena.

All smartphones come equipped with numerous micro-sensors that are adept at collecting critical environmental information, such as temperature, atmospheric pressure, humidity, magnetic field, luminosity, sound, geographic location, acceleration, gravity, and more. This data aids us in navigation and location identification while alerting us when the battery is overheating or when the device encounters moisture — all in real time, without storing the information.

Professor Colin Price alongside student Hofit Shachaf from the Porter School of the Environment and Earth Sciences utilized data gathered from the global public via the WeatherSignal app (OpenSignal) to create a methodology for assessing the risk of wildfires solely based on data collected from smartphones by the public.

The findings were published in the journal Natural Hazards and Earth System Sciences (NHESS).

The researchers illustrated that effective utilization of such data could facilitate early warnings for significant wildfire incidents, especially because millions or even billions of data points are amassed globally each day through our smartphones. Currently, early warning systems in isolated forested regions often lack information due to the unavailability of weather stations in these remote areas. (Some researchers are utilizing goats, dogs, and elephants to forecast the weather!)

While we take our devices everywhere, the data is usually discarded and not retained.

Collin Price

Nevertheless, several companies have started amassing smartphone data for various applications, with the consent of users. The researchers assert that this vast data resource could assist in predicting extreme weather patterns and natural calamities.

A crucial factor in assessing wildfire probability is the moisture content present in vegetation (essentially the combustible material for the fire), which, in turn, is influenced by the temperature and humidity levels of the surrounding atmosphere. Both temperature and relative humidity can be readily gathered from the smartphones of the public. However, smartphone data may contain inaccuracies. The temperature reading could reflect the air conditioning from your office, while the humidity sensor might detect moisture when the user is taking a shower.

The researchers assert that the substantial volume of data gathered from smartphones permits the elimination of anomalies from the data set. Moreover, since the micro-sensors are not calibrated before integration into our devices, it was essential to initially calibrate the local smartphone data against established meteorological stations. This process turned out to be fairly uncomplicated, requiring only a single calibration to rectify a smartphone’s readings.

Subsequent to calibrating or “training” the device, the researchers examined significant wildfire incidents: a notable one was the massive blaze in Portugal in July 2013. The outcomes were unexpected, with smartphone data acquired from the public revealing considerable anomalies prior to and during these major wildfires.

“It’s astonishing, but even though each smartphone harbors its own errors and biases, with the aggregation of large volumes of data from multiple devices, we can average out those errors while still maintaining valuable data,” Shachaf remarks. “The extensive quantity of data serves to counteract the issues linked with individual smartphones.”

“Considering the swift rise in the number of smartphones globally, we advocate utilizing this data source to enhance early alerts to the public and disaster management teams regarding impending natural crises,” Professor Price concludes. “Improved early alerts could avert natural hazards from escalating into natural disasters.”

Comments

comments