A Singular Value Decomposition (SVD) based turbulence velocity structure function construction method was proposed. The velocity structure function constructed by the method was fitted with a turbulence model to realize the turbulence identification of a laser radar. First, the spatial data of lidar scanning was divided into distance gate sectors. Singular value decomposition was then performed on the turbulent wind field in each subsector, and the characteristic velocity reference value and turbulent pulsation velocity of each distance gate were obtained to construct the velocity structure function. The standard von Kármán turbulence model function was selected as the fitting constraint, and the cube root of the eddy current dissipation rate was obtained to assess the intensity of the turbulence. Finally, through measured data obtained from Lanzhou Airport, the performance of the velocity structure function and local average method of the SVD method under different turbulence intensities were compared and analyzed. The turbulence data reported by the crew were compared and analyzed, and the SVD method was used to predict the turbulence warning, which could reach 85.2%. This method is of great significance for improving airport turbulence detection and identification.