As the use of groundwater for drinking water (bottled water) increases, the government is implementing an artificial intelligence (AI)-based real-time monitoring system. This is due to findings that 62% of 2,000 surveyed groundwater wells failed to meet drinking water standards, and concerns have been raised about the inadequacies in the response system for fluctuations in groundwater levels and quality.
On the 16th, according to related departments and the industry, the National Institute of Environmental Research under the Ministry of Environment plans to establish an 'AI-based real-time monitoring system for drinking water' by the end of this year.
First, after investigating domestic and international cases of AI and groundwater management, a sensor-IoT-based automatic measurement network will be established, followed by pilot development of water quality and quantity prediction algorithms using machine learning.
The Institute of Science plans to install sensors measuring 12 items, including water level, electrical conductivity (EC), pH, and temperature, 24 hours a day at one pilot groundwater observation well, and create a model to detect anomalies early by integrating field and laboratory data. The sensor data will be automatically calibrated and sent to the cloud, where AI will learn patterns and provide warnings of water quality deterioration or quantity depletion in advance.
This project aligns with the government's national AI strategy and digital new deal, which advocates for 'the expansion of 5G and AI across all industries.' The Ministry of Environment plans to establish measures for the mandatory automatic measurement of drinking water intake points, data standardization, and real-time public platform creation based on the pilot results.
The Board of Audit and Inspection noted in an audit in 2021 that 'the installation and data management of automatic measurement devices for drinking water quality are inadequate,' calling for institutional improvements. Around 330,000 residents in areas without public water supply use groundwater for their daily needs, but the compliance rate for the legally mandated water quality testing every two years is only 3.8% at best.
According to an investigation of groundwater quality in areas where public water supply is not available, the average rate of non-compliance with water quality standards for surveyed wells from 2015 to 2019 was 55.7%, and in a 2023 survey by the National Institute of Environmental Research, 61.9% were found unsuitable for drinking. Among well owners, 75% indicated that they did not undergo testing due to 'testing expenses being burdensome' or 'not knowing the procedures.'
In particular, groundwater quality has sharply changed with seasonal and rainfall variations, making manual testing insufficient. Therefore, the Institute of Science explained that if the AI prediction model is commercialized, it could allow for preemptive decisions on whether to halt water extraction or invest in additional purification facilities.
A representative from the National Institute of Environmental Research stated, 'By installing sensors in groundwater wells to automatically measure data such as water level and quality, we will receive that data via LTE networks and use AI algorithms such as random forest and LSTM (long short-term memory) to predict changes in water quality or reductions in quantity.' They added, 'Previously, it took time to collect samples and analyze them, but with AI, we can detect early signs of water quality changes and quickly recognize and respond to potential contamination risks.'
This representative further added, 'While this pilot project is targeted at specific wells that use groundwater, plans will be made to consider expanding its application to the entire bottled water industry based on the results.'