Sensitivity of pluvial flood exposure to the selection of intensity-duration-frequency data

Pluvial floods are and will remain an important driver of flood risk, especially in an urban context. Recently, several floods triggered by extreme rainfall made the news and led to many casualties, such as those in Valencia and Nepal in 2024. To better prepare for such disasters, urban planners may use pluvial flood maps to assess flood risk and plan accordingly. Typically, such maps are produced by distributing rainfall over topography using a hydraulic model which solves some variation of the shallow water equations. While the decision for a specific hydraulic model may impact pluvial flood maps, here we will focus on the role of pluvial input data.

Typically, intensity-duration-frequency (IDF) data is used to drive these models, yet these data are highly uncertain due to, for instance, the absence of accurate rainfall observations or the application of extreme value statistics.

Here, we present results of a sensitivity analysis in which we employed a range of global and national IDF data sets, such as NOAA Atlas 14, KOSTRA-DWD, BURGER, GPEX, PPDIST and PXR. Each data set is unique in the amount of data it was produced with, the spatial extent, the spatial regionalization of point-based estimates, the extreme value distribution used, and so forth. All IDF datasets were fed into a hydraulic model (LISFLOOD-FP) using the Chicago Design Storm (CDS) method to produce consistent and comparable maps of pluvial flood hazard for several test cases. Subsequently, the (dis-)agreement of the flood maps obtained is assessed.

To convert flood maps into impact, they are intersected with exposure data to obtain an estimate of average annual exposure (AAE) to pluvial floods, which is a better measure for assessing the impact of these floods.

While we expect that intensities extracted from the different IDF data sets will differ markedly, this study will shed light on the impact these differences may have on flood hazard and flood exposure estimates.