AbstractWe investigate how a tropical cyclone (Hurricane Isaac in 2012) generated seismic ground motions using seismic and barometric data from the Earthscope network. In the frequency band 0.01–0.02 Hz, seismic and surface pressure amplitudes show a systematic decreasing trend with distance from the center of the hurricane. However, the decreasing rate is much higher for seismic waves than for pressure. We develop a stochastic theory of seismic wave excitation by surface pressure that connects these two observed data sets; surface pressure is the excitation source, and seismic data are the resulting seismic wave field. This theory contains two parameters: (i) the pressure power spectral density (Sp) and (ii) the correlation length in the pressure field (L). Using the formula, we solve for the spatial variation of correlation lengths. The solution shows that longer correlation lengths in pressure are near the hurricane center. Because seismic wave excitation is proportional to L2Sp, the excitation for seismic waves becomes effectively more localized closer to the center. Also, the scaling relation between L and Sp leads to an excitation source which is approximately proportional to the third power of surface pressure. This centralized source for seismic wave excitation explains why the decreasing rate with distance is higher for seismic data than for barometric data. However, this spatial coherence mechanism may not be the only process, as strong turbulence near the center may cause transient bursts of pressure and also induce higher temporal correlation. These alternative mechanisms need to be carefully analyzed in the future.