Marine Chlorophyll Prediction and Driver Analysis based on LSTM-RF Hybrid Models

2025-08-07

AI Summary

To address the need for harmful algal bloom (HAB) early warning and marine ecosystem health monitoring, this study proposes an LSTM-Random Forest (LSTM-RF) hybrid model to improve high-frequency time-series forecasting of marine chlorophyll-a concentration. Methodologically, the model synergistically integrates LSTM’s capability in capturing long-range temporal dependencies with RF’s interpretability and robustness in modeling nonlinear, multi-source ecological drivers (e.g., temperature, salinity, dissolved oxygen), enhanced by input standardization and sliding-window feature engineering. Experimental results demonstrate superior performance on the test set: R² = 0.539, MSE = 0.0058, MAE = 0.057—significantly outperforming standalone LSTM or RF baselines. The key contribution lies in the first application of such a hybrid architecture to chlorophyll-a dynamics prediction, achieving enhanced temporal generalization without sacrificing model interpretability. This advances high-fidelity monitoring of marine carbon cycling and supports rapid, evidence-based ecological risk response.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Humans and AI: Human-in-the-loop Machine Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Marine chlorophyll concentration is an important indicator of ecosystem health and carbon cycle strength, and its accurate prediction is crucial for red tide warning and ecological response. In this paper, we propose a LSTM-RF hybrid model that combines the advantages of LSTM and RF, which solves the deficiencies of a single model in time-series modelling and nonlinear feature portrayal. Trained with multi-source ocean data(temperature, salinity, dissolved oxygen, etc.), the experimental results show that the LSTM-RF model has an R^2 of 0.5386, an MSE of 0.005806, and an MAE of 0.057147 on the test set, which is significantly better than using LSTM (R^2 = 0.0208) and RF (R^2 =0.4934) alone , respectively. The standardised treatment and sliding window approach improved the prediction accuracy of the model and provided an innovative solution for high-frequency prediction of marine ecological variables.

Problem

• Predict marine chlorophyll concentration accurately for ecosystem monitoring
• Overcome single model limitations in time-series and nonlinear modeling
• Enhance prediction accuracy using multi-source ocean data and hybrid models

Innovation

• LSTM-RF hybrid model for time-series and nonlinear features
• Multi-source ocean data training improves accuracy
• Standardized treatment and sliding window enhance prediction