Using one year ground atmospheric electric filed data by the AMEO340 installed in Nanjing University of Information Science and Technology,this paper calculates the ground atmospheric electric field sampling series by the fast Fourier transform in order to get power spectrum of the ground atmospheric electric field series.The ground atmospheric electric field data of sunny and lightning days are analyzed with seven-layer decomposition of the sym5 wavelet function,which decreases the overlapping degree of ground atmosphere electric field data waveform to some extent.The ground atmospheric electric field signals of lightning days are decomposed by the seven-layer decomposition.The low-frequency component of the ground atmospheric electric field signals not only can highlight its main change trend,but also can clearly identify the number of strong positive and negative ground flashes.So it can supply more effective information for the lightning nowcasting by using the characteristics of the ground atmospheric electric field singnals.