With ample evidence of a future destined for warming temperatures, atmospheric shifts, and extreme weather, climate change is perhaps the most critical issue facing our planet. By calling on AI and machine learning tools for help, we’re in an excellent position to leverage the most powerful computing systems ever designed to help tackle the problems facing a globe already covered in sensors.
Several initiatives are in the works, but Microsoft’s “AI for Earth” has committed $50m in funding, so far awarding 35 grants in 10 countries for projects focused on using AI in climate conservation.
On land, this technology has been leveraged in projects monitoring nearly 30,000 acres of forests to better understand which species of trees survived and which were damaged in the wake of a hurricane.
This work requires advanced image recognition technology to classify tree species from aerial images and AI and is perfect for this task as the massive stores of image data need to be collected, analyzed, and extrapolated to draw meaningful conclusions on the condition of trees throughout the rest of the forest.
With the present pace of rising sea levels, change in circulation patterns, and water chemistry, the planet’s entire reef ecosystem could collapse in the next thirty years, threatening $375b to the global economy and putting 500m people at risk of food insecurity.
The 50 Reefs Project is a combined effort using 360-degree imaging technology and AI to gather and analyze the shape and texture of a reef to gauge health. These automated systems work in seconds, saving untold hours as compared to scientists manually reviewing the tens of thousands of images that make up the data set.
Apart from analyzing the observable results of climate change, Deep Learning and AI systems can also be harnessed for predictive powers and ascertain the likely impact from increasingly common extreme weather. With no practical limit to the variables at play in these models, the ability to run so many simulations is critical to provide more meaningful conclusions.
IBM’s Deep Thunder relies on machine learning tools for a global forecasting model while aligning with researchers to predict events like floods and mudslides. Traditionally, atmospheric models relied on predictions in which simulations were weighted equally, but AI helps to assign more probable weighting to improve predictive accuracy. The information gleaned helps determine when, where, and how severely a storm can be in advance, allowing repair crews and emergency services to mobilize proactively.
The number of natural disasters has quadrupled since 1970 and in another decade nearly 1.5 billion people will live in cities at high risk of natural disasters, so a collaborative effort consisting of a combination of AI techniques will be called upon to address the greatest dangers facing our ecosystem, the planet, and its populace.
See the full story here.