In order to study the effects of parameterization of physical processes of climate models on global energy balance, monsoon simulation, simulation uncertainty and other scientific issues, we conducted a parameter perturbation ensemble test on the climate model CAS-FGOALS of the Institute of Atmospheric Physics, Chinese Academy of Sciences. At the same time, 34 physical parameters were disturbed and 1024 numerical simulation test samples were generated. In addition to different physical parameters, each sample uses the same test conditions for continuous integration for 5 years, thus building a parameter disturbance dataset. The dataset consists of monthly atmospheric temperature, pressure, humidity, wind, precipitation, radiation and other elements, daily precipitation, 6-hour precipitation and radiation, and 3-hour precipitation.