Using AI to Predict and Prepare for Natural Disasters
With over 6 inches of rain in three hours on Ellicott City in Maryland, the Main Street looked like river rapids, cars were tossed around like rubber ducks, for the second time in two years, in 2018. According to the predictions of the National Weather Service, such storms are supposed to occur once in a thousand years. Floods are common in Ellicott City, where the Patapsco River meets its tributaries, but they are getting worse since a natural sponge forest has been replaced by a concrete jungle in the name of development. The city was selected for a pilot program that would use automated sensors to deliver flood warnings to its residents.
A high-resolution map showing the ground plan from New York to southern Virginia until it drains into the Chesapeake Bay was created to ease the process of predicting and preparing for future floods. The map was generated from aerial imagery, using AI, to show the precise placement of objects on the ground. This was an expensive and time-consuming process but turned out to be highly effective. Microsoft aided in the training of AI to identify objects by itself, and eventually, AI can be used to refresh and update the map.
Detailed maps that are up-to-date with land usage will present in the planning of drainage systems that accommodate more water. Eventually, the map is expected to offer live dashboards and automated alert systems that issue a warning before additional development overwhelms the storm water management capacity. Additionally, the map has been used by the Urban Forestry Administration in Washington DC to identify places where stagnant water accumulates and for planting more trees in those areas.
These techniques are being used across the globe. For example, Global Forest Watch offers weekly and monthly deforestation alerts using AI-powered algorithms developed by the University of Maryland. It is also being used in the oceans to map the world’s coral reefs. The amount of data gathered by the satellites employed to map the earth’s surface would have been indecipherable without the use of AI-based algorithms.
While mapping services have been used for conservation for a long time, data utilization has improved with the use of AI. Cloud computing and AI are providing deeper insights into planet health using a combination of sensor technologies and high-resolution imaging, resulting in an almost real-time readout of the planet’s vital signs. Microsoft is making its AI-mapping tool widely available, which is highly convenient for a society obsessed with the digital. AI on a planetary scale can be expected to become a commodity very soon.