Classification Machine Learning to Detect de Facto Reuse and Cyanobacteria at a Drinking Water Intake

Adapting to environmental challenges that impact source water quality is critical for utilities in providing safe and reliable drinking water for their customers. Two of these emerging concerns are the presence of harmful algal blooms (HABs) and de facto water reuse (DFR). As climate change raises water temperatures and lowers water levels, both the frequency and intensity of HAB and DFR events are expected to increase.

Because of this, early detection and response are more important than ever. A recent study co-authored by a team of experts including Carollo’s national PFAS lead, Kyle A. Thompson, Ph.D., PE. presents groundbreaking research on the use of machine learning in HABs and DFR early detection challenges. The study, titled “Classification Machine Learning to Detect de Facto Reuse and Cyanobacteria at a Drinking Water Intake,” was published in Science of the Total Environment.

Using supervised machine learning (SML), the researchers created a real-time monitoring system at the intake level that could alert water utilities to early warning signs. This multi-class SML allowed the researchers to differentiate between HABs and DFR events, incorporating online instrumentation that alerted utilities to potential threats, even at low levels of contamination.

In detecting low concentrations of DFR and HABs at the intake level, the SML technology allows utilities to respond swiftly to potential threats, helping to keep drinking water safe and protect human health. This proactive approach is particularly beneficial for utilities with intakes in rivers or smaller reservoirs, where spikes in DFR are more likely to occur.

As we navigate the challenges posed by climate change and increasing demands on water resources, innovative solutions like those presented in this study offer hope for more resilient and reliable water systems. For a deeper dive into this groundbreaking research, read the full article.