Wake effects impose significant aerodynamic interactions a- mong wind turbines. Advanced wind farm control consider- ing the wake dynamics has become of great importance for the wind energy integration into the power grid. To address this issue, a control-oriented dynamic wind farm model is essential, which needs to be able to capture the dominant aerodynamic characteristics while ensuring a high computa- tional efficiency. In this paper, a data-driven reduced order method is adopted to develop a low-order dynamic WF mod- el. The control-oriented model captures the dominant flow dynamics seen by high-fidelity simulations from the data per- spective. Besides, the original input-output relation is well preserved. Thus, the proposed low-order surrogate model is promising to be used in wind farm dynamic control.