Exploring Long-Term Flood Variability in the Lower French Broad River Using Fluvial Paleoflood Hydrology

Floods are a natural hazard affecting everyone globally. Extreme floods, defined here as floods with an annual exceedance probability ≤ 0.001, are underrepresented in streamflow records. This research aims to reconstruct past flood events from flood deposits and examine flood variability over longer timescales. In doing so, the research provides insights into past climate, drivers of extreme floods, and changes in flood frequency and magnitude. This study focuses on the French Broad River, a major tributary of the Tennessee River, in Sevier County Tennessee, < 1 km downstream of Douglas Dam. Layers of floodplain sediments were used to infer the size of past floods (paleofloods) based on peaks in coarse sand volume referred to as "flood peaks". A Bayesian age-depth model using long optical stimulated luminescence dating of quartz and radiocarbon-dated wood from distinct flood units was used to estimate paleoflood ages. Particle size data was statistically analyzed to identify "flood peaks" using End-Member Analysis and Locally Estimate Scatterplot Smoothing (LOESS) regression analysis. LOESS positive residuals plotted above the regression trendline were interpreted as sand peaks corresponding to large and extreme flood event deposition and were plotted against the age/depth model provided by rbacon to examine changes in frequency and magnitude of floods over the last 2500 Cal yr B.P. The results suggested that large and extreme flood frequency increased during the last 2,000 years of the Holocene but that flood magnitude decreased during the same timeframe. Seven peaks of extreme floods were observed during the following timespans: 452-460 Cal yr B.P., 469 Cal yr B.P., 604-672 Cal yr B.P., 791 Cal yr B.P., 1567-1589 Cal yr B.P., 1900- 1922 Cal yr B.P., and 1958-1997 Cal yr B.P. Flood activity decreased became notably less frequent between 1000-1250 years Cal yr B.P and a decrease in magnitude was observed in the last century. Analyzing observed and reconstructed strong/very strong El Niño events, they do not uniformly coincide with positive flood residuals, though some periods show overlap. Deviations from this pattern, suggest human settlement of the Tennessee River Valley and orbitally induced climate changes are also contributing factors to extreme flood variability.