558 - At-a-Station and Spatial Variability of Hershfield Rainfall Sampling Adjustment Factors: Implications for Engineering Practice

For storm durations relevant to stormwater design, DDF values should be derived from “truly continuous” rainfall data. However, most rainfall records are aggregated over fixed durations, which introduces underestimation. Indeed, if current DDF values were obtained from such totalized data, how were biased, “fixed maxima” converted into “true,” continuous maxima, before performing frequency analysis? Traditionally, we used Hershfield, also known as rainfall sampling adjustment factors (SAFs). These can only be derived at locations with higher resolution rain gauges, where almost continuous maxima can be extracted, allowing comparison with the “fixed” maxima extracted from totalized data.
We find several important issues and research gaps with SAFs in current practice: (i) multiple procedures have been advanced, but without comparing them, (ii) nobody has studied their at-a-station uncertainty, (iii) SAFs are determined and then applied as a mean across multiple locations, without considering their variability, and (iv) no one has considered how SAF uncertainty affects our predictions of extreme rainfall.
We use 862 German and 147 U.S. stations with continuous (1-min) rainfall data, performing a detailed analysis of the above issues. We compare SAFs using all proposed procedures, study SAF variability both at a station and in space, and then propose a unified engineering methodology for dealing with the effects of rainfall totalization on the estimation of extreme precipitation. Our results suggest that: (i) there is a preferred procedure for computing SAFs, (ii) at-a-station variability of SAFs is large enough to be relevant for engineering design considerations, and (iii) SAFs display spatial structure.