Rice cultivation is one of the major anthropogenic methane sources in China and globally. However, accurately quantifying regional rice methane emissions is often challenging due to highly heterogeneous emission fluxes and limited measurement data. This study attempts to address this issue by quantifying regional methane emissions from rice cultivation with a high-resolution inversion of satellite methane observations from the Tropospheric Monitoring Instrument (TROPOMI). We apply the method to the largest rice-producing province (Heilongjiang) in China for 2021. Our satellite-based estimation finds a rice methane emission of 0.85 (0.69 – 1.03) Tg a⁻¹ from the province, or an average emission factor of 22.0 (17.8 – 26.6) g m⁻² a⁻¹ when normalized by rice paddy areas. The satellite-based analysis reveals a 2 to 4 times lower bias in widely used global and national inventories, which lack up-to-date regional information. The inversion reduces the uncertainty of regional rice emissions by 73% relative to bottom-up estimates based on field flux measurements. The satellite inversion also shows that the highest rice methane emissions occur in June during the tillering stage of rice, decreasing towards ripening, indicating that the predominant water management practice in the region involves drainage and intermittent flooding after initial flooding. Process-based modeling further suggests that this practice can lead to a reduction of methane emissions by more than 50% compared to continuous flooding rice paddies and natural wetlands.

水稻是中国乃至全球最主要的人为甲烷排放源之一，然而由于排放通量的高度时空异质性且观测数据稀缺，区域水稻甲烷排放的准确量化面临极大挑战。本研究基于TROPOMI卫星甲烷观测数据，构建高分辨率反演模型对水稻种植区域的甲烷排放进行量化，从而有效解决这一难题。我们将该方法应用到2021年中国最大水稻生产省份——黑龙江，发现该省的水稻甲烷排放量为0.85 (0.69–1.03) Tg a⁻¹，经种植面积标准化后的平均排放因子为22.0 (17.8–26.6) g m⁻² a⁻¹。基于卫星观测的反演分析揭示，广泛使用的全球和国家排放清单存在2–4倍低估，这与清单缺乏最新的区域信息相关；相较于基于稻田实地通量观测作为排放因子的“自下而上”估算，反演方法可将区域水稻排放的不确定性降低73%。卫星反演结果还显示，水稻甲烷排放的峰值出现在6月水稻分蘖期，随后排放逐渐减少，印证该地区主要的水管理方法为初期淹水后实施排水晒田和间歇灌溉。基于过程模型结果进一步验证，与连续灌溉的稻田和自然湿地相比，该管理方式可减少50%以上的甲烷排放。

甲烷排放,水稻,反演模型,卫星观测,中国东北