Desertification poses a significant threat to grasslands in arid regions, necessitating the development of efficient seeding technologies for ecological restoration. This study integrates Computational Fluid Dynamics (CFD) with the Discrete Element Method (DEM) to investigate the airflow distribution characteristics and seed transport mechanisms within a multi-channel pneumatic seeding system. CFD analysis was initially used to model and optimize the airflow distributor, focusing on key parameters such as Reynolds number, inlet-outlet area ratio, branch spacing, and chamber diameter. The optimized design achieved nearly uniform outlet velocities of 17-18 m/s with minimal pressure loss. Subsequently, CFD-DEM coupling simulations were applied to the seed feeder, using coated Caragana korshinskii seeds, to analyze transport velocity and mass flow rate under varying structural parameters such as mixing section outlet height, inclined input angle, and input height. Regression analysis highlighted the significant influence of these parameters, with the optimal configuration yielding an outlet velocity of 12.47 m/s and a mass flow rate of 19.764 g/s. In field tests conducted in simulated desertified grasslands, seed distribution was uniform, with variation coefficients below 6.54 % at forward speeds of 1-3 m/s, and the lowest observed value of 5.88 % at 2.5 m/s. These results validate the proposed device's ability to achieve efficient, uniform multi-row seeding under harsh conditions, enhancing operational performance. This research provides valuable insights into the design of pneumatic seeding devices and can serve as a reference for optimizing other agricultural machinery.