Temperature changes, in the context of the phenomenon of climate change, are one of the important issues of the current societies. It is important to study the trend of temperature changes on a global, regional, and local scale. The general atmospheric models of the atmosphere, including the instrument of simulation and the assessment of future climate-temperature conditions, are considered under various scenarios. One of the main topics in the application of these models is their spatial separation. This problem can be explored with the aid of a downscaling method and can be solved to a large extent. One of the tools of statistical downscaling (SDSM) is using multivariate linear regression technique. In this study, using SDSM, at stations (Yazd and Shiraz) and according to the RCP 8.5 scenario for the upcoming periods 2018-2048, 2078-2078 and 2079-2099, using the output of the change model the canESM2 climate can be monitored for temperature variations. The average daily temperature observation for the 1976-2005 period has been modeled as input. Output results of the downscaling model showed that in the upcoming periods and based on the scenario under review, the average temperature increase would increase. This increase will increase by almost 1 degree Celsius in the first two stages, with a 30-year-old step, and in the last-20-year-old period. In the third period (2108- 2079) the largest increase, compared to the baseline is expected. The highest seasonal increase, For the summer season is and then autumn. Meanwhile, the first place is for Shiraz and Yazd stations in July. The months of June and August are jointly in the next place.