Estimating Operational Costs of Activated Carbon for Water Treatment Plants by Predicting the Rise of Harmful Algal Blooms Under Climate Change in Korea Using Machine Learning

TL;DR

This study uses machine learning to predict how climate change will increase harmful algal blooms in South Korea and raise water treatment costs.

Contribution

A novel method combining machine learning and cost analysis to estimate future water treatment expenses due to climate-driven algal blooms.

Findings

Cyanobacteria density is projected to increase 3–5 times by 2100 under the SSP5–8.5 climate scenario.

Water treatment costs could triple, reaching $22.1/month/household by 2100 due to higher PAC usage.

Proactive measures are needed to address rising health risks and treatment burdens from algal blooms.

Abstract

The escalating frequency of harmful cyanobacterial blooms (HCBs), driven by climate change and eutrophication, poses risks to ecosystems, water resources, and public health. Given South Korea's heavy reliance on surface waters, increasingly affected by HCBs producing microcystins and taste and odor compounds (geosmin and 2‐methylisoborneol), this study used machine learning to predict cyanobacterial proliferation by 2100 under climate scenarios. It also estimates increases in treatment costs, assuming water treatment plants (WTPs) respond to increased bloom intensity solely by modifying their usage of powdered activated carbon (PAC). A random forest (RF) model trained on 28 years of Nakdong River data projected HCB occurrences under Shared Socioeconomic Pathway 5–8.5. The RF indicated significant increases in HCB magnitude and variability (cyanobacteria densities from 1.6 × 104 to 6.3 × 104 cells/mL; coefficient of variation from 1.60 to 1.77), corresponding to a 6.7°C increase in mean annual air temperature. Analysis of WTP operational data and prior studies revealed a correlation between PAC use and HCB events, suggesting the increase in HCBs necessitates significantly higher PAC doses to treat projected secondary metabolites, particularly microcystins. Under the worst‐case scenario, the projected cost burden for water treatment could triple from current levels, potentially reaching $22.1/month/household by 2100, supporting proactive implementation of advanced treatment facilities in high‐risk regions. These findings underscore the need for enhanced preparedness to address more complex HCB patterns under climate change, ensuring water safety, economic stability, and human health. Additionally, this study provides a methodological blueprint for other countries facing similar climatic and environmental challenges.

Investigating future operational costs in water treatment (WT) facilities.Random forest projected 3–5 times more abundant cyanobacteria by 2100 (SSP5–8.5).Increase in microcystins demanded substantially more powdered activated carbon.WT costs to rise from $3–12 to $6–22/household/month under SSP5–8.5.Proactive measures are necessary to mitigate health risks and WT burdens.

Investigating future operational costs in water treatment (WT) facilities.

Random forest projected 3–5 times more abundant cyanobacteria by 2100 (SSP5–8.5).

Increase in microcystins demanded substantially more powdered activated carbon.

WT costs to rise from $3–12 to $6–22/household/month under SSP5–8.5.

Proactive measures are necessary to mitigate health risks and WT burdens.

This study investigated future operational costs in water treatment (WT) plants. Random forest modeling projected 3–5 times more abundant cyanobacteria by 2100 due to climate change. Increased microcystins demanded substantially more activated carbon and, as a result, WT costs could rise to $6–22/household/month. Proactive measures are necessary to mitigate health risks and WT burdens.

Figures