AI for Earth 2 - Inland Floods Prediction
Combine hydrological sensor data with satellite images to predict flood propagation with machine learning.
Premium participation: Get 1-on-1 Mentoring & GuidanceTry a free mentoring session
Preventing human and economic losses by monitoring rivers and predicting flooding events
Challenge completed. Watch the recording of the AI for Earth 2 final results presentations to discover that the teams of AI for Inland Floods Prediction and AI for Forest Health came up with:
About the Challenge
The floods and landslides in South Africa last month have claimed hundreds of lives. In July last year, several European countries were affected by one the the most catastrophic natural disasters - floods causing a similar number of casualties. An important reason why these floods claim so many lives is their unpredictability, especially in the case of flash floods.
In this Challenge, we combine data from a network of hydrological sensors within a river estuary in France with open satellite data. Using time series data of water level and speed, hyperspectral cameras and satellites, we’ll try to predict the propagation of floods and help mitigate their impact across a river’s course.
Models developed in this Challenge can help warn of flash floods in any place with a network of sensors along a river. Early warning systems can help evacuate populations earlier and prevent deaths.
You’ll also explore models for mapping a river’s tributaries via satellite data, which can provide crucial information for flood propagation in countries where such maps don't exist.
Our challenge partner
GOAL: Develop models for flash floods prediction based on water level sensor and satellite data
Models developed in this Challenge can help warn of flash floods in any place where there is a network of sensors along a river. Early warning systems can help evacuate populations earlier and prevent deaths.
➡️ We will build ML models to estimate propagation speed of flood events when detected upstream by sensors that collect data on:
- Surface speed
- Water level
- Camera footage
➡️ We will be measuring percentage of surface water based on images (for small or dried up rivers)
- Combine with water height & flow speed
- See if there’s jams in the river, other anomalies
- Identify events that register on multiple sensors
➡️ We’ll augment the hydrological sensor data with satellite data
➡️ And we will develop a model to map rivers & its tributaries based on satellite data (e.g. in a directed graph)
- Use labeled data of countries that have the rivers mapped to create a model for countries who do not
Who are we looking for?
We are looking for data science & AI engineers with experience with time series analysis & forecasting. Having worked with satellite data, you would be a great addition to the team. Are you interested in entrepreneurship? We might have a special surprise for you too.
You will collaborate with a diverse team of up to 50 international collaborators in subteams. You can join as a contributor (~12 hours per week commitment for 10 weeks) or coach (2-4 hours per week, only for experienced ML professionals)
We’ll organize 2 masterclasses on relevant topics during the challenge to bring you up to speed.
Did you know
🌊 River flooding affects 21 million people worldwide
💰 River flooding affects $96 billion in GDP each year.
Floods are the most frequent type of natural disaster
🌡 After extreme temperatures, floods kill the most people of all natural disasters.
⛈ Flash floods are caused by rapid and excessive rainfall that raises water heights quickly, and rivers, streams, channels or roads may be overtaken.
🏔 River floods are caused when consistent rain or snow melt forces a river to exceed capacity.
🌀 Coastal floods are caused by storm surges associated with tropical cyclones and tsunami.
Application Deadline: 19 June 2022
Challenge Kick-off: 20 June 2022
Midterm Presentations: 01 August 2022
Final Presentations: 22 August 2022