Abstract:

As an important and new dating method, cyclostratigraphy is extensively used on analyzing data of deep sea core and terrestrial well logs. According to the procedure of data processing during cyclostratigraphic analysis, we write a program to recognize Milankovitch cycles in sediments by combining Matlab functions such as filter, interp1, mean, detrend, variance and pwelch. When used to estimate the power spectra of time series, a better estimation result could be gotten if the parameter settings satisfy certain conditions. These conditions are parameter window should be about equal to the wavelength of the lowest frequency signal among useful signals, parameter noverlap  be equal to 50% of the window length and parameter nfft be equal to the smallest power of two and must be great than the whole length of time series. The program can not be used to estimate the background red noise spectra as well as to estimate the confidence level of spectra peaks should be the shortcomings of the program.

Key words: cyclostratigraphy, Milankovitch cycles, spectral estimation, Matlab