Abstract

Stormwater management facilities are important elements of the civil infrastructure that can be sensitive to climate change, particularly to precipitation extremes that generate peak runoff flows. The design and anticipated performance of stormwater infrastructure is based on either the presumed characteristics of a “design rainstorm” or the continuous simulation of streamflow driven by a time series of precipitation. Under either approach, a frequency distribution of precipitation is required, either directly or indirectly, together with an underlying assumption that the probability distribution of precipitation extremes is statistically stationary. This assumption, and hence both approaches, are called into question by climate change. We therefore examined both historical precipitation records and simulations of future rainfall to evaluate past and prospective changes in the probability distributions of precipitation extremes across Washington State. The historical analyses were based on hourly precipitation records for the time period 1949–2007 from weather stations surrounding three major metropolitan areas of the state: the Puget Sound region (including Seattle, Tacoma, and Olympia), the Vancouver (WA) region (including Portland, OR), and the Spokane region.

Changes in future precipitation were simulated using two runs of the Weather Research and Forecast regional climate model (RCM) for the time periods 1970–2000 and 2020–2050, statistically downscaled from the ECHAM5 and CCSM3 Global Climate Model and bias-corrected against the SeaTac Airport rainfall record. Downscaled and bias-corrected hourly precipitation sequences were then used as input to the HSPF hydrologic model to simulate streamflow in two urban watersheds in central Puget Sound. Few statistically significant changes in extreme precipitation were observed in the historical records, with the possible exception of the Puget Sound. RCM simulations generally indicate increases in extreme rainfall magnitudes throughout the state, but the range of projections is too large to predicate engineering design, and actual changes could be difficult to distinguish from natural variability. Nonetheless, the evidence suggests that drainage infrastructure designed using mid-20th century rainfall records may be subject to a future rainfall regime that differs from current design standards.