In this paper

we applied subordinate function to obtain the seismic precursor anomalies of subsurface fluid in Guangdong and Fujian. The data used in this paper are over 10 years′ continuous records including water level of Huaan

Shaoguan

Huadu

Meizhou and radon in the water of Huaan

Zhangzhou

Quanzhou

Yongchun. We first calculated 13-point sliding mean of monthly average value

and then calculated the subordinate function. After transforming the subordinate function into 0-1

the anomaly-frequency-superposition were calculated. The diagram obtained was the object of studiy.The diagram shows that the subordinate function of data appeared to be of positive anomalies that were more than 0.5. Up to 29 positive anomalies of them appeared from January 1989 to September （2003.6 ） are aseismic anomaly

and the others correspond with the earthquakes （about M5） occurred in Guangdong and Fujian or the strong earthquakes （M≥6） occurred in Taiwan. After studying the anomaly-frequency-superposition diagram

we found that the value of anomaly-frequency-superposition of subordinate function can be separated into 5 anomaly periods with cluster feature. There was a statistical correlation between the length of the anomaly period and the intensity of earthquake. We may predict the earthquake intensity depend on the length of the anomaly period. The prediction index is whether the length of anomaly periods is up to 10 months or not. The superposition method was easier to eliminate the interference created by accidental factor compared with the single anomaly method.