The most challenging problem for dealing with optimal control problem of stochastic flow is the formidable computational complexity. In the lecture,  we will explore a low-dimensional approximation to reduce the degrees of freedoms of the control system. Several methods, including POD (proper orthogonal decomposition ), CVT (centroidal Voronoi tessellation), CVOD (CVT-based POD) and PCE (polynomial chaos expansion), are considered and compared to form our reduced-order modeling (ROM).  Numerical tests of various situations are performed to validate our conclusion.