Abstract: The farmland-drainage ditch system has been widely used to intercept and assimilate the non-point source pollution in modern agriculture. This study aims to quantitatively assess the ecological buffering of the farmland-drainage ditch system. An ecological buffer zone module was constructed to fully consider the geometric features, hydraulic connectivity, and buffering process of farmland and its ditch networks. The complex, irregular field plots and multi-level ditch layouts were converted into a simplified, representative rectangular source area and buffer strip using geometric modeling and parameter optimization. The key parameters of the source area were derived from the weighted calculations of individual plot dimensions, including the final converted length (Ls), width (Ws), and slope (Ss). The ditch system was conceptualized as the second buffer zone. The parameters, such as the length (Lv), width (Ws), were conceptualized as the wetted perimeters. The slope (Sv), vegetation stem spacing (Svv), and vegetation height (Hvv) were transformed to incorporate the coefficients (e.g., \boldsymbol\alpha _\boldsymboli,\boldsymbolj , \boldsymbol\beta _\boldsymboli,\boldsymbolj , and \boldsymbol\gamma _\boldsymboli,\boldsymbolj ) for the effects of channelized flow and confluence. The vegetative filter strips modeling system (VFSMOD) was then coupled to simulate the runoff generation, sediment deposition, and pollutant interception. Sub-modules of the VFSMOD model also included the infiltration, surface runoff, sediment filtration, and pollutant transport. The preliminary calibration and validation were performed on the measured data from a rice cultivation area. The better performance was achieved with the relative deviations for the total nitrogen and total phosphorus interception rates below 20%. The parameters were calibrated for the application, such as a stem spacing of 2.5 cm, soil organic matter at 2.5%, and specific Manning's coefficients. Subsequently, a case study of the coupled model was applied in the high-standard farmland construction projects in Liyang City, Jiangsu Province, China. Sensitivity analysis was also conducted to determine the optimal thresholds for the key design parameters. The results showed that the interception rates of the runoff (R_R), sediment (R_S), and the pesticides with the varying mobility (R_P1, R_P2, and R_P3) were in a relatively optimal state when the buffer-to-source area ratio (R_BTS) reached 10%. Among them, R_R reached about 16.0%, R_S reached between 71.5% and 99.8%, and the pesticide interception rates ranged from 33.5% to 58.8% depending on the mobility. The optimal threshold for Svv was determined to be 2.5 cm. In addition, three sub-areas (Banzhu, Laiyang, and Gudu) were also selected to verify the model in the practical application. The converted (R_BTS) for Banzhu and Gudu were 6.5% and 6.3%, respectively, below the 10% threshold, while Laiyang was at 9.5%. The interception rates were lower after the simulation, compared with the optimal scenario. Therefore, the drainage ditches 01 and 07 in Banzhu were converted into ecological ditches, and then the ecological ditch section in Gudu was extended after engineering adjustments. All indicators were improved under the scenarios after simulation; for example, R_BTS, R_R, R_S, and R_P1 for Banzhu increased to 10.3%, 9.7%, 90.4%, and 37.0%, respectively. In conclusion, this finding can provide a framework and a coupled modeling tool to quantify the ecological buffer function of farmland drainage systems. Specifically, quantifiable thresholds can also be used for planning on ditch systems in high-standard farmland projects in green agriculture.