Assessment of uncertainty in regional and at-site precipitation frequency analysis for the localized region of Ellicott City, Maryland

Recent heavy rainfall events and flash floods in Ellicott City, Maryland, highlight the importance of having robust and up-to-date estimates of precipitation frequency. NOAA Atlas 14 is the US standard for estimating precipitation frequencies. Volume 2 of the Atlas, which provides precipitation frequency estimates for Maryland, was last updated in 2006 using data through 2000. The present study explores and compares the uncertainty in precipitation frequency estimates arising from factors such as choice of distribution, parameter estimation method, and length of the record (including consideration of more recent data) on estimates from at-site frequency analysis considering stations within 20 km radius of Ellicott City. Further, this study considers choice of distribution, number of stations, and length of the record on the estimates of regional frequency analysis considering the stations within radii of 20, 40 and 60 km of Ellicott City. The results of the at-site analysis show that choice of distribution and length of record are drivers of uncertainty, while, in regional frequency analysis, the main driver of uncertainty is choice of distribution. The effect of inclusion of more recent data had mixed impacts on the site-specific analysis and the effects on the regional frequency analysis were more limited. Comparing the hazard curves obtained from at-site and regional frequency analysis demonstrates the “averaging impacts” that may occur when using regional approaches and highlights how use of a regional approach may potentially mask localized effects in regions such that around Ellicott City.