Description
Flooding is a devastating natural hazard that claims many human lives and significantly impacts regional economies each year. To properly design civil infrastructure (e.g., bridges, culverts, storm sewers) to convey flood flows without failing, it is important to understand the potential flooding risk in terms of both recurrence interval (i.e., return period) and magnitude. Flood frequency analysis (FFA) is a form of risk analysis used to extrapolate the return periods of floods beyond the gauge record. Since discharge data for most catchments have been collected for periods of time less than 100 years, the estimation of design discharges for return periods larger than 100-years requires extrapolation. This study focuses on the assessment and modifications of flood frequency-based discharges for sites with limited sampling periods. Here, limited sampling period is intended to capture two issues: (1) limited number of observations to adequately capture the flood frequency signal (i.e., minimum number of annual peaks needed) and (2) climate variability (i.e., sampling period contains primarily "small" or "large" annual peaks only). A total of 34 gauges spread throughout the Susquehanna River basin (71,200 sq km) were used to investigate the impact of sampling period on flood frequency distributions. Subsets of a given annual maximum discharge series were created such that the sample period ranged from 10 years to the total number of years available. To estimate the flood frequency for each subset, the Log Pearson Type III distribution was fit to the logarithms of the instantaneous annual peak discharges. The resulting flood frequencies from these subsets were compared to the results from the entire record at each gauge. Based on the analysis, the minimum number of years required to obtain a reasonable flood frequency distribution was determined for each gauge. In addition, a method to adjust flood frequency distribution at a given gauging station with limited data based on "adjustment factors" obtained from other locations with longer periods of records was developed. The proposed method was applied to a subset of data study stations to evaluate the effectiveness of the adjustment factors. In addition, the correlation between the adjustment factors and rainfall anomalies was investigated. For the study watersheds, the results indicate that the adjustment factors are correlated to rainfall anomalies resulting in larger adjustments for periods sampling "dry" years relative to smaller adjustment factors for "wet" periods.
