Based on Algorithmic Differentiation (AD), we present an application of the adjoint model in FVCOM (Finite Volume Community Ocean Model) generated by the AD-enabled NAG Fortran compiler. As opposed to Finite Difference (FD), AD yields derivatives accurate up to machine precision. In numerical simulations, the adjoint models allow the efficient computation of sensitivities for functions in which the number of inputs far exceeds the number of outputs. Our progress of the adjoint model implementation is illustrated by a basic case study of FVCOM.