Autonomous Underwater Vehicles (AUVs) rely on integrated navigation systems and corresponding filtering algorithms to ensure mission success and the spatiotemporal accuracy of sampled data. Among these, the ensemble Kalman filter (EnKF) combines Monte Carlo methods with the Kalman filter, which is particularly suited for nonlinear systems. This study proposes an enhanced adaptive EnKF algorithm to improve the smoothness and accuracy of the filtering process. Instead of the conventional Gaussian distribution, this algorithm employs a Laplace distribution to construct the system state vector and observation vector ensembles, enhancing stability against non-Gaussian noise. Additionally, the algorithm dynamically adjusts the number of vector members in the ensemble using adaptive mechanisms by specifying thresholds during filtering to adapt the requirements of real-world observational settings. Using field trial data from DVL, GPS, and electronic compass measurements, we optimize the algorithm’s parameter settings and evaluate the overall performance of the algorithm. Results indicate that the proposed adaptive EnKF achieves superior accuracy and smoothness performance. Compared to the conventional EnKF and EKF, it not only reduces the average positioning error by 30% and 44%, respectively, but also significantly improves the filtering smoothness and stability, highlighting its advantages for AUV navigation.