O.B.L.I.V.I.O.N| Spot That Fire V3.0
Awards & Nominations
O.B.L.I.V.I.O.N has received the following awards and nominations. Way to go!
Spot That Fire V3.0
O.B.L.I.V.I.O.N: Spot That Fire V3.0
Summary
When fires burn too hot and uncontrollable or when they’re in the “wild land-urban interface” (places where woodlands and homes or other developed areas meet), they can be damaging and life threatening (Wildfire).Our aim is to predict the burnt area of forest fires, based on the spatial, temporal, and weather variables from where the fire is spotted. This prediction can be used to calculate the urgency and the intensity of the fire.Later in, an app front end along the above back end help us in informing the authorities, getting people to a safe location and bringing forward volunteers.
How We Addressed This Challenge
We basically started off, developing a model to predict the area that would be affected in the case of a reported fire. We then decided to incorporate this idea into an android app. In addition to this, we also plan to add features like report a fire, volunteer to help and safe routes.
Predicting the amount of area that the fire will affect is crucial in evacuation processes, among many other things. This can give the authority and the firefighters a clear view on the location of the fire and the people living in the affected places, if present. In addition, anyone can simply capture a photo and report a fire near them. We also provide the option to volunteer, which could be a huge relief for the firefighters, Volunteering will not include dangerous tasks but will concentrate more on guiding people to safe spots and such.
Even though the app hasn't been fully developed, we hope to complete this in the future, because this app can prove to be a difference and can help increase the overall effectiveness when dealing with a scenario of a wildfire.
How We Developed This Project
Over the past few years, we have all seen news of wildfires across the world resulting in huge death counts, both human and wildlife. We have all wondered why this is happening so often. Climate change is not doing any favors so it is upto us, and us only, to make a difference.
This is what we are striving for, to make a difference and help people who're going through this. That is why, as soon as we saw this challenge, we knew we had to take this because this is something the world needs right now - a sustainable and effective solution or preventive measures or even post-disaster management.
We are building an android application that enables the users to report forest fire near their areas so that others using that application will come to know about it and they can volunteer for rescue operations there. We've integrated an ML model with the app which uses previous forest fire datasets and predict the intensity of the fire, thereby classifying disaster-prone and safe areas. The result we get from the model help volunteers in rescue operation as they can find out which all areas (nearby the place where the forest fire occurred) are safe and then do the rescue operations accordingly. We've used KNN algorithm to predict and then linear regression to classify the areas into safe ones and fire prone ones. This was done in colab (using python) and then integrated into the android application (Android application was built in Android studio using java).
We achieved an overall 50.26% accuracy in the model we developed, and the only problem we faced was we couldn't complete the app development. Again, all the progress made open source, so we can and will complete this in the future.
Our App's UI:
How We Used Space Agency Data in This Project
To determine the intensity of the fire that occurred, which is our model's main goal, we require data such as Temperature, Weather, Climate, Wind and FWI data. Active fire data were obtained from NASA's Fire Information for Resource Management System (FIRMS). Using GeoNEX's geostationary satellite sensors, temperature and weather of the specific region are obtained (particularly with the aid of Geostationary Operational Environmental Satellite (GOES)).
For developing the app, the above mentioned resources along with NASA's satellite imaging and google maps is vastly used to procure a safe route for the endangered people. The app in later stages can be developed on alerting the higher authorities if the fire is deemed dangerous and also, provide an interface for volunteers to come together. We have used open-source code, developed by Spot that Fire version 1.0 contestants, to create the app and add new features into it.
Later on, we plan to use NASA's MOPITT data to find the concentration of CO gas, develop a model based on its spread, integrate it into the app, thereby alerting the citizens on different places' safety according to their search.
Project Demo
PowerPoint presentation detailing our approach :
https://docs.google.com/presentation/d/17ee3TGBNf5xDRYKB7kxQfQyi7RleegdpB5C2a00ipxw/edit?usp=sharing
The video addressing the problem and our approach in finding the suitable solution:
Data & Resources
- P. Cortez and A. Morais. A Data Mining Approach to Predict Forest Fires using Meteorological Data. In J. Neves, M. F. Santos and J. Machado Eds., New Trends in Artificial Intelligence, Proceedings of the 13th EPIA 2007 - Portuguese Conference on Artificial Intelligence, December, Guimaraes, Portugal, pp. 512-523, 2007. APPIA, ISBN-13 978-989-95618-0-9. Available at: http://www.dsi.uminho.pt/~pcortez/fires.pdf
- Marj Tonini, Mirko D’Andrea, Guido Biondi, Silvia Degli Esposti, Andrea Trucchia and Paolo Fiorucci, "A Machine Learning-Based Approach for Wildfire Susceptibility Mapping. The Case Study of the Liguria Region in Italy", Geosciences (Switzerland), March 2020 10.3390/geosciences10030105
- NASA FIRMS - https://earthdata.nasa.gov/earth-observation-data/near-real-time/firms
- NASA GeoNEX - http://www.nasa.gov/geonex
- Spot that Fire version 1.0 - https://2018.spaceappschallenge.org/challenges/volcanoes-icebergs-and-asteroids-oh-my/real-time-fire-app/details