In order to prevent the world's coral reefs from disappearing forever, there have been many coral reef restoration projects. But how can we more effectively assess the effectiveness of these programs? The answer may lie in listening to coral reefs Usually, to measure the progress of these projects, snorkeling or scuba diving scientists will visually inspect the coral reefs. They will not only check the growth of new corals, but also observe the number and type of fish and other animals. Generally speaking, the healthier the coral, the greater the biodiversity of the coral reef.
However, many difficulties will be encountered in the actual operation. Swimming across the whole coral reef may be very laborious and time-consuming. In addition, many creatures may not be seen because they hide under corals or only move at night.
In search of an alternative, scientists from the University of Exeter in the United Kingdom began underwater recording on some coral reefs in Indonesia. These recordings include the sounds of marine life, many of which are inaudible to the unaided human ear.
Using machine learning based algorithms, the researchers then created an artificial intelligence computer program that can distinguish between sound features recorded on healthy coral reefs and those recorded on poor coral reefs. When the program was subsequently used to evaluate new recordings, it proved to be 92% accurate in determining whether each recording came from a healthy or unhealthy coral reef.
Dr Tim Lamont, co-author of the study, said: "sound recorders and artificial intelligence can be used around the world to monitor the health of coral reefs and find out whether attempts to protect and restore them are effective. In many cases, it is easier and cheaper to deploy hydrophones on coral reefs and leave them there than expert divers who repeatedly visit coral reefs to investigate them - especially in remote areas.".