Forecasting Volcanic Radiative Power (VPR) at Fuego Volcano Using Bayesian Regularized Neural Network

2025-03-25

AI Summary

To address the urgent need for volcanic hazard early warning at Fuego volcano, this paper proposes a short-term forecasting method for Volcanic Radiative Power (VPR) based on Bayesian Regularized Neural Networks (BRNN). This work represents the first systematic application of BRNN to VPR time-series modeling, leveraging prior probability distributions over network weights to mitigate overfitting under limited observational data and thereby enhance model generalizability. Compared with standard scaled conjugate gradient (SCG) and Levenberg–Marquardt (LM) optimization algorithms, the BRNN approach achieves significantly improved prediction accuracy—reducing mean squared error to 1.77×10¹⁶ W² and attaining an R² score of 0.50—demonstrating its effectiveness in characterizing dynamic thermal activity. The proposed method establishes a novel paradigm for quantitative, thermal-infrared remote sensing–driven volcanic activity forecasting, thereby strengthening volcanic hazard risk assessment and real-time response capabilities.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Forecasting volcanic activity is critical for hazard assessment and risk mitigation. Volcanic Radiative Power (VPR), derived from thermal remote sensing data, serves as an essential indicator of volcanic activity. In this study, we employ Bayesian Regularized Neural Networks (BRNN) to predict future VPR values based on historical data from Fuego Volcano, comparing its performance against Scaled Conjugate Gradient (SCG) and Levenberg-Marquardt (LM) models. The results indicate that BRNN outperforms SCG and LM, achieving the lowest mean squared error (1.77E+16) and the highest R-squared value (0.50), demonstrating its superior ability to capture VPR variability while minimizing overfitting. Despite these promising results, challenges remain in improving the model's predictive accuracy. Future research should focus on integrating additional geophysical parameters, such as seismic and gas emission data, to enhance forecasting precision. The findings highlight the potential of machine learning models, particularly BRNN, in advancing volcanic activity forecasting, contributing to more effective early warning systems for volcanic hazards.

Problem

• Predicting volcanic radiative power using neural networks
• Comparing Bayesian Regularized Neural Network with other models
• Improving volcanic hazard forecasting accuracy with machine learning

Innovation

• Bayesian Regularized Neural Networks for VPR prediction
• Comparison with SCG and LM models
• Integration of geophysical parameters for accuracy