article · Environmental Science & Technology
Accurate forecasting of chlorophyll-a levels is critical for managing eutrophication in large lake ecosystems sustainably. This study investigated the ability of Kolmogorov-Arnold Networks to forecast time-series chlorophyll-a concentrations in the world's largest lakes. Using monthly remote-sensing data captured by Aqua-MODIS between September 2002 and April 2024, the network was evaluated alongside traditional machine learning algorithms and established neural network architectures. Over an unseen forecasting period from March to August 2024, the Kolmogorov-Arnold Network consistently demonstrated superior predictive power. It proved more capable of tracking overall trends, dynamic fluctuations, and peak chlorophyll-a levels than comparative models, including long short-term memory networks and support vector regression. Statistical rankings across multiple lake sites confirmed the model's robustness, showing that architectures built on the Kolmogorov-Arnold representation theorem handle complex nonlinearity and long-term dependencies in environmental data more effectively than standard universal approximation methods.
Chlorophyll-a concentrations indicate the onset of eutrophication, a process that severely degrades freshwater quality and threatens lake ecosystems. By delivering more reliable forecasts of these concentrations, this method helps water managers anticipate algal blooms and water quality changes. Advanced machine learning tools that accurately capture complex environmental cycles offer valuable support for the sustainable conservation of major freshwater bodies worldwide.
This work demonstrates an applied predictive framework that could assist environmental agencies, water utility managers, and lake conservation programmes in early-warning water quality systems. Based on evaluations using satellite records from 2002 to 2024, the model sits at an applied research stage. To reach operational use, it would need integration into routine environmental monitoring software and lake management decision-support platforms.
AI-generated from the published abstract. Always read the original work before citing.
Accurate prediction of chlorophyll-<i>a</i> (Chl-<i>a</i>) concentrations, a key indicator of eutrophication, is essential for the sustainable management of lake ecosystems. This study evaluated the performance of Kolmogorov-Arnold Networks (KANs) along with three neural network models (MLP-NN, LSTM, and GRU) and three traditional machine learning tools (RF, SVR, and GPR) for predicting time-series Chl-<i>a</i> concentrations in large lakes. Monthly remote-sensed Chl-<i>a</i> data derived from Aqua-MODIS spanning September 2002 to April 2024 were used. The models were evaluated based on their forecasting capabilities from March 2024 to August 2024. KAN consistently outperformed others in both test and forecast (unseen data) phases and demonstrated superior accuracy in capturing trends, dynamic fluctuations, and peak Chl-<i>a</i> concentrations. Statistical evaluation using ranking metrics and critical difference diagrams confirmed KAN's robust performance across diverse study sites, further emphasizing its predictive power. Our findings suggest that the KAN, which leverages the KA representation theorem, offers improved handling of nonlinearity and long-term dependencies in time-series Chl-<i>a</i> data, outperforming neural network models grounded in the universal approximation theorem and traditional machine learning algorithms.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1021/acs.est.4c11113
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
