Development of an ANN-Based Urban Flood Alert Criteria Prediction Model and the Impact of Training Data Augmentation

Kang, Hoseon; Cho, Jaewoong; Lee, Han-Seung; Hwang, Jeong-Geun; Moon, Hye-Jin

doi:10.9798/kosham.2021.21.6.257

Cited by 3 publications

(1 citation statement)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ref [12] used ArcGIS and HEC-GEOHMS to divide the Nakdonggang River watershed into 2268 sectors, drew rainfall-peak flow curves for different initial loss scenarios and antecedent moisture conditions, and calculated the threshold rainfalls. Ref [13] estimated threshold rainfalls for different durations using events with damage caused by past rainfall in urban areas and others without such damage. Ref [14] stressed the need for impact forecast and estimated threshold rainfalls using the SWMM model.…”

Section: Introductionmentioning

confidence: 99%

Estimation of Threshold Rainfall in Ungauged Areas Using Machine Learning

Choo

Choi

et al. 2022

Water

View full text Add to dashboard Cite

In recent years, Korea has seen abnormal changes in precipitation and temperature driven by climate change. These changes highlight the increased risks of climate disasters and rainfall damage. Even with weather forecasts providing quantitative rainfall estimates, it is still difficult to estimate the damage caused by rainfall. Damaged by rainfalls differently for inch watershed, but there is a limit to the analysis coherent to the characteristic factors of the inch watershed. It is time-consuming to analyze rainfall and runoff using hydrological models every time it rains. Therefore, in fact, many analyses rely on simple rainfall data, and in coastal basins, hydrological analysis and physical model analysis are often difficult. To address the issue in this study, watershed characteristic factors such as drainage area (A), mean drainage elevation (H), mean drainage slope (S), drainage density (D), runoff curve number (CN), watershed parameter (Lp), and form factor (Rs) etc. and hydrologic factors were collected and calculated as independent variables, and the threshold rainfall calculated by the Ministry of Land, Infrastructure and Transport (MOLIT) was calculated as a dependent variable and used in the machine learning technique. As for machine learning techniques, this study uses the support vector machine method (SVM), the random forest method, and eXtreme Gradient Boosting (XGBoost). As a result, XGBoost showed good results in performance evaluation with RMSE 20, MAE 14, and RMSLE 0.28, and the threshold rainfall of the ungauged watersheds was calculated using the XGBoost technique and verified through past rainfall events and damage cases. As a result of the verification, it was confirmed that there were cases of damage in the basin where the threshold rainfall was low. If the application results of this study are used, it is judged that it is possible to accurately predict flooding-induced rainfall by calculating the threshold rainfall in the ungauged watersheds where rainfall-outflow analysis is difficult, and through this result, it is possible to prepare for areas vulnerable to flooding.

show abstract

Section: Introductionmentioning

confidence: 99%

Estimation of Threshold Rainfall in Ungauged Areas Using Machine Learning

Choo

Choi

et al. 2022

Water

View full text Add to dashboard Cite

show abstract

Development of a Deep Learning-Based Flooding Region Segmentation Model for Recognizing Urban Flooding Situations

Yoo,

Lee,

Jeung

et al. 2024

Sustainability

View full text Add to dashboard Cite

Urban flooding has become increasingly frequent due to the rising intensity of rainfall driven by urban development and climate change. Effective prevention measures are crucial to mitigate the significant human and material damages caused by such events. Rapid and accurate pre-detection techniques can help to reduce the impacts of urban flooding. With the advancement of deep learning, deep neural networks (DNNs) have been successfully applied across various domains, including computer vision and speech recognition. In particular, DNNs for computer vision demonstrate high performance with relatively low computational costs. In this paper, we propose a flooding region segmentation model for urban underpasses based on the U-Net architecture. To train and evaluate the model, we collected datasets from the Mannyeon, Oryang, and Daedong underpasses in Daejeon. The proposed method achieved Dice coefficients of 98.8%, 94.03%, and 93.85%, respectively. This model demonstrates high segmentation performance in detecting flooded regions and can be integrated into continuous flood monitoring systems.

show abstract

Development of an Urban Flood Forecast Model Using Lumped Pipe Networks

Kim¹,

Cho²

2022

J. Korean Soc. Hazard Mitig

View full text Add to dashboard Cite

Typical urban flood simulation is conducted using physical models such as the 1D storm water analysis model and 2D inundation analysis model. Although 2D inundation analysis can predict flow velocity, inundation depth, and inundation area throughout an inundated urban area, it is difficult for it to produce a near real-time urban flood forecast for a metropolitan area such as Seoul. In this study, a physical urban flood forecast model was developed using lumped pipe networks to produce a near real-time urban inundation forecast. The dense pipe networks within a drainage basin were simplified as a single conceptual lumped pipe that has drainage and storage functions, and new pipe networks were constructed using lumped pipe networks. The model was applied to the August 2018 storm events in Seoul and showed a prediction accuracy of 0.71. The results demonstrated that the model can obviate the limitations of the near real-time operation of existing physical flood forecasting models to yield useful information for urban flood response, though showing room for improvement.

show abstract

Development of an ANN-Based Urban Flood Alert Criteria Prediction Model and the Impact of Training Data Augmentation

Cited by 3 publications

References 3 publications

Estimation of Threshold Rainfall in Ungauged Areas Using Machine Learning

Estimation of Threshold Rainfall in Ungauged Areas Using Machine Learning

Development of a Deep Learning-Based Flooding Region Segmentation Model for Recognizing Urban Flooding Situations

Development of an Urban Flood Forecast Model Using Lumped Pipe Networks

Contact Info

Product

Resources

About