A Systematic Review of Methane Emission Estimation Challenges in Open and Unmanaged Landfills

Landfills generate substantive greenhouse gas (GHG) emissions. Methane is one of the GHGs generated by landfills. The landfill emissions are more pronounced in developing countries where most of the waste disposal sites are open and unmanaged. Unmanaged landfills are non-engineered sites that are poorly managed. This review critically examines the suitability of existing models and algorithms in estimating methane emissions from open and unmanaged landfills. The study reviewed process-based approaches, artificial intelligence (AI)/Machine Learning (ML) methods, and traditional models such as IPCC, LandGEM, and Scholl Canyon. The review revealed that traditional models are based on generalized assumptions that often does not apply to all sites, leading to uncertainties in methane emission estimations. Secondly, improved traditional models such as Capturing Landfill Emissions for Energy Needs (CLEEN) and Solid Waste Emissions Estimation Tool (SWEET) shows reduced estimation errors but have not been validated across regions and sites. Also, process-based approaches like the California Landfill Methane Inventory Model (CALMIM) showed improved estimation of methane emissions but require extensive site-specific data. However, data scarcity in many open and unmanaged landfills inhibits their adaptability. Further, the utilization of machine learning algorithms demonstrated high predictive performance. They have ability to capture nonlinear relationships between methane generation and site conditions. However, they require extensive historical data for model training. The review noted that hybrid physics-AI/ML approaches offer better results than standalone ML because they enhance interpretability of the results. The review concluded that process-based models will be ideal for sites with extensive site-specific data while AI/ML approaches will be ideal for sites with extensive historic emission data. Finally, when methane oxidation and lag time are included in a first order model, it could enhance methane estimation for open and unregulated landfills with limited data. The model is also considered less costly and adaptable when validated using site specific data. The findings provide a basis for improving methane emission estimation from open and unmanaged landfills.