Abstract: To clarify the carbon sinks and their uncertainty components in the temperate secondary forest ecosystem (mosaics of natural secondary forests and plantations). We selected three typical forest stands including secondary mixed broadleaved forest (T1-MBF), secondary Mongolian oak forest (T2-MOF), and larch plantation (T3-LPF). The net primary productivity (NPP) and soil heterotrophic respiration (R_h) were monitored for 4 years (2020–2023) by both inventory and chamber methods, and net ecosystem productivity (NEP, carbon sink; NEP = NPP − R_h) and its uncertainty were further calculated for three forest stands. The results showed that the NEP were 1.99 ± 1.78, 1.87 ± 2.06 and 2.68 ± 1.42 t ha⁻¹·a⁻¹ for T1-MBF, T2-MOF, and T3-LPF, respectively, with high relative uncertainties (89.33%, 109.98% and 52.93%, accordingly). Specifically, fine root NPP dominated uncertainty (58.05–78.63%), followed by R_h (2.20–30.45%) and leaf (2.47–9.58%), while stable pools (e.g., stem, coarse root) contributed minimally. Importantly, to improve reliability, we developed a revised method to estimate low-uncertainty carbon sink, which focuses on low-uncertainty carbon pools, including stem, coarse root, and soil carbon. Low-uncertainty carbon sinks reached 1.82 ± 0.21, 1.79 ± 0.34, and 2.56 ± 0.54 t ha⁻¹·a⁻¹ for T1-MBF, T2-MOF, and T3-LPF, respectively. Notably, relative uncertainties dropped sharply to 11.52%, 18.77%, and 21.15%. This study provides a robust framework for quantifying low-uncertainty carbon sinks in the temperate secondary forest ecosystem by integrating resilient carbon pools, significantly improving estimation reliability while reducing uncertainty by 60.0–87.1% compared to conventional approaches.

Key words: Temperate forest, Inventory method, Carbon sink, Uncertainty