Associate Professor University of Wisconsin-Madison
Methods that combine modern multi-decadal gridded precipitation datasets and stochastic storm transposition (SST) are promising ways of improving estimates for extreme precipitation frequency. Hydrologists use SST to augment limited precipitation record lengths by including storms from a larger region when performing traditional statistical methods. SST and probable maximum precipitation (PMP) rely on many of the same principles but have slightly different focuses. SST results in annual exceedance probabilities for many return periods, whereas PMP generally aims to provide a physical-reasonable upper bound on the amount of rainfall a storm can produce. We believe our work on defining transposition domains for SST—specifically, identifying regions within which rainstorm properties are approximately homogeneous—can be applied to PMP estimation as well. We have developed a method that compares the ratio distributions of two locations’ extreme precipitation against a simulated ratio distributions calculated from the location and a sample of itself. We then perform a false discovery rate field significance test at the 5% level on the results to see if there is an acceptable level of false positives for the transposition domain. We have found that a strictly statistical method for delineating transposition domains based on annual maxima records is limited due to statistical similarities in precipitation that in some cases lead to transposition domains that are larger than we feel is reasonable. By including additional meteorological “ingredients” of extreme precipitation (e.g., precipitable water and temperature), we can make a self-contained process for delineating meteorologically homogeneous regions and lay the groundwork for exploring how domains might be altered due to changing climate.
.jpg "Daniel B. Wright, PhD photo")