Tactile texture presentation on touch panels enhances their usability and realizes immersive user interfaces. This study develops a tactile texture rendering method for electrostatic friction displays. The method combines two rendering models for material textures compared with previous studies which focused on either of these two models. One of these models is a physical model that simulates low-frequency frictional signals depending on the exploratory finger velocities and contact loads. The other is an autoregression-based data-driven model for high-frequency textural friction. For user studies, we compared combining the two models with using only the physical model for the four types of materials. Although the effectiveness varied across the materials, the subjectively judged realism and identification of the materials were improved for the combined condition. The new method combining high-frequency textural information and low-frequency physical model-based friction is expected to provide realistic tactile textures for electrostatic surface tactile displays.