Due to the small spatiotemporal scale and large change rate of turbulence, and the lack of large-scale field observation, there are fewer researches on the characteristics of atmospheric turbulence in China. Since 2011, China has begun to store second-level high-resolution sounding data, which makes it possible to deeply study the characteristics of turbulence in various parts of China. When using radiosonde data to diagnose the characteristics of turbulence, it is necessary to consider the impact of instrument noise on the analysis results. This paper analyzes the impact of instrument noise on the analysis results in China, which proves that noise can reduce the average turbulent dissipation rate in the troposphere, indicating the necessity of removing noise. On this basis, the Thorpe analysis method is used in analyzing the relationship between the intensity of turbulence (turbulence dissipation rate) and the type of underlying surface in detail. The results reveal that the turbulence in the troposphere and the annual average grassland turbulence are the strongest, followed by those of dryland crops, paddy crops, shrubs and wasteland. In the lower stratosphere, it is still affected by the underlying surface, and the annual average turbulence of grassland is still the strongest, and the next is the turbulence of dry land crops. The turbulence shows obvious seasonality on different underlying surfaces, and the turbulence intensity and frequency in the upper troposphere are the highest in summer. The results of this paper would help to understand the turbulence characteristics of different underlying surfaces, and provide guidance for the improvement of turbulence parameterization schemes of aircraft flight safety guarantee, air pollution model and so on.