Greenhouse Gas Emissions from Large-Scale Dairy Farms: Interannual Dynamics, Farm-Level Heterogeneity, and Key Determinants

The transition to low-emission (LE) dairy production is essential for mitigating agricultural greenhouse gas (GHG) emissions. However, the interannual variability and principal drivers of farm-level GHG footprints across intensive dairy farms remain insufficiently characterized, constraining targeted mitigation strategies. This study aimed to (i) quantify interannual variation in on-farm GHG emission intensity, (ii) characterize management contrasts between high- and low-emission farms, (iii) clarify the underlying mechanisms, and (iv) identify optimization pathways. A supplementary sensitivity analysis encompassing upstream emissions associated with feed production was performed using an expanded system boundary to assess the effects of system boundary selection on the estimated GHG emissions. Using 3-year (2021–2023) field data from 22 intensive dairy farms, a standardized farm-level GHG inventory was used to quantify operational GHG emission intensity per unit of fat- and protein-corrected milk (FPCM). The production-weighted mean on-farm operational GHG intensity was 0.86 kg CO2-eq kg−1 FPCM, with enteric methane constituting the dominant emission source. Incorporating upstream emissions associated with feed production contributed an additional 0.427 kg CO2-eq kg−1 FPCM, yielding a total emission intensity of 1.287 kg CO2-eq kg−1 FPCM under the expanded system boundary, representing a 49.7% increase relative to the operational inventory. Feed upstream emissions accounted for 33.2% of the total emissions estimated within the expanded system boundary. Both herd size and energy consumption exhibited increasing trends throughout the study period, whereas the proportion of lactating cows relating to the total herd declined from 46.01% in 2021 to 44.32% in 2023, suggesting that the expansion of the overall herd exceeded the corresponding growth in the lactating cow population. Under the operational system boundary, high-emission (HE) farms exhibited relatively greater contributions to their GHG emission profiles from enteric methane production, electricity consumption, and manure management activities. Conversely, LE farms attained improved performance through higher proportions of lactating cows, greater feed efficiency, and manure anaerobic digestion. Correlation analysis linked herd size, energy consumption, and dry matter intake positively to on-farm GHG emissions. Dietary starch was associated with manure N2O emissions, while feed efficiency was associated with higher energy-related GHG emissions. Overall, herd composition, feeding management practices, energy utilization, and manure management strategies were identified as principal determinants of on-farm GHG emission performance. Furthermore, the sensitivity analysis employing an expanded system boundary demonstrated that incorporating upstream emissions associated with feed production substantially elevates the overall GHG emission intensity expressed per kg FPCM. This study helps address the lack of multi-year farm-level evidence for China’s dairy sector and proposes integrated strategies—herd structure adjustment, precision feeding, clean energy exploitation, and manure management—alongside sustained long-term monitoring, given the pronounced interannual variability. These findings highlight the roles of farm-level GHG inventories in identifying targeted on-farm emission mitigation opportunities and expanded system boundary assessments in quantifying the overall GHG burden associated with the entire dairy production supply chain.