Based on the ozonesonde data in Guangdong, Hong Kong, and Macao regions from 2022 to 2023, the vertical distribution characteristics of O₃ concentration were analyzed, and the applicability of Aqua satellite AIRS O₃ vertical profile product and ERA5 reanalysis O₃ vertical profile product were evaluated by using self-organizing map neural network (SOM) method. The vertical distribution of O₃ in the Greater Bay Area exhibited significant seasonal variations. In spring, summer, and winter, the ozone vertical distribution displayed a unimodal structure, with peak concentrations located near 700, 950, and 300 hPa, respectively. In contrast, the vertical distribution in autumn showed a bimodal structure, with peaks near 925 and 400hPa. The vertical differences in O₃ between stations were relatively small, with deviations between Guangdong and Hong Kong stations ranging from -3.2% to 11.0%. The quality of AIRS and ERA5 data in autumn and winter within the troposphere was better than that in spring and summer. At 850~200hPa, both AIRS and ERA5 data showed relatively good quality, with seasonal relative average deviations (Rad) ranging from 16.5% to 25.8% for AIRS and 15.1% to 25.7% for ERA5. The average correlation coefficients (r) for the seasons ranged from 0.47 to 0.75 for AIRS and 0.23 to 0.74 for ERA5. Below 850 hPa, the quality of AIRS and ERA5 data was relatively bad, with average value of r were 0.34 and -0.15. The vertical distribution of O₃ was categorized into 5 types. Among these, the data quality of AIRS and ERA5 was best under the type 1distribution structure, while it was worst under the type 2 and type 3 structures. Type 1 occurred more frequently in autumn (43%) and winter (61%), whereas type 2 and type 3 were more common in summer (66%). Type 4 and type 5 occurred more frequently in spring (85%).