In the last two decades, spectral decomposition as a dynamic analysis method for seismic and prediction of effective thicknesses has been firmly established in the geological ex-ploration workflow. The reason for this is high efficiency in forecasting in conditions of high lateral and vertical variabil-ity of reservoir properties. However, the existing methods for interpreting the results of spectral decomposition have disadvantages. In our study, we tried to solve these disad-vantages and created a new technique for interpreting spec-tral decomposition, which was called "clustering of spectral curves".