This project is an evolution of http://fires-map.com/. The aim of the project is to diminish consequences of smoke created by wildfires by providing a tool for prediction of smoke propagation.
Smoke from landscape fires is one of the most dangerous consequences of fires, leading to pulmonary and heart disease in populated areas [1]. Forest fires smoke contains 20% of all worldwide pollutants (inc. PM2.5) [2]
Smoke from wildfires kills approximately 340,000 people each year (e.g. only ~265 000 direct deaths from all types of fires) [3].
During the hackathon we have:
- Met each other and built a strong team
- Identified and verified the need for the service described
- Discovered data sources for the model and implemented data fetcher
- Choose the Siberian last year fires in Russia as a pilot region for model training
- Built the heuristic to classify smoke
- Developed requirements for a spatiotemporal prediction model (Including density-based clustering for hotspots, autoregression and convolutions for matrix data processing)
- Developed mockup for a mobile prototype of SmokeAlert
- Prepared visual demo
- Discovered and fixed multiple bugs in georasters Python library
To learn more about our project
Our source code
Our team:
- Georgy Potapov, creator of fires-map.com, Crisismap
- Dmitry Pogodin, backend-developer
- Maria Kolos, UX / UI developer
- Oleg Urzhumtsev, data scientist, Skoltech, Cisco
- Rustam Akhtyamov, PhD-student, Skoltech Space Center
[1] http://ec.europa.eu/environment/integration/research/newsalert/pdf/294na2_en.pdf
[2] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3346787/
[3] http://www.who.int/mediacentre/factsheets/fs365/en...