Space Apps Challenge 2018

Carpe Ignis, meaning Seize the Fire, is a web-based crowd-sourcing tool for early wildfire detection and visualization conceived by a team of computer geeks in Montreal.

In an era of climate change and increasing urbanization, with the deforestation that comes with it, wildfires are posing an increasingly bigger risk to not only residents nearby, but also to infrastructure and wildlife. It is why we believe that the early detection of such wildfires, and fires in general, is of the utmost importance in helping humans deal with them.

Carpe Ignis lets people submit either textual or multimedia information to declare a wildfire in their vicinity. This submitted information is then stored in Carpe Ignis' database, where location is stored as a latitude and longitude.

The report is first run against several checks. First, machine learning uses NASA data to compute the likelihood of it being a fire or not depending on the specific location and time of year, going off on the assumption that wildfires in Northern Canada, for example, are pretty rare. If a picture or a video has been submitted, machine learning is used to determine with reasonable certainty whether the image depicts actual fire. If the report cannot be rejected with an extremely high confidence level, a notification then goes out to a randomly selected group of users nearby so that they can confirm whether a fire is indeed present or not. If enough positive reports are received, the status of the fire is updated to confirmed and Carpe Ignis takes charge of notifying concerned parties.

To notify people, Carpe Ignis operates mostly on a voluntary basis. People can sign up for notifications by entering their name, email and phone number. Location data is then accessed with the user's approval using HTML5 functionalities. This latitude and longitude is then stored, and when a new fire report is confirmed, the users within a 50 km radius are notified by text and email.

Finally, Carpe Ignis also provides a visualization widget embedded into the website, where people can see the localisation of fires detected by NASA satellites during the current day. It is possible to scale the map and scroll to it to in order to visualize fires located in other countries too.

We have some improvements we would like to release for Carpe Ignis:

1. Integration of CO levels measured by MOPITT to give better advice to users about evacuation and health effects of nearby wildfires
2. Pair NASA data with data from Hadfield's Rapid Burned Area Mapping to have an extra check
3. Improve notifications with Google Maps APIs to target non-subscribers
4. Integrate MODIS (Moderate Resolution Imaging Spectroradiometer) data for increased redundancy check of fire reports.
5. Enable certain persons to get verified (firewatchers, government agencies) so as to be able to skip some redundancy checks for more rapid detection.

Our open repository is visible here

Special thanks to:

• Canadian Space Agency (Raw data - MOPITT, R-BAM)
• NASA (Raw data - FIRMS, InciWeb, MOPITT, MODIS/VIRIS)
• StackOverflow
• MDN Web Docs (Form submission)
• W3Schools (HTML, CSS, JavaScript, PHP)
• GeeksForGeeks (Sending emails using Java)
• Google Maps API for the web map
• PorkBuns for the website