Problem / Objective:

Flooding is the most common natural disaster on the planet, affecting hundreds of millions of people and causing between 6,000 and 18,000 fatalities every year – of which 20 percent are in India. Reliable early warning systems have been shown to prevent a significant fraction of fatalities and economic damage, but many people don’t have access to those types of warning systems.

The Google Flood Forecasting Initiative aims to use Google’s infrastructure and machine learning expertise for providing accurate real-time flood forecasting information and alerts to those in affected regions. This is made possible through AI and physics-based modelling which create accurate and scalable inundation models in real-world settings. With reliable on-ground data obtained from governmental agencies, Google's river flood forecasting models can more accurately predict not only when and where a flood might occur, but the severity of the event as well.

Through an innovative approach for inundation modelling, the initiative aims to provides unprecedented lead time, accuracy and clarity in flood forecasting.

Solution / Approach:

To simulate the water behaviour across a floodplain, inundation modelling uses as inputs measurement or forecast of river water levels and high-resolution elevation maps. Real-time river measurements and forecasts are obtained for this initiative through Google's collaboration with the Central Water Commission (CWC), the technical organisation of India in the field of water resources.

Google has devised a new approach for inundation modelling, called a morphological inundation model, which combines physics-based modelling with ML to create more accurate and scalable inundation models in real-world settings. In comparison to classical physics-based models, this morphological model improves accuracy by 3%, which can significantly improve forecasts for large areas, while also allowing for much more rapid model development by reducing the need for manual modeling and correction.

Additionally, their alert-targeting model allows identifying areas at risk of flooding at unprecedented scale using end-to-end machine learning models and data that is publicly available globally. They have also developed HydroNets – a specialised deep neural network architecture built specifically for water levels forecasting – which allows the utilisation of some exciting recent advances in ML-based hydrology in a real-world operational setting. 

Impact / Implementation:

First piloted in the Patna region of Bihar in 2018, Google’s flood forecasting initiative has been extended to the whole of India by 2020, covering 200 million people across more than 250,000 sq km. Google technology is being used to improve the targeting of every alert the government sends; around 30mn notifications have been sent to people in flood affected areas, to date.

The alerts sent out include three tiers of risk: some flood risk, greater flood risk, and greatest flood risk. For better accessibility, the information is provided in different formats so that people can both read their alerts and see them presented visually; they have added support for Hindi, Bengali and seven other local languages, too. 

Google.org has started a collaboration with the International Federation of Red Cross and Red Crescent Societies (IFRC) to build local networks that can get disaster alert information to people who wouldn’t otherwise receive smartphone alerts directly. A partner notification infrastructure has been established to provide these forecasts for the CWC and other organisational partners that can use it to prepare for disaster management and relief efforts.

Google’s initiative is providing people with information about flood depth: when and how much flood waters are likely to rise. And in areas where it is possible to produce depth maps throughout the floodplain, they are sharing information about depth in the user’s village or area. The recent improvement to the forecasting model has allowed them to double the lead time of many of their alerts – providing more notice to governments and giving tens of millions of people an extra day or so to prepare.

Image from Wikimedia Commons