SEISMIC IDENTIFICATION OF HURRICANES: SEEKING THE SIGNAL NEEDLE IN THE NOISE HAYSTACK

The short and incomplete observational record of North Atlantic hurricanes limits understanding of the link between rising sea surface temperatures and the frequency or intensity of those storms. North Atlantic hurricane records were based entirely on ship logs and land observations before aircraft reconnaissance started in 1944, and even then only limited areas were monitored. Hence the potential for sampling problems exists before satellite-based observation began in the mid-1960’s, implying that an undercount in the historical record is likely. To address this issue, we are developing methodology to improve the record of the number of North Atlantic hurricanes through the analysis of their signals recorded on decades of historical seismograms. Microseisms--signals derived from natural sources not related to earthquakes, also known as ambient seismic noise--are generated by atmospheric energy and so have been used as a proxy for oceanic wave climate and an indication of decadal-scale climate variability. Microseisms should therefore be usable to detect historical hurricanes that may have gone unobserved.

We use digital data from the HRV (Harvard, Massachusetts) and SJG (San Juan, Puerto Rico) seismic stations to study microseisms generated by well-characterized North Atlantic hurricanes. Although powerful hurricanes such as Andrew (August 1992) can easily be identified, doing so for storms of intensities less than category 3 is challenging. That is because the amplitude and frequency content of microseisms generated by hurricanes are controlled by factors beyond intensity alone. For weaker storms, additional variables--a storm's velocity over the ocean, hurricane-seismic station distance and velocity structure along the path between the two, ocean depth along that path, and the areal extent of the hurricane--become increasingly important. Hence although we are making progress, our present understanding of the "transfer function" between hurricane intensity and the microseism signal is too imprecise for these signals to be used as a discriminant for hurricanes of all intensities.