We investigated sediment-water heat flux estimation in relation to applications (hydrodynamics simulation, evaporation studies, or global change effects assessment) in which sediment temperatures are not available because of technical complications (difficulties in installing sensors, for example) or because of methodology used (remote sensing, for example). We used field sediment temperature data measured every 10 min to 1 m depth at Doñana National Park marshland to obtain sediment thermal properties and to calculate diel sediment-water heat exchange through Beck’s sequential function specification method. We compare four models for the simulation of sediment-water heat flux by using surface temperatures. Two models need initial estimated inside sediment temperatures; the other two do not. Influence of estimated initial temperature profiles depends on the temperature distribution assumption and is significant for three days or fewer at the daily timescale. A model that does not use initial sediment temperatures provides accurate estimations with low computation time.