Modelling CH₄ emission from rice ecosystem: A comparison between existing empirical models

Rice is a staple food for more than three billion people and accounts for up to 11% of the global methane (CH₄) emissions from anthropogenic sources. With increasing populations, particularly in less developed countries where rice is a major cereal crop, production continues to increase to meet demand. Implementing site-specific mitigation measures to reduce greenhouse gas emissions from rice is important to minimise climate change. Measuring greenhouse gases is costly and time-consuming; therefore, many farmers, supply chains, and scientists rely on greenhouse gas accounting tools or internationally acceptable methodologies (e.g., Intergovernmental Panel on Climate Change) to estimate emissions and explore mitigation options. In this paper, existing empirical models that are widely used have been evaluated against measured CH₄ emission data. CH₄ emission data and management information were collected from 70 peer-reviewed scientific papers. Model input variables such as soil organic carbon (SOC), pH, water management during crop season and pre-season, and organic amendment application were collected and used for estimation of CH₄ emission. The performance of the models was evaluated by comparing the predicted emission values against measured emissions with the result showing that the models capture the impact of different management on emissions, but either under- or overestimate the emission value, and therefore are unable to capture the magnitude of emissions. Estimated emission values are much lower than observed for most of the rice-producing countries, with R correlation coefficient values varying from −0.49 to 0.87 across the models. In conclusion, current models are adequate for predicting emission trends and the directional effects of management, but are not adequate for estimating the magnitude of emissions. The existing models do not consider key site-specific variables such as soil texture, planting method, cultivar type, or growing season, which all influence emissions, and thus, the models lack sensitivity to key site variables to reliably predict emissions.