Urban pluvial flooding prediction by machine learning approaches – a case study of Shenzhen city, China

Qian Ke,Xin Tian,Jeremy Bricker,Zhan Tian,Guanghua Guan,Huayang Cai,Xinxing Huang,Honglong Yang,Junguo Liu

doi:10.1016/j.advwatres.2020.103719

Abstract

Urban pluvial flooding is a threatening natural hazard in urban areas all over the world, especially in recent years given its increasing frequency of occurrence. In order to prevent flood occurrence and mitigate the subsequent aftermath, urban water managers aim to predict precipitation characteristics, including peak intensity, arrival time and duration, so that they can further warn inhabitants in risky areas and take emergency actions when forecasting a pluvial flood. Previous studies that dealt with the prediction of urban pluvial flooding are mainly based on hydrological or hydraulic models, requiring a large volume of data for simulation accuracy. These methods are computationally expensive. Using a rainfall threshold to predict flooding based on a data-driven approach can decrease the computational complexity to a great extent. In order to prepare cities for frequent pluvial flood events – especially in the future climate – this paper uses a rainfall threshold for classifying flood vs. non-flood events, based on machine learning (ML) approaches, applied to a case study of Shenzhen city in China. In doing so, ML models can determine several rainfall threshold lines projected in a plane spanned by two principal components, which provides a binary result (flood or no flood). Compared to the conventional critical rainfall curve, the proposed models, especially the subspace discriminant analysis, can classify flooding and non-flooding by different combinations of multiple-resolution rainfall intensities, greatly raising the accuracy to 96.5% and lowering the false alert rate to 25%. Compared to the conventional model, the critical indices of accuracy and true positive rate (TPR) were 5%-15% higher in ML models. Such models are applicable to other urban catchments as well. The results are expected to be used to assist early warning systems and provide rational information for contingency and emergency planning.

Highlights

We first elaborate on how the proposed Machine learning (ML) model estimates the rainfall threshold better than the current empirical threshold provided by the local authority (SMB, 2019) The threshold suggests any event is regarded as a pluvial flood if either 30-min rainfall depth is over 20 mm or 3-h rainfall depth is over 80 mm
The results show that ML models, especially linear discriminant analysis, can classify flooding and non-flooding by two principle components, raising the ACC and true positive rate (TPR) to 96% and 58%, respectively; and lowering the false alert rate to 25%
Despite uncertainty about the inundation records and ML models, this data-driven method provides a basis for generating rainfall thresholds for flood early warning and emergency response in Shenzhen

Summary

Introduction

Its impact, including loss of life and damages to both public and private properties, can be further deepened by climate change and accelerated urbanization (Falconer et al, 2009). This type of flooding usually occurs when intense rainfall exceeds the capacity of an urban drainage system. Recent extreme precipitation events have raised awareness from both authorities and citizens to the challenges of predicting and managing urban pluvial floods. In the UK, about 40% of damages and associated economic losses in cities are estimated to result from pluvial flooding (Douglas et al, 2010). Urban pluvial flood prediction and management is a critical topic in the context of urban water management

Objectives

Results

Discussion

Conclusion