The influence of temporal rainfall distribution and storm movement on flood depth in urban pluvial cloud burst modeling

Abstract: Pluvial floods are the most difficult and to date least investigated phenomena in urban hydrology. While efforts are being made to increase the knowledge base concerning this type of flooding, a large part of the difficulty lies in the nature of the precipitation. Convective storms represent most of the larger intensity short term rainfall in urban areas and is also the raintype, that is expected to increase the most in the future. The rain cells of this type have a more distinct boundary, larger intensity, a smaller extent and a shorter life span, than frontal rains. Combined with the low availability of densely spaced rain gauge networks and also low temporal resolution of measurements in 15 minutes intervals at best, makes this rain type still very difficult to analyze and even harder to predict. The resolution of cloud radar images at 2x2km and taken every 15 minutes is too coarse and the error reduction algorithms for radar based precipitation (HIPRAD) images to analysera in patterns are not sufficient by them selves to analyze the characteristics of such rainfields and the processes occurring within these fields. The spatial variation of raincells, their development and decay, the distance between them, and the velocity and direction of their movement can however be investigated employing a combination of densely spaced rain gauges and radar images to reach a more realistic representation of short-term precipitation for the use of in hydraulic models. The movement of rain fields has been investigated with two main areas of focus: The influence of direction or directional bias, often with an interest in the most crucial case referred to as the resonance effect, and in context of areal reduction of point rainfall. Most of these studies have been carried out with statistical methods and in laboratory experiments. In this study a hydraulic model was built on the terrain model of a realcity, a 28 km area in the city of Falun, to test the recently gathered information about the temporal variation of five empirical hyetographs with different peak arrival times and peak intensities, which are representative of Swedish climate. The hyetographs were produced and provided by SMHI. The empirical rain types were derived from 20 years of rain gauge observations and confirmed by radar images. For reference purposes, a standard Chicago design storm (CDS) rain was modeled as well. The simulated scenarios were modeled as a MIKE 21 hydraulic model, as a stationary scenario and in four movement directions. It was foundthat the empirical rain types produced lower inundation depth than the CDS, in a range of 20 to 50 % lower. The effect of modeling rainfall in motion produced on average only about 4-20 % lower water depths than the corresponding non-moving scenario. In a few instances, in a single evaluation point, the moving scenarios resulted in a relative water depth of a maximum of just above 1%. It was concluded that the conceptual approach of areal reduction from movement seems to be accurate and could help improve modeling rainfall in general, and specifically for cloud burst scenarios of shorter durations in urban catchments. It was also found that further investigation of the physical processes in rainfields could serve to increase the accuracy in areal reduction of precipitation for more realistic hydraulic models and in turn reduce over design.