Rain Gauge vs. Radar Measurements - Modelling an Extreme Rain Event with High Spatial Variability

Abstract

This work investigates the usage of both rain gauge measurements and radar data of a 500-year event that happened in Innsbruck in July 2016. The occurring intensities varied strongly, e.g. they could be classified into a return period of 500 and 25 years within a distance of 2 km. The underlying research question is, whether a change of model performance can be expected due to a better representation of the spatial variability of this specific rain event. This further leads to the question, whether a spatially accurate representation of rain events is still decisive for model performance when modelling such intense rainfalls. For high return periods, other impacts may gain importance instead. Performance assessment is done by comparing simulated water levels to in-sewer measurements. Results show, that measured peak intensities and runoff volumes are significantly higher for radar measurements than for gauged precipitation which leads to a significant overestimation of water levels. Consequently, radar rainfall input did not improve the model performance. A reason could be found in the missing radar image at peak time, different temporal resolutions of the data sets, in the accuracy of the measuring methods during such heavy rainfall intensities or even in the underlying ideas of the initial model establishment.