Abstract: The complex conditions within closed fire zones make gas monitoring challenging,resulting in varied sampling results. To address this,our study evaluates the reliability and priority of different monitoring indicators. We used temperatureprogrammed experiments to optimize early warning indicators and developed a hierarchical early warning model. The grey correlation method was applied to assess the relationship between selected indicators and parameters within the closed area, ensuring their suitability. The model was then validated using field data. Results identified four key characteristic temperatures for Tingnan coal:30 ℃, 70 ℃, 100 ℃, and 200 ℃. The volume fraction of CO,as well as the initial detection temperatures of C₂H₄ and C₂H₂,were chosen as primary early warning indicators. Additionally,the commonly used composite index R_C,O was confirmed as an effective warning indicator. Strong correlations were found between the CO volume fraction,R_C,O,O₂ volume fraction,and the pressure difference (ΔH) between the inside and outside of the closed area. The correlation was highest with O₂ volume fraction (0. 801 3 for CO and 0. 785 8 for R_C,O), followed by ΔH (0. 696 8 for CO and 0. 672 1 for R_C,O). Continuous monitoring showed that R_C,O exceeded the minimum threshold for a blue warning on the second day and continued to rise, successfully triggering the model’ s blue warning alert. These findings demonstrate that the hierarchical early warning model is effective and can be reliably applied in practice.