Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks
Abstract Computational complexity has been the bottleneck for applying physically based simulations in large urban areas with high spatial resolution for efficient and systematic flooding analyses and risk assessment. To overcome the issue of long computational time and accelerate the prediction pro...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2021-03-01
|
Series: | Journal of Flood Risk Management |
Subjects: | |
Online Access: | https://doi.org/10.1111/jfr3.12684 |
id |
doaj-28f55d46f1654760b8a4e13067512405 |
---|---|
record_format |
Article |
spelling |
doaj-28f55d46f1654760b8a4e130675124052021-02-15T15:02:32ZengWileyJournal of Flood Risk Management1753-318X2021-03-01141n/an/a10.1111/jfr3.12684Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networksZifeng Guo0João P. Leitão1Nuno E. Simões2Vahid Moosavi3Chair of Digital Architectonics, Institute of Technology in Architecture (ITA), Department of Architecture Swiss Federal Institute of Technology Zurich (ETHZ) Zürich SwitzerlandDepartment of Urban Water Management Swiss Federal Institute of Aquatic Science and Technology (Eawag) Dübendorf SwitzerlandINESC Coimbra, Department of Civil Engineering University of Coimbra Coimbra PortugalChair of Digital Architectonics, Institute of Technology in Architecture (ITA), Department of Architecture Swiss Federal Institute of Technology Zurich (ETHZ) Zürich SwitzerlandAbstract Computational complexity has been the bottleneck for applying physically based simulations in large urban areas with high spatial resolution for efficient and systematic flooding analyses and risk assessment. To overcome the issue of long computational time and accelerate the prediction process, this paper proposes that the prediction of maximum water depth can be considered an image‐to‐image translation problem in which water depth rasters are generated using the information learned from data instead of by conducting simulations. The proposed data‐driven urban pluvial flood approach is based on a deep convolutional neural network trained using flood simulation data obtained from three catchments and 18 hyetographs. Multiple tests to assess the accuracy and validity of the proposed approach were conducted with both design and real hyetographs. The results show that flood prediction based on neural networks use only 0.5% of the time compared with that of physically based models, with promising accuracy and generalizability. The proposed neural network can also potentially be applied to different but relevant problems, including flood analysis for flood‐safe urban layout planning.https://doi.org/10.1111/jfr3.12684convolutional neural networkdata‐driven emulationfast water depth predictionflood modelling |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Zifeng Guo João P. Leitão Nuno E. Simões Vahid Moosavi |
spellingShingle |
Zifeng Guo João P. Leitão Nuno E. Simões Vahid Moosavi Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks Journal of Flood Risk Management convolutional neural network data‐driven emulation fast water depth prediction flood modelling |
author_facet |
Zifeng Guo João P. Leitão Nuno E. Simões Vahid Moosavi |
author_sort |
Zifeng Guo |
title |
Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks |
title_short |
Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks |
title_full |
Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks |
title_fullStr |
Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks |
title_full_unstemmed |
Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks |
title_sort |
data‐driven flood emulation: speeding up urban flood predictions by deep convolutional neural networks |
publisher |
Wiley |
series |
Journal of Flood Risk Management |
issn |
1753-318X |
publishDate |
2021-03-01 |
description |
Abstract Computational complexity has been the bottleneck for applying physically based simulations in large urban areas with high spatial resolution for efficient and systematic flooding analyses and risk assessment. To overcome the issue of long computational time and accelerate the prediction process, this paper proposes that the prediction of maximum water depth can be considered an image‐to‐image translation problem in which water depth rasters are generated using the information learned from data instead of by conducting simulations. The proposed data‐driven urban pluvial flood approach is based on a deep convolutional neural network trained using flood simulation data obtained from three catchments and 18 hyetographs. Multiple tests to assess the accuracy and validity of the proposed approach were conducted with both design and real hyetographs. The results show that flood prediction based on neural networks use only 0.5% of the time compared with that of physically based models, with promising accuracy and generalizability. The proposed neural network can also potentially be applied to different but relevant problems, including flood analysis for flood‐safe urban layout planning. |
topic |
convolutional neural network data‐driven emulation fast water depth prediction flood modelling |
url |
https://doi.org/10.1111/jfr3.12684 |
work_keys_str_mv |
AT zifengguo datadrivenfloodemulationspeedingupurbanfloodpredictionsbydeepconvolutionalneuralnetworks AT joaopleitao datadrivenfloodemulationspeedingupurbanfloodpredictionsbydeepconvolutionalneuralnetworks AT nunoesimoes datadrivenfloodemulationspeedingupurbanfloodpredictionsbydeepconvolutionalneuralnetworks AT vahidmoosavi datadrivenfloodemulationspeedingupurbanfloodpredictionsbydeepconvolutionalneuralnetworks |
_version_ |
1724268821766733824 |