Opens in a new tab

Forecasting Chlorophyll-a in Lakes

Lanxi Li, Melanie Po-Leen Ooi, Guilherme Weigert Cassales, Whitney Woelmer, Sepideh Heydari, and Deniz Özkundakci. 2026. A Systematic Literature Review of Machine Learning Methods to Forecast Chlorophyll-a in Lakes. IEEE Access.

Go to article

Abstract

Accurately forecasting chlorophyll-a (Chl-a) concentrations has been an essential task for monitoring and managing eutrophication in lake ecosystems. Nutrient enrichment and changes in climate can accelerate algal growth, often measured via the concentration of Chl-a (component of algae). Rapid algae growth could also lead to harmful algal blooms (HABs), posing risks to a major concern for ecological and human health. In recent years, machine learning (ML) approaches, particularly deep learning (DL), have been increasingly used to improve Chl-a prediction. However, Chl-a data acquisition methods and data standards are often inconsistently reported across studies, limiting the reproducibility of machine learning and deep learning approaches for forecasting Chl-a. This paper presents a systematic literature review and mapping study of 43 primary studies selected from 3,338 screened papers to analyze current ML and DL approaches for Chl-a forecasting in lakes. The review identified four major research gaps: inconsistent data acquisition and preprocessing protocols, lack of methodological standardization, unclear trade-offs between model complexity and generalizability, and limited application of explainable AI (XAI) for operational use. This paper provides a roadmap for future research. Key recommendations include standardizing data acquisition and reporting protocols, establishing benchmark datasets to improve reproducibility, and promoting more spatio-temporal and seasonal studies to enable cross-study comparisons. The study also highlights the importance of integrating XAI and co-design with ecosystem managers to improve the robustness, transparency, and real-world applicability of Chl-a forecasting models.