How WeatherSignal can contribute to forest fire monitoring

Knowing that a paper was published using our data always gives us a thrill. We enjoy the fact that our work can meaningfully contribute to different fields of research. Above all, we see it as proof of the value of crowdsourced data and the reliability of smartphones’ sensors.

This is certainly the case with Sagi Dalyot and Shay Sosko’s most recent paper, “Towards the Use of Crowdsourced Volunteered Meteorological Data for Forest Fire Monitoring”. You might remember these two researchers at Technion from a previous blogpost on our academic partners. We recently learned that their work was distinguished with the Best Paper Award by IARIA, the International Academy, Research, and Industry Association. Our congratulations go to them!

A big cheer for Sagi and Shay!

A big cheer for Sagi and Shay!

Sagi and Shay’s study focus on the evaluation of smartphone-gathered meteorological data, as a means to complement weather station data for the purpose of early fire detection. Fire spreading simulations, as well as fire danger rating systems, are largely based on two types of meteorological data: ambient temperature and relative humidity. Nowadays, some smartphones have incorporated sensors to measure both. Sagi and Shay chose the Samsung Galaxy S4, one of the first cellphones to include these sensors – integrated in the SHTC1 chip made by Sensirion. The app used to collect the data was our very own WeatherSignal.

The accuracy of the measurements was evaluated by implementing three different scenarios to gather the data, varying in duration and location – from series of short measurements to long periods of continuous collection, with the phone alternatively situated in the shade and exposed to direct sunlight. The data thus collected was then compared to that from weather stations.

The researchers found that when the smartphone was placed in a shadowed space, the measurements were accurate and reliable. And although exposing the device to direct sunlight resulted in some erroneous readings, a calibration algorithm developed by the authors permitted them to correctly identify these and discard them. Sagi and Shay went on to map how temperature readings crowdsourced by WeatherSignal can complement weather stations’ data, filling the gaps in areas not covered by meteorological stations.

Their conclusions are very positive and point to the possibilities offered by crowdsourced data from smartphones’ sensors: “with relatively small post-processing, and without having the need to use reference data to analyze the correctness of the data, the collection device can function as a reliable and accurate ‘dynamic geosensor station’ that serves as supplementary data source that is external and independent to the static network”.

In a time when new sensors and smart devices, released on a daily basis, are being tested as earthquake detectors and all-encompassing health-monitoring tools, their promise is turning more and more into reality.

This entry was posted in Academic, Crowdsourcing, Sensors, WeatherSignal. Bookmark the permalink.

Leave a Reply