A support vector machine-based machine learning method is proposed to estimate atmospheric optical turbulence profiles. Using sounding data collected in coastal areas, the measured temperature, pressure, relative humidity, wind speed, wind speed shear, and temperature shear profile data are used to estimate the atmospheric optical turbulence profiles on different days. The estimated profiles are compared with the actual measured values. Error analysis results show that the root mean square errors of the estimated atmospheric optical turbulence profile and actual measurement profile are 0.4461 and 0.3939 and the correlation values are 70.42% and 62.17% on 2018-05-05 and 2018-05-10, respectively. The results demonstrate that a support vector machine model trained and learned using actual measured data can accurately estimate the atmospheric optical turbulence profile in coastal areas. Despite some errors, the general trend is consistent. The feasibility of estimating atmospheric optical turbulence profile by support vector machine method is verified, which lays a foundation for directly estimating atmospheric optical turbulence profile by using conventional meteorological sounding data and establishing relevant models.