Abstract：Currently used seismic inversive technology mostly is the way controlled by seismic dominant frequency that makes the potential of relatively higher frequency and lower frequency in useful band without full utilization, so that the inversive results are unsatisfactory. Differing from the current inversive methods, the frequency-divided inversion is relying on logging and seismic data, u-sing the methods of support vector computer to study the amplitude response in different detective frequencies (amplitude versus frequency, simply called AVF), taking AVF as independent information of inversion and creating the non-linear mapping relationship between wave impedance curves of logging data and seismic waveforms, fully using the information of low, medium and high frequencies in seismic data to reduce the uncertainty of thin-layer inversion, so that the high-resolution inversive results could be achieved. The computation of models and real data processing showed the effectiveness and correctness of the methods. Frequency-divided inversion is also the high-resolution non-linear inversive method without wavelet detection and without initial model.
