Abstract: Urban sewage pipelines are significant sources of greenhouse gas (GHG) emissions. However, the mechanisms underlying GHG production and release at the sediment-water interface, particularly in real-world sewage networks characterized by highly variable water quality, remain under-researched. This study aimed to clarify the dynamic patterns and driving mechanisms of these processes. From November to December 2024, sediment and overlying water samples were collected from representative sewage pipelines in Shenzhen. To systematically investigate the physicochemical and microbial processes at the sediment-water interface, a comprehensive approach was adopted, incorporating water quality analysis, headspace gas chromatography measurements of dissolved GHGs (CH₄, CO₂, N₂O), 16S rRNA gene amplicon sequencing, and quantitative PCR. The results indicated that the dissolved CH₄ concentration did not respond immediately to changes in water quality, but exhibited significant metabolic lag effects, demonstrating that CH₄ production and release at the sediment-water interface follow a delayed response to environmental changes. In contrast, dissolved CO₂ generation showed distinct patterns: in the overlying water, CO₂ was positively correlated with several water quality indicators, such as chemical oxygen demand (COD), volatile fatty acids (VFAs), and nitrogen compounds, suggesting that its source is linked to various microbial and biochemical processes. In the sediment, CO₂ was primarily associated with COD and VFAs, indicating production mainly through fermentation. N₂O was detected only in pipeline sections with relatively higher dissolved oxygen (DO) levels, confirming that DO is a critical environmental factor governing the types of GHGs produced. Microbial analysis further highlighted that organic load is a central factor driving the differentiation of microbial community structure and the distribution of carbon metabolism pathways. High-organic-load conditions favored the enrichment of microbial communities specializing in the degradation of large organic molecules, with representative genera such as Syntrophorhabdus, leading to increased carbon flow toward CO₂ production. Under moderate-organic-load conditions, microbial communities that utilize small-molecule substrates, including genera like Lactivibrio, became more abundant. These communities showed a significant positive correlation with the abundance of the methane-producing gene mcrA-1, which is associated with increased CH₄ emissions in the overlying water. Microbial communities appeared to regulate the balance of CH₄ and CO₂ emissions through a "synergy-competition steady-state" mechanism, a dynamic regulation influenced by both the type and amount of organic matter present in the system. Finally, daily dynamic monitoring of GHG emission fluxes further confirmed that microbial community function plays a crucial role in regulating the timing and magnitude of GHG emissions.