For illegal pollution monitoring AI could help farms from space to look

For illegal pollution monitoring AI could help farms from space to look

 

Comprising around 40 per cent of the country’s livestock by US Concentrated animal feeding operations (CAFOs) and these intensive farms often contain about 2500+ pigs or 120,000+ chickens per facility, generating around 335 million tonnes of waste per year. The vast amount of waste usually goes untreated before it makes its way into the waterways, also farms produce pollution via the (bovine) livestock. The US government has been making use of the country’s recent developments in Artificial Intelligence (AI) to detect farms that may be illegally polluting the waterways. Also, The approach has been trialled across Europe to monitor and inspect farmland.

Despite the US’ Clean Water Act stating that those who dump waste into a waterway require a federal permit, it is estimated that nearly 60 per cent of CAFOs do not have one. Not only does that make this sort of dumping illegal, but it also greatly harms the environment and all the species that live in it.

Of course, the US realised that this was a problem and nominated a team of experts to scan and track which farms were illegally dumping waste into waterways, using satellite imagery. However, two members of this team, Daniel Ho and Cassandra Handan-Nader of Stanford University, were able to train a neural network to scan publicly available satellite images for CAFOs and identify certain shapes such as rectangular barns or outdoor manure pits. Amazingly, while a human would take six weeks to identify the farms around the country, the AI machine took just two days and found 15 per cent more farms. Ho believes that these computer algorithms are the first step towards cleansing the world of waste-stained waterways.

Similarly to the US, computer algorithms are monitoring the health of vineyards in Tuscany, Italy, and observing whether farmers which received governments subsidies in Lithuania and Estonia are maintaining the good condition of their lands, reducing the need for physical visits from inspectors.It is estimated that using these algorithms may save around €500,000 every year in manual inspection costs and false payments made to non- compliant farmers.