<p>Accurate quantification of methane (CH₄) emissions from rice paddies is essential for greenhouse gas mitigation. The DeNitrification-DeComposition (DNDC) model, widely used for biogeochemical simulation, often exhibits limited accuracy under diverse cultivation conditions due to structural constraints in representing soil temperature, root exudations, and crop growth stages. Its flexible structure, however, enables improvement via equation refinement and calibration. In this study, key CH₄-related process equations in DNDC were revised to enhance simulation accuracy. The plant growth index (PGI) was redefined from transplanting-to-maturity to transplanting-to-heading, and the dissolved organic carbon (DOC) equation was revised to incorporate observed root biomass carbon (Root C), soil organic carbon (SOC), and bulk density. Measured soil temperature replaced the default air-temperature assumption. Model calibration and validation used three years of field data from four transplanting dates (5/10, 5/25, 6/9, 6/24). The modified model increased R² from 0.661 to 0.788 and reduced RMSE from 1.308 to 1.091&#xa0;kg C ha^− 1 day^− 1; cumulative CH₄ emissions closely matched observations, whereas the original model reproduced on average only 74.2% of the observed cumulative emissions across treatments. Sensitivity analysis identified soil temperature as dominant, with CH₄ emissions changing + 39.6% per + 1&#xa0;°C and − 18.2% per − 1&#xa0;°C. Overall, PGI- and DOC-based refinements improved dynamic representation of CH₄ emissions and may provide a preliminary methodological basis for regional-scale validation; however, broader multi-site evaluation would be necessary before considering application within Tier 3 inventory frameworks.</p>