Underwater Acoustic Sensor Networks (UASNs) are the main technology of the underwater Internet of Things (IoT), providing a better technical means and information sensing platform for applications such as marine ecological environment monitoring and underwater search and rescue. In the application of UASNs, localization is crucial because data collection without accurate location information will be of no use. However, the presence of unfavorable factors, such as path loss, absorption loss, uncertainty in device transmit power, and unknown parameters of the underwater environment, renders it more challenging to achieve robust and precise localization in complex dynamic ocean environment. To this end, this study proposes a Coarse-to-Fine Localization method for UASNs under Unknown Multi-Parameters (CFL-UMP). First, the original nonlinear and nonconvex localization problem is transformed into an alternating nonnegative constrained least squares framework (ANCLS) using a Taylor first-order expansion and several approximation operations. Subsequently, the coarse localization stage is based on the Golub-Kahan bi-diagonalized least squares solution algorithm LSMR. However, LSMR typically only converges rapidly to a locally optimal solution. Consequently, the dichotomy method is employed in the fine localization stage. The approximate solution derived from the coarse estimation in the preceding step serves as the initial value for the dichotomy method, and the exact solutions for the underwater target location, the path loss factor, and the transmit power are simultaneously obtained through iterations. Furthermore, to demonstrate the superiority of the CFL-UMP algorithm, the computational complexity of the CFL-UMP algorithm is analyzed and the Cramér?Rao Low Bound (CRLB) is derived. Finally, compared with the selected benchmark algorithms, the simulation results confirm that CFL-UMP achieves optimal localization accuracy in different underwater simulation scenarios, effectively reducing the underwater localization error.