Abstract: The sea clutter data is processed by the Hilbert-Huang transformation to obtain the Hilbert marginal spectrum of the sea clutter, and the characteristic of its spectrum and the effect of the target on the spectrum are analyzed. It is found that the spectrum of the sea clutter is wider and the low frequency components are weaker when no target is hit, while it is narrower and the low frequency components are stronger when the target is hit. To describe the effect of the target on the spectrum, the membership degree is introduced, which is calculated by the partial Hilbert marginal spectrum where the target is hit, and it is found that the membership degree is smaller obviously when the target is hit, so the method based on the membership degree of the partial Hilbert marginal spectrum is proposed to detect the weak target. Shown as the simulation results of the real and simulating target data, the slowly fluctuate target in the sea clutter is detected effectively by the method whose detection performance is better than the method directly using the box dimension, the method to detect the target in the frequency domain and the multi-pulses CA-CFAR method.