In this paper, we proposed a lattice hydrodynamic traffic model to study the congested traffic patterns on two-lane with the partial reduced lane via considering the upstream–downstream density difference and lane-changing rule. Using the lattice hydrodynamic model with the partial reduced lane for numerical simulation, various complex congested traffic patterns such as triggered stop-and-go (TSG), pinned localized cluster (PLC), oscillatory congested traffic (OCT), homogeneous congested traffic (HCT), homogeneous synchronized traffic (HST) and moving localized cluster (MLC) can be reproduced. Moreover, it is found that the sensitive coefficient λ of the upstream–downstream density difference plays a main role in generating these congested traffic patterns.