Peakflow causative mechanisms and frequencies at Reynolds Creek, southwestern Idaho

Abstract

Peakflow frequency analysis is used in forest hydrology for designing stream crossing structures such as culverts and bridges. The traditional approach to frequency analysis is based on the assumption that the observed sample of peakflows represents a homogeneous population. It is not clear if using the "homogeneous" population of the annual series correctly represents the frequency distribution of high flows resulting from different causal mechanisms. This later situation may present a serious problem when extrapolation of the flow frequency curves is required, which is nearly always the case for the estimation of the 50- and 100-year peakflow. The objective of this study was to determine if frequency distributions are different when peakflows are separated by seasonal mechanisms. Additionally, it was desired to compare the peakflow estimates from the traditional annual series results with the separated distributions for a range of exceedence probabilities. Streamfiow data were obtained for three gages at Reynolds Creek Basin, Idaho. For water years 1964 to 1996, the instantaneous maximum flow was identified separately for peakflows that occurred in the winter, prior to snowmelt; in spring, during snowmelt; and in summer, after the snowmelt period. Causative mechanisms were inferred from the seasonal groups and included rain, rain-on-snow, rain on frozen ground in the winter, clear-sky or rain-on-snow in the springtime, and thundershowers in the summer. Separating peakflows resulted in differences between the flow frequency distributions. At the 0.40 km² headwater basin, Reynolds Mountain East, snowmelt caused the highest flows for exceedence probabilities between the 99% and 1.4%, whereas winter peakflows caused the highest flows for exceedence probabities less than 1.4%. At the 55 km² basin, Reynolds Creek at Tollgate, snowmelt caused the largest flows for exceedence probabilities between 99% and 7%, and winter mechanisms caused the largest flows for exceedence probabilities less than 7%. At the 239 km² basin, the Reynolds Creek Outlet, snowmelt caused the greatest flows for exceedence probabilities between 99% and 45%, and winter rainfall, rain-on-snow or rain on frozen soil caused the largest flows for exceedence probabilities equal to or less that 45%. At the Outlet basin, thundershowers caused higher flows than snowmelt mechanisms for exceedence probabilities less than or equal to 4%. The data suggest that stratifying peakflows based on season of occurrence provides a more accurate estimate of the peakflow frequency distributions. The individual curves provide a better idea of the distributions, especially when the curves are extrapolated to low, rarer, exceedence probabilities. Damage to structures may be reduced if causative mechanisms are considered when designing structures using peakflow analysis. Constructing a composite curve to incorporate the probability of each type of seasonal mechanism, in any given year, is an appropriate tool to consider where mixed mechanisms are evident in the streamflow record.