Modellering av urbane pluviale flommer med MIKE 21 : metodeutvikling

Abstract

Extreme rainfall events combined with the densification of urban areas increase the probability of urban pluvial flooding. These floods may cause substantial damage to infrastructure and inhabited areas which are costly to society. It is necessary to map potential problem areas. An important tool in this process is storm water modeling which can be carried out using GIS-analysis or two-dimensional hydraulic models. A GIS-analysis considers topography only and since no hydraulic calculations are made, this tool mainly provides a rough estimate of potential risk areas and flow paths. A program that can perform two-dimensional hydraulic surface flow calculations is MIKE 21 Flow Model. To execute these analyses a digital terrain model was established using ArcGIS.

In MIKE 21 the topography can be represented by squares structured in a grid (raster) or unstructured triangles and squares (flexible mesh). Models based on the former are less challenging to establish. However, the flexible mesh allows the user to vary the resolution across the site. This is a huge advantage since special attention and computer power can be directed at areas essential to the surface flow, while less important areas can be given a coarser representation. The main objective of this thesis is to find a method to increase the resolution of a raster without reducing the area of analysis nor increase the simulation period. The study was conducted based on a 100-year flood in an area of 27km2 located south of the Drammen river in the municipality of Drammen. The capacity of the drainage system was estimated to equal a 10-year flood; Thus the 100-year flood was reduced accordingly. Infiltration was not considered.

The investigated methods can be divided into two groups according to procedure. The first attempts used results from the entire catchment area as input for models that only covered the area of interest. Other attempts were based on a single setup where terrains of different resolutions were used simultaneously. DHI Denmark was contacted after several attempts without satisfying results. It appeared that raster models are not suitable for modelling urban pluvial floods with the given prerequisites.

Due to these results an additional objective was added, which aim was to analyze how MIKE-21 models react to changes in input data. The same area of analysis and general setup was used. As time was a limiting factor, only a water balance analysis was performed according to the modified parameter without calibrating the results any further. The purpose was to uncover the model’s reactions to change and investigate which underlying reasons could cause instability in the water balance. The differences between a model with an outlet and a closed model were investigated, as well as how a model responds to a change in the height of buildings in the digital terrain model. The heights investigated were 2 meters and 25 meters. Furthermore, it was considered how the model was affected by placing sinks to simulate drains in the catchment area and whether it would be a preferred method to using a reduced 100-year flood to represent the drainage system.

The results showed an instability in the water balance for all setups. The model with an outlet and the closed model gave very similar results, with the main difference being that the model with an outlet was a bit more time consuming to create. When the terrain using buildings of 2 meters were used MIKE 21 added more “correction water” than for the terrain with buildings of 25 meters. Correction water is added if the program detects negative water depths, which indicates an unstable area in the model. The flooding was also more severe when the buildings were given a height of 2 meters.

The sinks impact on the results were highly dependent on their location. In one setup 30 sinks were spread throughout the catchment area, while in another the same amount were placed exclusively in the area around Marienlyst stadium. The area around the stadium have showed excessive flooding in previous GIS-analyses and MIKE 21 simulations. The sinks were given a constant capacity equivalent to a 10-year flood. There was no noticeable effect on the flooding results when spreading the sinks in the catchment area. In contrast the setup containing sinks around Marienlyst stadium had visible effect on flooding in that area. In comparison, the reduced 100-year flood had a huge impact on the flooding results in the entire catchment area. However, the results from the setups with sinks contained such large instabilities that no conclusion could be drawn as to whether the method should substitute the reduced 100-year flood in representing the drainage system.

Ekstreme nedbørhendelser kombinert med fortetting av byer øker sannsynligheten for urbane flommer av pluvial karakter. Slike flommer kan forårsake store skader på infrastruktur og annen bebyggelse og påfører med dette samfunnet betydelige kostnader. Det er nødvendig med gode redskaper for å kartlegge potensielle problemområder. Et viktig verktøy i denne sammenheng er overvannsmodellering, som kan gjennomføres ved bruk av enkle GIS-analyser eller todimensjonale hydrauliske modeller. En GIS-analyse tar kun hensyn til topografi og siden ingen hydrauliske beregninger utføres vil dette verktøyet hovedsakelig gi en grov oversikt av potensielle risikoområder og strømningsveier. Et dataprogram som kan utføre todimensjonale hydrauliske overvannsberegninger er MIKE 21 Flow Model. I denne oppgaven ble ArcGIS benyttet for å etablere en terrengmodell til anvendelse i GIS-analyser og modellering i MIKE 21.