Sea depth monitoring is a critical task to ensure the safe operation of harbors. Traditional schemes largely rely on labor-intensive work and expensive hardware. This study explores the possibility of deploying networked sensors on the surface of sea, measuring and reporting sea depth of given areas. We propose a Restricted Floating Sensors (RFS) model, in which sensor nodes are anchored to the sea bottom, floating within a restricted area. Distinguished from traditional stationary or mobile sensor networks, the RFS network consists of sensor nodes with restricted mobility. We construct the network model and elaborate the corresponding localization problem. We show that by locating such RFS sensors, the sea depth can be estimated without the help of any extra ranging devices. A prototype system with 25 Telos sensor nodes is deployed to validate this design. We also examine the efficiency and scalability of this design through large-scale simulations.