Traces of Impact: multi-proxy analyses of major storm signatures in New Hampshire lake sediments

Thumbnail Image
non-tropical storms, climate change, hazard mitigation, instrumental meteorology in New England, sub-catastrophic but high-impact events
Tifft, William
AN ABSTRACT OF THE THESIS OF William D. Tifft for the degree of Master of Science in Environmental Science and Policy presented on November 20, 2020 Title: Traces of Impact: multi-proxy analyses of major storm signatures in New Hampshire lake sediments Abstract approved: Lisa A. Doner Major tropical and non-tropical storms cause extensive damage throughout the Northeast U.S. Recent climate change may be driving increases in their frequency, but large uncertainties exist around average return-intervals of so called 50-year, 100-year and larger storms, because of gaps in the historical record. Effective community resilience plans, especially for hazard mitigation, rely on the accuracy of the century-scale storm record, but instrumental meteorology in New England began in the late 1800s, too short a time for prediction of rarer, stochastic events. Longer records, that improve the accuracy of long-term storm frequency data, encompass varying climate regimes, and identify sub-catastrophic but high-impact events, will enable communities to better anticipate and mitigate storm effects. To improve documentation of major storms, and to test several hypotheses about the sensitivity of lake sediment archives to storms of varying intensities, I develop and compare four new paleolimnological data sets in central New Hampshire. This research primarily asks, do major storms, documented in meteorological and historical records as destructive and highly impactful, leave distinctive traces in lake sediment records? It also asks: what magnitude or duration of storm is needed to create a trace in the sediment record; do storm traces vary across watersheds experiencing other types of disturbance; are storm signatures consistent from watershed to watershed; do storm frequencies vary across climate regimes; and which analytical practices best capture storm histories across multiple lake basins? I address these research questions using new 210Pb-dated sediment core records from four central New Hampshire lakes: Norway Pond, Hancock; Pleasant Lake, Deerfield; Spofford Lake, Chesterfield; and Newfound Lake, Bristol and the list of major storms included in the 2013 State of New Hampshire Multi-Hazard Mitigation Plan. Sediment datasets include loss-on-ignition, dry bulk density, magnetic susceptibility, particle-size, stable isotopes, and geochemistry. Although time series of these datasets fail to show coherence around known storm events, multi-variate statistical analyses, including analysis of variance (ANOVA), principal component analysis (PCA), and binomial regression analysis, show that, in all the lakes except Pleasant Lake, there are statistically significant storm-coincident sediment changes. The ANOVA results indicate that zinc and lead are significantly different (0.05 confidence) between storm and non-storm events in all four lake records, and other heavy metals: Ba, Cd, Fe, Mn, Ni, and S, and High Frequency Mass Susceptibility are significantly different between storm and non-storm events in three of the four lakes. The PCA results reveal that the association of variables changed over time, with storms of the last 30-years clustering near each other. The binomial regression supports the concept that it is possible to use individual variables to identify the presence or absence of major storms in each sample. This technique has value for deriving the minimum long-term frequency of major storms in a given region.