AbstractThis article examines some of the difficulties associated with the determination of C2n over water in a coastal region using a bulk model. The analysis shows the need to supplement bulk models with elements that do not belong to traditional Monin–Obukhov surface-layer theory. A reexamination of the scintillation measurements collected during the Electro-Optical Propagation Assessment in a Coastal Environment (EOPACE) campaign leads the authors to include 1) the nonuniformity of sensible heat and humidity fluxes and 2) a deficit of the scintillation that seems to depend only on the characteristic virtual potential temperature. It is suggested that the anomalous scintillation deficit represents an alteration of a characteristic length related to the optical turbulence, likely caused by the interaction of the surface layer with the sea surface. The new parameters are estimated using Bayesian regression methods applied to the EOPACE data. Predictions obtained with this new model are then compared with scintillation measurements obtained during the recent Validation Measurement for Propagation in the Infrared and Radar (VAMPIRA) campaign. Better agreement is obtained with the new model than with a conventional bulk model. The implications of the modifications made to the calculation of C2n are discussed.