Abstract: By using wavelet transform (WT),we analysed part of the LTD data collected from the equatorial western Pacific Ocean,the results show that the temperature finestructures has selfsimilarity and may be described with fractal.The singularity exponents of temperature series and their probability distributions are also calculated.The probability distribution functions (pdfs) have nearly a universal form.The nonzero probability at negative exponents implies the possible existence of singular points in the temperature profiles.When computing the average period of the WT,we found that the mean period △ of WT of the temperature series has a power dependence on the transform scale u,that is,△～α^β with β≈1.This nuggests that a simple dichomotous cascading mechanism may govern the finestructure dynamics.In addtion,since the amplitude to of the wavelet transform is related with the strength of temperature fluctuations,the vertical distribution of fluctuation energy can be calculated with wavelet transform.Compared to the FFT spectrum,the WT spectrum is more smooth and is capable of giving more accurate spectral slope.