In view of the problem that the variation of sound velocity in different depth intervals has not been taken into account in the existing methods for the modeling of small time-scale sound velocity field in local sea areas, a method for the layering of sound velocity profiles based on statistical characteristics of the measured deep-sea sound velocity profiles is proposed. Furthermore, a method for the construction of layered time-varying model of local small-scale sound velocity profiles is proposed based on the Empirical Orthogonal Function (EOF). The diurnal variation characteristics of the first mode coefficient of layered-EOF and the equivalent average sound velocity of the sound velocity profiles are analyzed by using the measured data of whole deep sound velocity profiles in the South China Sea, and the accuracy of different fitting models is compared. Finally, the factors affecting the periodic variations of the sound velocity profiles are analyzed by using temperature and tide data in the experimental sea area. The results show that: ①Both the first mode coefficient of layered-EOF and the equivalent average sound velocity of the sound velocity profiles are characterized with daily periodic variation, with the daily periodic variation of sound velocity being not obvious in the upper layer, more obvious in the middle layer and small but still present in the lower layer. ②The influence of long-term variation term should be taken into account in fitting the local and small time-scale sound velocity. ③The change of the first mode coefficient of the EOF of the sound velocity profiles in the experimental sea area is significantly correlated to temperature. And the extracted time-varying characteristics of the sound velocity profiles are basically consistent with the periodic feature of tide in the experimental sea area.