Awards & Nominations
Smokebusters has received the following awards and nominations. Way to go!
Automated Detection of Hazards
Smokebusters
Summary
Our proposal is to provide an interactive and free platform for the user with the objective of informing and providing real-time data on climatological aspects and fires with the use of an extensive network of data and machine learning, where the program will get us provide real-time predictions of possible fire outbreaks as well as the distinction between their types, whether natural or criminal, enabling the issuing of alerts to the competent authorities in order to combat it in the most agile and efficient way possible, using the data acquired by the machine in real time (relative humidity, active air masses, local vegetation status, etc.)
How We Addressed This Challenge
Through a website, we will provide an interactive display with the user so that it can analyze outbreaks of fires and also have the possibility to check other variables such as wind. The solution will be to analyze the pixels of the images and understand the history of fires that occur in this region so that the software can understand the possibility of this pixel catching fire. As a confirmation loop, we will also use heat radiation analysis, known as a hotspot, and possibly aerosol emission spots.
Across automated fire detection, it is possible to alert local authorities in an agile way, enabling them to act earlier and earlier, reducing the consequences generated by fires. With a smaller burnt area it is possible to more easily determine the origin of the fire, facilitating the discovery of those responsible for starting it.
How We Developed This Project
As a result of the intense fires that took over much of Brazil throughout the year, our team felt challenged to take up this challenge for themselves and look for applicable solutions to predict and reduce the damage caused by fires, which affect not only biodiversity but also the all of us. To develop the project we were inspired by information programs such as GreenPeace's Global Fire Dashboard portal and NASA's Fire Information for Resource Management System, which have a wide network of information that we use to compose our library and carry out the analysis and learning of machine. In addition, we developed a prototype of the website where you can have a central idea of how the program works. The difficulties we encountered were to consolidate the automation of processes, which after countless brainstorms and research we managed to find a solution.
How We Used Space Agency Data in This Project
We used NASA GIBS and JAXA Imagery data to observe pixels history and teach an algorithm when a wildfire happens. The NASA data has wildfires reports that can indicate where and when a wildfire happened on the map. Moreover JAXA data can indicate temperature on a surface area and aerosol levels. Putting all those information together we can determine a pattern that happens when there's a wildfire on the map.
Project Demo
https://we.tl/t-1dpy0zUJST
Data & Resources
NASA GIBS and JAXA ASMR, An insight into machine-learning algorithms to model human-caused
wildfire occurrence (RODRIGUES, 2014) and Spatiotemporal prediction of fine particulate matter during
the 2008 northern California wildfires using machine learning (Reid, 2015)