500 Years of Tropical Cyclone Precipitation Variability Along the Gulf Coast: Multidecadal Patterns, Trends, and Climate Connections

Abstract:

The complex role of tropical cyclone precipitation (TCP) in regional hydroclimates presents a unique challenge for predicting how TCP will change in a warming world. However, our understanding of the full range of TCP variability is limited by the short instrumental record. Here, we present a ~500-year (1540–2012 CE) reconstruction of hurricane season TCP totals for southern Mississippi using latewood width (LW) measurements of longleaf pine. This is the longest reconstruction of TCP and the first on the Gulf Coast. We developed a reference chronology using cores from living individuals and stumps in the De Soto National Forest. Additionally, we leveraged wood from a nearby historical house and coffins to extend our record. We used a two-thirds spline to standardize the growth series and remove non-climatic growth trends. Our regression model was calibrated using LW and instrumental TCP during the common period (1948–2012 CE). The reconstruction generally matches the instrumental TCP, capturing ~40 % of the variance. Our trend analysis suggests that TCP has significantly increased (~0.1 mm/decade) at our site over the last 500 years. Further, we see significant trends in most quartiles, including the third (Q3; positive) and fourth (Q4; negative) quartiles. Examining the frequency of Q4 years (i.e., the wettest years) using a 50-year moving window, we found active and quiescent multidecadal intervals and an increase in Q4 frequency. We examined our reconstruction alongside climate indices for the Atlantic Multidecadal Oscillation and Bermuda High (BH). Our results suggest that TCP variability, especially on multidecadal intervals, is connected to large-scale climate patterns, likely through their influence on environmental variables important for TC genesis, intensity, or tracks (e.g., SST anomalies, steering winds). Further, while pre-instrumental connections between TCP and the BHI remain unexplored, the link we document suggests that the BHI may influence TCP prior to 1948. Finally, several signals correspond to known TCs that resulted in floods, such as the 1888 Louisiana hurricane and the 1934 Central America hurricane. Given that TCs can result in severe flooding, our record can offer additional insight into paleofloods across the Southeast US during an interval of increased flood probability.