Abstract: Ordinary kriging interpolation methods can be used in predicting submarine stratum thickness and obtaining its spatial distribution characteristics, using profile data as source data and borehole data as validation data. In this paper, three variogram models and different sampling interval are chosen to estimating submarine stratums thickness. The results show that ordinary kriging is an effective method of predicting submarine stratum thickness. The variogram model with the best structural analysis is not the model with minimal error. The choice of variogram model should be figured out through comprehensive analysis of interpolation output. The most suitable model for estimation of submarine stratums thickness in this well field is spherical model. Gauss model comes the second. For spherical model, the influence to regression effect in undulating stratum is greater than gentle stratum. An analysis of the variation ratio of the SE prediction show that reducing the sampling interval can greatly reduce the interpolation error for undulating stratum, and has little effect on improving the interpolation accuracy for gentle stratum.