Seismological observations on the 2019 March 21 accidental explosion at Xiangshui chemical plant in Jiangsu, China

On 2019 March 21, an explosion accidentally occurred at a chemical plant in Xiangshui, Yancheng City, Jiangsu Province, China. Using broad-band digital seismic data from East China, South Korea and Japan, we investigate properties of the Xiangshui explosion as well as two nearby chemical explosions and four nearby natural earthquakes in Jiangsu Province, East China. From Lg and Rayleigh waves recorded by regional networks, both body wave magnitude m(b) (Lg) and surface wave magnitude M-s (Rayleigh) are calculated for these events. The magnitudes of the Xiangshui explosion are m(b) (Lg) = 3.39 +/- 0.24 and M-s = 1.95 +/- 0.27, respectively. Both the empirical magnitude-yield relation for buried explosion and empirical yield-crater dimension relation for open-pit explosion are adopted for investigating the explosive yield. The result from the yield-crater dimension relation is approximately 492 ton, which is consistent with the ground truth and considerably larger than that from the buried source model. This also reveals that, for Xiangshui explosion, the explosion to seismic energy conversion rate is approximately one-third compared to a similar sized fully confined explosion. By comparing the body wave and surface wave magnitudes from explosions and nearby earthquakes, we find that the m(b):M-s discriminant calculated at regional distances cannot properly distinguish explosions from natural earthquakes. However, the P/S spectral ratios Pg/Lg, Pn/Lg and Pn/Sn from the same data set can be good discriminants for identifying explosions from earthquakes.

Automated summary

The study analyzed the Xiangshui chemical plant explosion in Jiangsu Province, China, and nearby chemical explosions and natural earthquakes using broad-band digital seismic data. Calculations of explosion magnitude and energy were consistent with ground truth, highlighting the characteristics of the explosion. Additionally, discriminants from P/S spectral ratios were found to effectively distinguish explosions from earthquakes.