Quantifying Patterns of Streamflow Peaks over the Southeastern United States Using a Long-term Retrospective Data Set
Raczynski, K., & Dyer, J. (2023). Quantifying Patterns of Streamflow Peaks over the Southeastern United States Using a Long-term Retrospective Data Set. Hydrological Processes. 37(8), e14960. DOI:10.1002/hyp.14960.
Utilizing simulated daily flow values for 61 948 stream nodes from the National Water Model (NWM) v.2.1 retrospective data set, which covers the period 1979–2020, streamflow peaks (SFP) were determined using an objective threshold method based on the Fisher-Jenks breakpoint algorithm. This approach provides estimates of low probability, high in-channel water level conditions. Strong spatial relations in temporal streamflow peak distribution were observed, with four main spatial clusters covering: (1) the central parts of Mississippi to South Carolina; (2) the south and north parts of Mississippi to South Carolina; (3) Florida, the Gulf Coast, and parts of North Carolina and Virginia; and (4) the Atlantic Coast and parts of the Gulf Coast. The first area is characterized by high multiannual SFP occurrence with long total durations, high contribution to total discharge, and occurrence throughout the year with peaks in February and March. Increasing annual peak trends are present locally. The second area is characterized by similar parameters with increased flash flooding potential. In the third area, lower occurrence and share in total discharge are observed, with two peaks of occurrence in the September–October and March periods. In the fourth region, the lowest occurrences and shortest durations are present, along with strong seasonality manifested in September–October occurrences that constitute over 80% of all SFP volumes. Decreasing trends in maximal monthly flows for the June–October period are also observed here. The results provide valuable regionalized information about historical high flow characteristics across the southeast United States, which can be used as a basis for improved long-term flood prediction based on seasonality models and constitute a baseline for defining regional changes in high flow frequency and distribution.