11 Jan Artificial Intelligence and Environmental Disasters: Flood Hub and Nestore
Over the past two decades, 90% of disasters on Earth have been caused by extreme weather events, such as floods, storms, hurricanes and heat waves.
Although artificial intelligence cannot yet fully defend the planet from these catastrophes, new technologies can still help to cope with the natural disasters that increasingly affect the Earth.
In fact, Artificial intelligence can provide timely and detailed methods to limit its effects and consequences.
AI allows to “regulate” a large amount of data and analyze patterns of several natural events: record rainfall, detect seismic events, predict the possibility of fires, floods, storms, and so on.
Google Project: Flood Hub
A good example of AI that aims to prevent environmental disasters is Flood Hub.
In 2018 Google developed Flood Hub, a watercourse monitoring system whose aim is predicting floods and inundations to limit any consequences.
This forecasting system consists of four subsystems:
– data validation
– phase forecasting
– flood modelling
– distribution of warnings.
The monitoring system collects data on flood events, then uses machine learning and artificial intelligence systems to develop estimates and forecasts of possible future floods.
This forecasting system is active in India and Bangladesh, and Google is working to expand these life-saving alarms to countries in South Asia and South America as well.
During the 2021 monsoon season, the alarm system covered flood-prone areas around rivers, with a total area of 287,000 km2, housing more than 350 million people. More than 100 million flood alerts have been sent to affected populations, authorities and emergency organizations.
Among the goals of the Flood Hub system, there is the global outreach, as well as the improvement of prediction and accuracy.
Artificial intelligence for the study of seismic sequences
Another relevant example concerns earthquakes: earthquakes are one of the most unpredictable and dangerous natural phenomena.
To date, it is impossible to determine with absolute certainty the place and precise moment of a seismic event; however, methods of preparation and mitigation of the consequences exist.
AI systems have been found to be crucial in analyzing and detecting seismic tremors. The National Institute of Oceanography and Experimental Geophysics and the National Institute of Geophysics and Volcanology have defined an algorithm capable of making an evaluation of the probability of aftershocks.
That algorithm, called Nestore, applies Machine Learning to calculate the probability that one earthquake tremor will be followed by others of different intensities.
The algorithms work by learning and require a large amount of data to be “trained.”
In this research, scholars have focused on data on earthquakes that occurred in California (area with high seismic activity) and thanks to this algorithm it was possible to predict, even well in advance, a large number of quakes, with a false alarm rate of less than 20%.
In conclusion, although humans cannot totally prevent environmental catastrophes, thanks to artificial intelligence we are finding ways to limit the damage and deal with these events as efficiently and safely as possible.