Research on arc fault location and detection needs to build an accurate and comprehensive impedance model that can reflect the characteristics of the arc. There is a problem in that existing arc impedance models cannot reflect intermittent and complete noise characteristics. Thus, an intermittent arc impedance model with segmented noise is proposed. First, the time-frequency domain characteristics of the experimental data of intermittent DC arc under resistive loads are analyzed, and the results show that the arc impedance varies with the current and is highly random, and the arc noise in the frequency domain presents differences in the spectral distribution of high and low frequency bands. Then, the impedance model with a probability distribution law is proposed to reflect the randomness and intermittency of the arc in the time domain, and the segmented noise model is used to reflect the characteristics of the noise spectrum distribution in the frequency domain. Finally, the performance of the model is verified based on experimental data, and the results show that the proposed model can accurately depict the random characteristics of the intermittent arc in both the time and frequency domains.