Hybrid time series forecasting of poultry manure–based biogas potential: a global renewable energy and greenhouse gas mitigation perspective

The rapidly expanding global poultry industry, which processes over 50 billion birds annually, generates vast quantities of manure that pose both a pressing environmental challenge and a significant, underutilised source of renewable energy. While anaerobic digestion offers a scalable pathway to convert this waste into biogas – mitigating greenhouse gas emissions and advancing circular bioeconomy goals – effective implementation requires accurate long-term forecasting of biogas potential (BP). Existing studies predominantly offer static or snapshot assessments, which are insufficient for long-term planning, investment decisions, and infrastructure development. To bridge this gap, this study introduces a novel hybrid time-series forecasting framework that synergistically combines classical statistical models (ARIMA, SARIMA) with advanced, genetically optimised machine-learning algorithms (GA-LSTM, GA-XGBoost). The central hypothesis – that hybrid approaches significantly outperform conventional methods in capturing complex, nonlinear temporal dynamics – is tested using a representative case study. The framework estimates a 2024 electricity-equivalent BP of 3,364 GWh and projects a rise to 3,427 GWh by 2035. The GA-XGBoost model reduced MAPE by approximately 83% compared to the SARIMA model. This work provides a transferable, bias-reducing forecasting tool that enables policy-grade planning, investment timing, and risk-aware deployment of manure-based biogas systems across livestock-intensive economies.