We use the Mars Weather Research and Forecasting (MarsWRF) general circulation model to simulate the atmospheric structure corresponding to the landing location and time of the Mars Exploration Rovers (MER) Spirit (A) and Opportunity (B) in 2004. The multiscale capability of MarsWRF facilitates high-resolution nested model runs centered near the landing site of each of the rovers. Dust opacity distributions based on measurements by Thermal Emission Spectrometer (TES) aboard the Mars Global Surveyor spacecraft, and those from an old version of the Mars Climate Database (MCD v3.1 released in 2001) are used to study the sensitivity of the model temperature profile to variations in the dust prescription. The reconstructed entry, descent, and landing (EDL) data from the rover missions are used for comparisons. We show that the model using dust opacity from TES limb and nadir data for the year of MER EDL, Mars Year 26 (MY26), yields temperature profiles in closer agreement with the reconstructed data than the prelaunch EDL simulations and models using other dust opacity specifications. The temperature at 100 Pa from the model (MY26) and the reconstruction are within 5°K. These results highlight the role of vertical dust opacity distribution in determining the atmospheric thermal structure. Similar studies involving data from past missions and models will be useful in understanding the extent to which atmospheric variability is captured by the models and in developing realistic preflight characterization required for future lander missions to Mars.