In this research calibration and uncertainty analysis of COUP model with focus on soil temperature simulation for 3-hours time scale have been performed for Hamedan synoptic station. The Generalized Likelihood Uncertainty Estimation (GLUE) was used for this object. In order to simulate the soil temperature, 22 parameters were chosen and by using the Monte Carlo stochastic sampling method from the uncertainty space of the parameters, 25000 scenarios were produced and model simulations were implemented. For separate behavioral and non-behavioral simulations, 3 criteria including Nash-Sutcliff, Mean Bias Error, and Root Mean Square Error were considered and acceptable thresholds for each criterion were defined. With applying the acceptable thresholds, 253 behavioral simulations were detected and used for calibration and uncertainty analysis of the model. Based on posterior parameter distributions some parameters were recognized as sensitive parameters. The median of behavioral simulations was considered for model calibration and the uncertainty analysis of the model was performed based on 90% confidence levels of behavioral simulation errors. The results showed that calibration of the model has considerably improved the performance of the model in comparison to default parameter values. In addition, the uncertainty analysis showed that the uncertainty of parameters has been considerably decreased in most cases with application of the GLUE method. Other differences between simulated and observed values were attributed to other sources of model uncertainty.