The acoustic monitoring of space-and timedependent structures in coastal waters constitutes a complex inverse problem which is attractive to physical oceanography. In this work the use of a nonlinear Kalman filtering scheme is investigated for tracking the evolution of the sound-speed field in a vertical section of a shallow water environment, taking into account the bottom acoustic properties. In support of the MREA/BP'07 experiment southeast of Elba, Italy, prediction results from the NCOM oceanic model provide realistic scenarios to test the tracking algorithm. Constraints in the form of rangedistributed observations such as the sea surface temperature (SST) are added in an attempt to obtain uniqueness of the solution. This paper shows that the formulation of the range-resolving inverse problem in a state-space model allows implementing a Kalman-based processor that effectively track the time variations of the 2-D sound-speed field. Furthermore, the estimations are enhanced by using an unscented Kalman filter (UKF) in place of the standard extended Kalman filter (EKF).