Although unmanned surface vehicles (USVs) have been extensively researched and applied to many typical scenarios, high performance autonomous control is still an open problem. One of their major problem stems from the difficulty in constructing a sufficient accurate and sufficient simple control-oriented dynamics model. Owing to the highly complicated physical mechanisms of hydrodynamics, USV systems possess strong nonlinearities and coupling, which make it exceptionally difficult to develop an accurate model structure. In real applications, extensively existing uncertainties, such as external disturbances due to wind, wave, current and measurement noise in the onboard sensor, will unavoidably influence identification precision with respect to the parameters in the USV’s model. Thus, in this paper, a new kind of nonlinear modeling scheme—the active modeling enhanced qLPV model—is proposed for a water-jet propulsion USV system. First, a qLPV structured model is derived by simplifying the real dynamics model, which has a simple quasi-linear structure and can approximate the nonlinearities of the hydrodynamics. Subsequently, a nonlinear version of the Kalman filter-based active modeling method is proposed to provide online estimates of the unstructured model to eliminate the errors due to the structure inaccuracy between the qLPV structured model and the real system. Finally, the newly proposed modeling scheme is tested on a real USV system and its superiority is shown through comparisons to some other traditional modeling methods.