2.5 days of using machine learning to support climate action by predicting and preventing forest disasters.
Our planet is already suffering the consequences of climate change. We have recently seen the devastating consequences of bush fires out of control in Australia, with researchers claiming that as temperature rises, do does the risk of wildfires around the world becoming a common pattern. There is an evident and urgent need to address the effects of climate change, and AI for Good can be a driving force to prevent disasters an help us act at the right time to preserve our planet.
Satellite imagery and remote sensor data present an opportunity to develop machine learning models that allow not only to predict potential threats and plan accordingly, but also to implement sustainable solutions for the future. By using computer vision and image analysis for example, robust predictions can be made about forest areas, their risk and their potential for sustainable development.
- Accurate risk assessments and prevention of forest fires
- Estimating carbon stock
- Categorisation of trees and impact on air quality in the area and other variables
- Developing effective mechanisms of fire resilience
- Insights for sustainable development
- Build machine learning models that would predict the likelihood of a severe disaster occurring: bush fires, etc.
- Predict the impact of those disasters
- Build machine learning models that allow to have preventive measures in place in case of disaster
- Predict the impact of preventive measures in place and formulate best practice approaches
- and more!
Participants will work in cross-functional teams, tackle one or more (but not limited to) of the following tasks based on their expertise:
- Categorisation of trees and areas based on satellite imagery
- Estimating tree height based on ML models simulating LiDAR
- Presence (or not) of trees and relationship with air quality
- Trees dependent on time for a given area
- Carbon density and its influence
- Fire prediction
In the following list you can find suggested datasets for obtaining satellite image data, air quality, trees, and more. If there are more datasets relevant for the challenge, feel free to open a new issue and we'll add it to the list.
- EDI Ecosystem LiDAR: High resolution laser ranging of Earth’s forests and topography from the International Space Station:
- An account has to be created: https://urs.earthdata.nasa.gov/users/new
- Copernicus Conventional Data Access Hubs
- Waterbase – water quality from EU Open Data Portal
- NASA’s “Modern-Era Retrospective analysis for research and Applications V.2”
- Some data on renewable energy capacity from Renewables.ninja project
- Global Tree Search database of Tree Species
- The World Bank: World Carbon Stocks data
- Fire Information For Resource Management System (FIRMS) from NASA
- EDO - European Drought Observatory
- EFFIS: European Forest Fire Information System
- Copernicus Emergency Management Service, providing information for emergency response for different disasters.
- Luetjens, B., Liebenwein, L., & Kramer, K. (2019). Machine Learning-based Estimation of Forest Carbon Stocks to increase Transparency of Forest Preservation Efforts.
- Hou, X., Wang, B., Hu, W., Yin, L., & Wu, H. (2019). SolarNet: A Deep Learning Framework to Map Solar Power Plants in China from Satellite Imagery.
- Victor Schmidt, A. L. (2019). Visualizing the Consequences of Climate Change Using Cycle-Consistent Adversarial Networks.
- Ramachandra, V. (2019). Causal inference for climate change events from satellite image time series using computer vision and deep learning.
- Carrasco, J., Pais, C., Shen, Z.-J. M., & Weintraub, A. (2019). Adjusting Rate of Spread Factors through Derivative-Free Optimization: A New Methodology to Improve the Performance of Forest Fire Simulators.
- Subramanian, S. G., & Crowley, M. (2018). Using Spatial Reinforcement Learning to Build Forest Wildfire Dynamics Models from Satellite Images. Frontiers in ICT.
For a more general overview of opportunities for tackling climate change with machine learning, the paper "Tackling Climate Change with Machine Learning" by a group of renowned researchers (https://www.climatechange.ai/) presents several cases.
Know of good academic resources relevant for the topic of the Hackathon? Share it on a new issue.
Our partner AWS will provide credits for each team to work on AWS Sagemaker to build, train, and deploy ML models quickly.
What to do?
- Create an AWS before the hackathon weekend
- You will receive credit codes via your team channel on Slack on the first day of the hackathon
- Instructions for pre-work and setup of Sagemaker can be found here
- In the previous instructions, one of them is to raise the Sagemaker default limits for GPU instances. This takes 2 working days and it's important that you do this before the hackathon