The object of this paper is to compare the relocation results between the catalog data and the cross-correlation data of the Yiliang earthquake sequence occurred in September 2011, then discuss the impact on the location of different velocity models and confirm the seismogenic fault structures. We adopted double difference algorithm for earthquake hypocenter relocation by using catalog data and cross-correlation data respectively. We proposed the time domain multi-channel correlation detection function and used it to calculate waveform cross-correlation travel time difference data. In general, the number of the waveform cross-correlation data is less than the catalog data, so the number of relocation earthquakes is small. In order to overcome this shortcoming, we used the catalog data to supplement the waveform cross-correlation data and set the weight of catalog data less than the waveform cross-correlation data. To ensure that the relocation results were stable and reliable, we used multiple criteria to filter the travel time difference data. The proportion of two kinds of data was about 17. 5 1 in our cross-correlation data used in relocation. In order to find the most suitable velocity model for the regional crust structure and ensure the relocation results were reliable, we used four velocity models and chose the best results. Among the 944 M-L >= 1.0 earthquakes of Yiliang earthquake sequence from September 7 to October 7, 2012, we obtained 652 relocation events. The location of M(s)5. 7 main earthquake was 27. 516 degrees N, 103. 951 degrees E, and 6. 9 km in the depth. The location of M(s)5. 6 main earthquake was 27. 543 degrees N, 104. 023 degrees E, and 7. 27 km in the depth. The Yiliang earthquake sequence was located between Zhaotong-Ludian fault and Yiliang-Huize fault which were both NE trending, and the two main earthquakes occurred at opposite ends of the sequence. Relocation results show that the earthquake sequence was more concentrated along the nearby fault, the seismic strip extended for less than 20 km, the depth distribution became shallower than before and was mainly located between 5 km and 8 km depth and the cluster was slightly tilted. The average relative errors of east-west direction, south-north direction, depth and origin time ware 55. 2 m, 43. 0 m, 186. 7 m and 0. 01 s, the travel time residual was 16 ms. We used bootstrap method to evaluate the location errors and the correlations of the four source parameters. The results show that the location errors of the results of cross-correlation data were smaller than catalog data and the location accuracy was more than an order of magnitude higher compared to catalog locations. The cross-correlation data can also reduce the correlations of the source parameters. The location results of different velocity models show that there were significant correlations between focal depth and velocity model and we should choose the velocity model carefully. Based on the distribution of the earthquake sequence and regional modern tectonic stress field, we confirmed that Yiliang-Huize fault was the seismogenic structure of Yiliang earthquake sequence.
