In this paper, the statistical prop-erties of parameters of each path in wireless channel models are analyzed to prove that there is the static part in channel state infor-mation (CSI) which can be extracted from huge amounts of CSI data. Based on the anal-ysis, the concept of the Tomographic Channel Model (TCM) is presented. With cluster al-gorithms, the static CSI database can be built in an off-line manner. The static CSI database can provide prior information to help pilot de-sign to reduce overhead and improve accuracy in channel estimation. A new CSI prediction method and a new channel estimation method between different frequency bands are intro-duced based on the static CSI database. Using measurement data, the performance of the new channel prediction method is compared with that of the Auto Regression (AR) predictor. The results indicate that the prediction range of the new method is better than that of the AR method and the new method can predict with fewer pilot symbols. Using measurement data, the new channel estimation method between different frequency bands can estimate the CSI of one frequency band based on known CSI of another frequency band without any feedback.