Reservoirs impact the riverine system through flow regulation and biogeochemical and thermal flux sequestration. Here we present a reservoir thermal stratification model that can be coupled to hydrological, land surface, and earth system models at regional or global scale. One-dimensional heat equation was solved numerically by considering thermal and water fluxes from/to a reservoir. Upstream inflow into a reservoir is treated as an additional source/sink of energy, while downstream outflow represented a sink. The mixing of layers (energy and mass) in the reservoir is driven by eddy diffusion, and reservoir inflow and outflow. To improve the representation of heat and mass fluxes we used a new storage-area-depth dataset that is unique to individual reservoirs. The stratification model was developed over the continental US (CONUS). Hourly atmospheric forcing from North American Land Assimilation System (NLDAS) Phase II and simulated daily streamflow from MOSART (river transport model of Accelerated Climate Modeling for Energy-ACME) were used as inputs for the model for simulations of 1400 reservoirs between 2001-2010. The model was validated using selected observed surface and profile temperature data over 132 reservoirs that are subject to various levels of regulation. Nash-Sutcliffe values were higher than 0.5 for about 77% of the validated reservoirs. Average values of root mean square error, mean absolute error, and bias were 3.6, 2.8, and -1.1°C respectively. The new reservoir storage-area-depth dataset has also improved the performance of the stratification model over 69% of the validated reservoirs as compared to the surface temperature results simulated using the current assumption of rectangular surface area. The reservoir stratification module completes the representation of riverine mass and heat transfer in earth system models, which is a major step towards quantitative understanding of human influences on the terrestrial hydrological, ecological and biogeochemical cycles.