# Intelligent Prediction Method for Waterlogging Risk Based on AI and Numerical Model

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Data and Methods

#### 2.1. Research Process

#### 2.2. Methods

#### 2.2.1. Construction of Rainfall Schemes

**Actual short-duration heavy rainfallscheme**

- 2.
**Characteristic rainfall schemes with different temporal and spatial distributions**

- 3.
- The rainfall process with the Chicago rainfall pattern

#### 2.2.2. Numerical Simulation Model

- One dimensional hydrodynamic model of river channel

^{2}), $Q$ is section flow (m

^{3}/s), $u$ is the velocity of lateral confluence (m/s), $t$ is time (s), $x$ is horizontal coordinates along the flow direction (m), $q$ is lateral flow (m

^{2}/s), $y$ is the water level (m), and ${S}_{f}$ is the friction gradient.

- 2.
- Surface runoff model

^{2}/s); $q$ is the rainfall intensity (m/s); $Z$ is the water level (m); and $u$ and $v$ are the average vertical velocity in $x$ and $y$ directions (m/s).

^{2}, which is shown in Figure 6.

- 3.
- Underground pipe network confluence model

#### 2.2.3. Training Samples

#### 2.2.4. Introduction to LSTM Neural Network Model

_{t}and three gates, the forget, input, and output gates, were added to the LSTM model to solve the problem of gradient disappearance or gradient explosion in the RNN model [30]. The structure of the model is shown in Figure 9.

#### 2.2.5. Waterlogging and Ponding Prediction Model Based on LSTM

_{t}was considered the prediction result. Ninety percent of the samples were used as training samples, and the remaining 10% were used as test samples. The rainstorm process without model training was forecasted by using the trained model.

## 3. Results and Discussion

^{2}. The closer the determination coefficient R

^{2}is to 1, the higher the fitting degree of the two curves.

^{2}, was greater than 0.85.

## 4. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- Tingsanchali, T. Urban flood disaster management. Procedia Eng.
**2012**, 32, 25–37. [Google Scholar] [CrossRef] [Green Version] - Huang, H.; Xi, C.; Zhu, Z.; Xie, Y.; Kai, L. The changing pattern of urban flooding in Guangzhou, China. Sci. Total Environ.
**2017**, 622–623, 394. [Google Scholar] [CrossRef] [PubMed] - Jha, A.; Lamond, J.; Proverbs, D.; Bhattacharya-Mis, N.; Barker, R. Cities and Flooding: A Guide to Integrated Urban Flood Risk Management for the 21st Century; World Bank Publications: Washington, DC, USA, 2012. [Google Scholar]
- Cheng, X.T. Urban Water Disasters and Strategy of Comprehensive Control of Water Disaster. J. Catastrophol.
**2010**, 25, 10–15. [Google Scholar] - Seyoum, S.D.; Vojinovic, Z.; Price, R.K.; Weesakul, S. Coupled 1D and Noninertia 2D Flood Inundation Model for Simulation of Urban Flooding. J. Hydraul. Eng.
**2012**, 138, 23–34. [Google Scholar] [CrossRef] - Zhou, H.-L.; Chen, Y.-B. 2D shallow-water simulation for urbanized areas. Adv. Water Sci.
**2011**, 22, 407–412. [Google Scholar] - Guo, K.H.; Guan, M.F.; Yu, D.P. Urban surface water flood modelling: A comprehensive review of current models and future challenges. Hydrol. Earth Syst. Sci.
**2021**, 25, 2843–2860. [Google Scholar] [CrossRef] - Zhang, H.P.; Wu, W.M.; Hu, C.H.; Hu, C.; Li, M.; Hao, X.; Liu, S. A distributed hydrodynamic model for urban storm flood risk assessment. J. Hydrol.
**2021**, 600, 126513. [Google Scholar] [CrossRef] - DHI. MIKE 1D, DHI Simulation Engine for 1D River and Urban Modelling; DHI: Singapore, 2012. [Google Scholar]
- DHI. MIKE 21 Flow Model FM, Hydrodynamic and Transport Module, Scientific Documentation; DHI: Singapore, 2007. [Google Scholar]
- Rangari, V.A.; Umamahesh, N.V.; Bhatt, C.M. Assessment of inundation risk in urban floods using HEC RAS 2-D. Model. Earth Syst. Environ.
**2019**, 5, 1839–1851. [Google Scholar] [CrossRef] - Huang, G.; Chen, W.; Yu, H. Construction and evaluation of an integrated hydrological and hydrodynamics urban flood model. Adv. Water Sci.
**2021**, 32, 334–344. [Google Scholar] - Jiang, C.; Zhou, Q.; Yu, W.; Yang, C.; Lin, B. A dynamic bidirectional coupled surface flow model for flood inundation simulation. Nat. Hazards Earth Syst. Sci.
**2021**, 21, 497–515. [Google Scholar] [CrossRef] - Hu, C.; Xia, J.; She, D.X.; Song, Z.; Zhang, Y.; Hong, S. A new urban hydrological model considering various land covers for flood simulation. J. Hydrol.
**2021**, 603, 126833. [Google Scholar] [CrossRef] - Soares-Frazao, S.; Lhomme, J.; Guinot, V.; Zech, Y. Two-dimensional shallow-water model with porosity for urban flood modelling. J. Hydraul. Res.
**2008**, 46, 45–64. [Google Scholar] [CrossRef] - Lecun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature
**2015**, 521, 436. [Google Scholar] [CrossRef] [PubMed] - Meer, P.; Mintz, D.; Rosenfeld, A.; Dong, Y.K. Robust regression methods for computer vision: A review. Int. J. Comput. Vis.
**1991**, 6, 59–70. [Google Scholar] [CrossRef] - Sundermeyer, M.; Schlüter, R.; Ney, H. LSTM Neural Networks for Language Modeling. In Proceedings of the Interspeech, Portland, OR, USA, 9–13 September 2012. [Google Scholar]
- Liu, Y.Y.; Li, L.; Zhang, W.H.; Chan, P.W.; Liu, Y.S. Rapid identification of rainstorm disaster risks based on an artificial intelligence technology using the 2DPCA method. Atmos. Res.
**2019**, 227, 157–164. [Google Scholar] [CrossRef] - Qing, L.; Hai-tao, X.; Jun-liang, Y. Daily Water Volume Prediction Algorithm of Urban Smart Water Based on Big Data. J. Beijing Univ. Posts Telecommun.
**2021**, 44, 82–88. [Google Scholar] - Barzegar, R.; Aalami, M.T.; Adamowski, J.F. Coupling a Hybrid CNN-LSTM Deep Learning Model with a Boundary Corrected Maximal Overlap Discrete Wavelet Transform for Multiscale Lake Water Level Forecasting. J. Hydrol.
**2021**, 598, 126196. [Google Scholar] [CrossRef] - Xinjun, W.; Xiaodong, Z.; Xi, D.; Zhentao, X.; Cheng, Q. CNN flood routing method based on data-driven training. J. Hydroelectr. Eng.
**2021**, 40, 8. [Google Scholar] - Ganggang, B.; Jingming, H.; Hao, H.; Junqiang, X.; Bingyao, L. Intelligent monitoring method for road inundation based on deep learning. Water Resour. Prot.
**2021**, 37, 6. [Google Scholar] - Gao, X.; Liu, J. Effect of Urbanization on River Hydrological Process in Shenzhen River Basin. Acta Sci. Nat. Univ. Pekin.
**2012**, 48, 153–159. [Google Scholar] - Liu, Y.-Y.; Li, Z.; Lei, L.; Yesen, L. Storm surge nowcasting based on multivariable LSTM neural network model. Mar. Sci. Bull.
**2020**, 39, 689–694. [Google Scholar] - Liu, Y.-Y.; Li, L.; Liu, Y.-S.; Chan, P.W.; Zhang, W.-H. Dynamic spatial-temporal precipitation distribution models for short-duration rainstorms in Shenzhen, China based on machine learning. Atmos. Res.
**2020**, 237, 104861. [Google Scholar] [CrossRef] - Leandro, J.; Chen, A.S.; Djordjevic, S.; Savic, D.A. Comparison of 1D/1D and 1D/2D Coupled (Sewer/Surface) Hydraulic Models for Urban Flood Simulation. J. Hydraul. Eng.
**2009**, 135, 495–504. [Google Scholar] [CrossRef] - Preissmann, A.; Cunge, J.A. Calcul des intumeseences sur machines electroniques. In Proceedings of the Ninth Convention of the International Association for Hydraulic Research, Dubrovnik, Croatia, 4–7 September 1961. [Google Scholar]
- Hochreiter, S.; Schmidhuber, J. Long Short-Term Memory. Neural Comput.
**1997**, 9, 1735–1780. [Google Scholar] [CrossRef] [PubMed] - Schmidhuber, J.; Gers, F.; Eck, D. Learning Nonregular Languages: A Comparison of Simple Recurrent Networks and LSTM. Neural Comput.
**2014**, 14, 2039–2041. [Google Scholar] [CrossRef] [PubMed]

**Table 1.**Comparison of maximum ponding depths of measured values and simulated results at waterlogging points during a heavy rainstorm in Shenzhen on 29 August 2018.

Serial Number | Waterlogging Point | Maximum Ponding Depth, Measured Values (cm) | Maximum Ponding Depth, Simulation Results (cm) | Error (%) |
---|---|---|---|---|

1 | Point A | 40 | 39.8 | 0.50 |

2 | Point B | 50 | 47 | 6.00 |

3 | Point C | 30 | 29.3 | 2.33 |

4 | Point D | 25 | 24.2 | 3.20 |

5 | Point E | 40 | 42.9 | 7.25 |

6 | Point F | 48 | 46.9 | 2.29 |

7 | Point G | 55 | 56.1 | 2.00 |

8 | Point H | 35 | 34.9 | 0.29 |

9 | Point I | 55 | 54.5 | 0.91 |

10 | Point J | 50 | 49.7 | 0.60 |

11 | Point K | 15 | 14.8 | 1.33 |

12 | Point L | 20 | 18.9 | 5.50 |

Average error | 2.68 |

**Table 2.**Comparison between the maximum ponding depths predicted by the numerical model and long short-term memory (LSTM) model and the measured values.

Serial Number | Waterlogging Point | Measured Values (cm) | Numerical Model Results (cm) | Error with Numerical Model (%) | LSTM Results (cm) | Error with LSTM (%) |
---|---|---|---|---|---|---|

1 | Point A | 40 | 39.8 | 0.50 | 39.2 | 2.00 |

2 | Point B | 50 | 47 | 6.00 | 51.2 | 2.40 |

3 | Point C | 30 | 29.3 | 2.33 | 28.3 | 5.67 |

4 | Point D | 25 | 24.2 | 3.20 | 23.5 | 6.00 |

5 | Point E | 40 | 42.9 | 7.25 | 41.2 | 3.00 |

6 | Point F | 48 | 46.9 | 2.29 | 46.9 | 2.29 |

7 | Point G | 55 | 56.1 | 2.00 | 56.1 | 2.00 |

8 | Point H | 35 | 34.9 | 0.29 | 35.7 | 2.00 |

9 | Point I | 55 | 54.5 | 0.91 | 55.2 | 0.36 |

10 | Point J | 50 | 49.7 | 0.60 | 50.7 | 1.40 |

11 | Point K | 15 | 14.8 | 1.33 | 15.9 | 6.00 |

12 | Point L | 20 | 18.9 | 5.50 | 19.3 | 3.50 |

Average error | 2.68 | 3.05 | ||||

Standard deviations of errors | 2.25 | 1.80 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Liu, Y.; Liu, Y.; Zheng, J.; Chai, F.; Ren, H.
Intelligent Prediction Method for Waterlogging Risk Based on AI and Numerical Model. *Water* **2022**, *14*, 2282.
https://doi.org/10.3390/w14152282

**AMA Style**

Liu Y, Liu Y, Zheng J, Chai F, Ren H.
Intelligent Prediction Method for Waterlogging Risk Based on AI and Numerical Model. *Water*. 2022; 14(15):2282.
https://doi.org/10.3390/w14152282

**Chicago/Turabian Style**

Liu, Yuanyuan, Yesen Liu, Jingwei Zheng, Fuxin Chai, and Hancheng Ren.
2022. "Intelligent Prediction Method for Waterlogging Risk Based on AI and Numerical Model" *Water* 14, no. 15: 2282.
https://doi.org/10.3390/w14152282