CS29C - Concurrent Session 29C: Development of Precipitation Frequency Areal Reduction Factors for Dams in British Columbia

An accurate estimate of watershed precipitation frequency (PF) is crucially important for computing flood-frequency for assessing spillway performance and conducting other dam safety studies that aim to protect critical infrastructure. While British Columbia (BC) has recently updated point-PF estimates, a missing piece of updated information is how to translate the point-based findings into PF information applicable to a specific watershed.

Spatial analyses of storm reconstructions in BC have been completed previously as part of probable maximum precipitation (PMP) studies which provide storm-focused areal reduction factors (ARFs). However, PF areal reduction factors (PF-ARFs) are needed for geographically fixed areas (watersheds) for developing watershed PF relationships for modeling watershed flood-frequency. For any given average recurrence interval (ARI) of precipitation, the ratio between the areal-average watershed PF curve and the areal-average point-PF curve is the PF-ARF. In particular, the PF-ARF values for a given watershed are not fixed values but vary with ARI.

In this study, PF-ARFs are developed for the BC region in two ways. First, a watershed-specific approach is used in which a stochastic storm generation algorithm is used to develop a watershed-PF curve for a given watershed. This is accomplished by sampling historical storms to re-create spatial patterns over the watershed and producing a simulated watershed-average annual maxima series which is then used to develop the watershed PF relationship. This method is conceptually correct, but computationally intensive, and provides a PF-ARF for only a single watershed area and duration, across multiple ARIs.

Therefore, an approach is needed that can take advantage of available spatial storm analyses to develop PF-ARFs applicable for watersheds across the entire BC region. A simplified approach is developed that utilizes significant historical storms of various durations and computes a storm-focused ARF as the ratio between the maximum point precipitation and the areal-averaged precipitation for various sizes of domains. By repeating this process and synthesizing results across multiple storms, durations, and area sizes, a statistically robust relationship for storm-focused ARFs as a function of watershed area can be developed. These results are then compared with the findings of stochastically generated watershed PF relationships as a calibration mechanism to identify adjustments needed to convert storm-focused ARFs to PF-ARFs for watersheds throughout British Columbia.