The one-dimensional (1D) lake model, a submodule in the Weather Research and Forecasting (WRF) system (WRF-lake) was evaluated in a large dimictic reservoir, Miyun Reservoir, in northern China. Another 1D lake model, Minlake, which has been successfully applied in this reservoir and many other lakes/reservoirs, was applied as a reference. Simulated results showed that Minlake was able to reproduce the whole temperature profile of Miyun Reservoir accurately. For WRF-lake, although we used carefully chosen parameterization (the same surface absorption fraction, light attenuation coefficient and initial temperature as with Minlake, as well as modified surface roughness lengths), the model still had imperfect surface temperature simulation and completely inaccurate simulation in the deep layers. Several numerical experiments were carried out to study the impact of three factors (thermal diffusivity, inflow-outflow and topography) on the two 1-D models’ performances, and we found: (1) Modifying the diffusion coefficient of WRF-lake can improve the simulation of deep layers but cannot influence the surface temperature. (2) Inflow-outflow and topography have a significant effect on the whole temperature profile. Overall, the WRF-lake model can be coupled with WRF when applied to reservoirs like Miyun, as it can reproduce surface water temperatures to some extent (Nash–Sutcliffe efficiency coefficient ＞ 0.9). However, for better model performance in reservoir physical processes description and more extensive application to other reservoirs with larger flow rates or larger storage capacity, optimizing the parameterization for thermal diffusivity, inflow-outflow and topography needs to be done in future work.