Incorporating a simple two-layer reservoir into a coupled land surface and river routing model to improve river temperature simulations in the Tennessee River Basin

Accurately simulating river temperatures is essential to predict the impact of land use and climate change on freshwater ecosystems. Coupled land cover, river routing, and river temperature models have been developed to this end. However, the majority of these models lack reservoir thermal stratification, which can decrease downstream temperature due to reservoirs providing a consistent source of cool water from deep layers. The objective of this research was to verify if incorporating a simple reservoir module improved the accuracy of river temperature simulations. We simulated river temperatures from 1949 to 2010 in the Tennessee River Basin in the U.S.A. We simulated surface hydrologic fluxes with the Variable Infiltration Capacity model and routed runoff fields from VIC with the RVIC routing model to produce streamflow estimates at multiple locations within the basin. We then used these streamflow estimates as well as meteorological variables as input to the River Basin Model (RBM) to produce river temperature estimates. Initial simulations demonstrated an ability of RBM to capture the seasonal variation in observed river temperature. However, these simulations overestimated summer temperatures, which we attributed to a lack of simulated temperature stratification in the reservoirs. We incorporated a new module into RBM to simulate reservoir stratification with a simple physically-based two-layer model. We compared river temperature simulations for RBM with and without this new reservoir module with continuous river temperature measurements and periodic reservoir measurements stratified at depth. Our simulations reveal that the two-layer module improves the model accuracy. The model also accurately simulated reservoir stratification improving its potential to estimate future impacts of climate and land cover change on river temperature.