Abstract:The procedure of usual deconvolution is that after the definition of expectation output, the deconvolution operator is derived from input seismic data, then convolved with seismic data to yield output data. My new method includes the steps: defining an area in a seismic section, calculating criterion with the use of predefined objective function, estimating wavelet by iterative method, deriving inverse wavelet from the known wavelet, and finally obtaining deconvolution output. Wavelet estimation is the key to the success of this method. Each iteration outputs deconvolution data. Then criterion is calculated, and next iterative parameter is determined and used to calculate correction value for improving wavelet. Good deconvolution section can be obtained finally.