Historical Seismograms and Hurricanes: Understanding Legacy Data and Developing Methods for Its Modern Usage

Abstract

Hurricanes pose great threats to society, and they may be getting stronger and exposing new areas to risk due to climate change. Prior to satellite observation in the 1980s, catalogs of hurricanes were based on ships’ logs, and thus, some events are missing. This incomplete record of activity limits the accuracy of hurricane behavior predictions. Therefore, novel sources of long-running quantitative observations over the last century that capture hurricane information are needed. A surprising candidate is ground motion. When ocean waves from hurricanes crash on the shore, they cause ground shaking known as “microseism” that is recorded by seismometers on land. Because seismic recording has been ongoing since the late 19th century, there is a unique opportunity to identify missing hurricanes through analysis of microseism.

While large archives of historical seismic data exist around the world, the conversion of these data on large sheets of paper recorded by century-old instruments into a resource compatible with modern seismology is not straightforward. This dissertation examines the feasibility of the opportunity to use these data with modern analyses, i.e., searches for missing hurricanes. Rigorous analyses to understand the fidelity of legacy seismic data are undertaken. Modern uses of legacy data require a comprehensive understanding of the properties of a wide variety of instruments. Data is examined, and features such as non-linear time distortions, amplitude uncertainties, and instrumental responses are identified and characterized. From these results, an understanding of how these features can arise from the multiple and various steps of the recording, scanning, and digitizing process is obtained, and either corrective measures or constraints on the uncertainties are suggested. A crucial outcome from these examinations is the determination that recording of microseism signals in historical data is robust.

Consequently, an algorithm that predicts the microseism that will arise from an oceanic storm is developed using an assimilation of modern seismic and hurricane data. It considers the propagation of energy from atmosphere, to ocean, to solid Earth, and gives a prediction of the frequencies, amplitudes, and arrival times of microseism originating from an oceanic storm. Comparing predictions made for storms from 2010 to 2022 to microseism observations from the seismic station in Harvard, MA, shows that the algorithm accurately captures the features of seismic observations. Moreover, results reveal important trends in how and where ocean swell forms, and suggest where various microseism excitation mechanisms are active. Application of the microseism prediction algorithm developed using the modern data to digitized historical seismograms from Harvard, MA, recorded in 1938 and 1939 provide proof of concept and lay the groundwork for systematic future analyses.