Flood Susceptibility Modeling in a Subtropical Humid Low-Relief Alluvial Plain Environment: Application of Novel Ensemble Machine Learning Approach

Pandey, Manish; Arora, Aman; Arabameri, Alireza; Costache, Romulus; Kumar, Naveen; Mishra, Varun; Nguyen, Hoang; Mishra, Jagriti; Siddiqui, Masood Ahsan; Ray, Yogesh; Soni, Sangeeta; Shukla, Uma Kant

doi:10.3389/feart.2021.659296

Cited by 31 publications

(9 citation statements)

References 151 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the context of SCAI values, higher values in very low susceptibility and lower values in very high susceptibility indicate higher reliability of the model 34) . This method is useful in assessing the reliability of the models.…”

Section: (2) Svm Model Resultsmentioning

confidence: 99%

A Novel Hybrid Approach of Spatial Prediction of Flood Susceptibility in Sri Lanka Using Bivariate Index of Entropy and Support Vector Machine Models

KURUGAMA,

KAZAMA,

HIRAGA

et al. 2023

Journal of JSCE

View full text Add to dashboard Cite

Frequent severe flood occurrences in Rathnapura city, Sri Lanka cause damages to both human lives and infrastructures. The existing flood mapping models have certain shortcomings that could be enhanced by utilizing more advanced and combined approaches. In this research, we suggested a new technique to enhance the prediction accuracy of flood susceptibility mapping by integrating the bivariate index of entropy (IoE) and support vector machine (SVM) as an ensemble method. The suggested approach was developed using four SVM kernels; polynomial , linear , sigmoid and radial basis function to examine the robustness of predictability of SVM technique. First a flood inventory map with 445 flood locations was created using satellite images, field survey and documentary sources. A spatial database was created with eight flood conditioning factors(FCFs) including altitude, slope, aspect, topographic roughness index(TRI), soil, Land use, distance from river and rainfall. Initially, IoE was utilized to assess the correlation between the various flood conditioning factors and flood occurrence. Afterwards, the results of the first step were utilized to perform SVM model. Model validation was carried out using seed cell area index(SCAI) and area under curve(AUC) methods. The highest success and prediction rates of 94.82% and 95.81% and lowest SCAI values of 0.18, 0.878 for very high and high susceptibility classes were achieved by ensemble IoE-SVM(RBF) model. Out of all the methods used, lowest accuracies were obtained by the individual IoE and SVM(RBF) method. Ensemble methods, enhance flood prediction capabilities and conditioning factor evaluation resulting in a more accurate final map.

show abstract

Section: (2) Svm Model Resultsmentioning

confidence: 99%

A Novel Hybrid Approach of Spatial Prediction of Flood Susceptibility in Sri Lanka Using Bivariate Index of Entropy and Support Vector Machine Models

KURUGAMA,

KAZAMA,

HIRAGA

et al. 2023

Journal of JSCE

View full text Add to dashboard Cite

show abstract

“…The AUROC curve ranges from 0.5 to 1.0, indicating varying degrees of accuracy. For further details on variable notations and their meanings, please refer to Pandey et al (2021).…”

Section: Model Performance Evaluation and Comparisonmentioning

confidence: 99%

Novel ensemble models and their optimization based flood susceptibility modelling in Indian Himalayan Foreland Basin

Pandey,

Costache,

Dash

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

This study focuses on the mapping of flood susceptibility in a specific region characterized by a low-altitude-range, sub-tropical monsoonal climate, and a riverine floodplain environment within the Middle Ganga Plain. To achieve this, four novel artificial intelligence model ensembles were employed. The aim was to compare the performance of these models in a distinct, tectonically active topoclimatic fluvial floodplain setting and determine the significance of different causative variables. The information obtained from this analysis can be highly valuable for planning and policymaking related to flood management principles. The entire Ganga Foreland Basin, which includes the Middle Ganga Plain, experiences frequent floods of varying magnitudes, underscoring the importance of this modeling exercise. To conduct this study, a comprehensive flood inventory and twelve selected flood conditioning factors were utilized in the development, testing (using 30% of the data), and validation (using another 30% of the data) of the four novel artificial intelligence models: LR-EBF, LR-FR, MLP-EBF, and MLP-FR. These models have been explored less in existing literature. The results revealed that the LR-based ensembles (LR-FR, LR-EBF: SRLR−FR = 86.7%, PRLR−FR = 83.9%, SRLR−EBF = 87.2%, PRLR−EBF = 84.7%) outperformed the MLP-based ensembles (MLP-FR, MLP-EBF: SRMLP−FR = 85.8%, PRMLP−FR = 82.8%, SRMLP−EBF = 86.4%, PRMLP−EBF = 84.4%) in the selected topoclimatic setting of the present study. Additionally, the LR-based ensemble with EBF demonstrated superior performance compared to the MLP-based ensemble with EBF. One notable finding of this study is the variation in performance among the four ensembles when applied in different topoclimatic and altitudinal range environments. The study revealed that the performance differences between LR-based ensembles employing both FR and EBF statistical models were consistent across all environments, exhibiting similar accuracy (in terms of AUROC) with less than a 5% disparity in success and prediction rates. This suggests that these models are likely to perform similarly across various environments. However, the performance differences in MLP-based ensembles, particularly with FR, were more significant, reaching up to a 10% disparity. Therefore, it is recommended to apply these models in high altitudinal range terrains with different topoclimatic settings to those considered in the present study. Notably, even the MLP-based ensemble with EBF displayed relatively consistent performance in varying topoclimatic and altitudinal range environments.

show abstract

“…Drainage systems often influences river overflow and continuous flooding in an area (Choubin et al, 2019). Higher drainage density results in lower infiltration and higher runoff (Arabameri et al, 2020;Pandey et al, 2021).…”

Section: River Densitymentioning

confidence: 99%

Data used for GIS-based Flood Susceptibility Mapping

Lee¹,

Rezaie

2022

GEO DATA

View full text Add to dashboard Cite

The dramatic increase in flood incidents as a significant threat to human life and property, environment, and infrastructure indicates the necessity of mapping spatial distribution of flood susceptible areas to reduce destructive effects of flooding. During the last decade, the integration of the geographic information system (GIS) with the remote sensing data provide efficient means to generate a more reliable and precise flood susceptibility map. The present study contains a review of 200 articles on the application of GIS-based methods in indicating flood vulnerable areas. The papers were reviewed in terms of influential variables, study area, and the number of articles published in the last 10 years. The review shows that the number of studies has increased since 2012. The total study areas covered 39 countries that were mostly located in Asia where the major developments and infrastructures have been constructed in the floodplains. The most common study areas was Iran (44 articles, 22%), followed by India (26 articles, 13%), China (26 articles, 11%), and Vietnam (15 articles, 7.5%). More than 90 variables were considered to map flood susceptible areas that the top 5 widely used flood conditioning factor are slope (98% of total articles), followed by elevation (92% of total articles), land use/land cover (79.5% of total articles), distance to the river (76.5% of total articles), and rainfall (73% of total articles). The review implies that many natural and anthropogenic factors affect flooding and the combination of both groups of factors is necessary to accurately detect and map flood-prone parts of the study area.

show abstract

Flood Susceptibility Modeling in a Subtropical Humid Low-Relief Alluvial Plain Environment: Application of Novel Ensemble Machine Learning Approach

Cited by 31 publications

References 151 publications

A Novel Hybrid Approach of Spatial Prediction of Flood Susceptibility in Sri Lanka Using Bivariate Index of Entropy and Support Vector Machine Models

A Novel Hybrid Approach of Spatial Prediction of Flood Susceptibility in Sri Lanka Using Bivariate Index of Entropy and Support Vector Machine Models

Novel ensemble models and their optimization based flood susceptibility modelling in Indian Himalayan Foreland Basin

Data used for GIS-based Flood Susceptibility Mapping

Contact Info

Product

Resources

About