Flood susceptibility mapping: Integrating machine learning and GIS for enhanced risk assessment

Flooding is a recurrent hazard occurring worldwide, resulting in severe losses. The preparation of a flood susceptibility map is a non-structural approach to flood management before its occurrence. With recent advances in artificial intelligence, achieving a high-accuracy model for flood susceptibility mapping (FSM) is challenging. Therefore, in this study, various artificial intelligence approaches have been utilized to achieve optimal accuracy in flood susceptibility modeling to address this challenge. By incorporating the grey wolf optimizer (GWO) metaheuristic algorithm into various models—including recurrent neural networks (RNNs), support vector regression (SVR), and extreme gradient boosting (XGBoost)—the objective of this modeling is to generate flood susceptibility maps and evaluate the variation in model performance. The tropical Manimala River Basin in India, severely battered by flooding in the past, has been selected as the test site. This modeling utilized 15 conditioning factors such as aspect, enhanced built-up and bareness index (EBBI), slope, elevation, geomorphology, normalized difference water index (NDWI), plan curvature, profile curvature, soil adjusted vegetation index (SAVI), stream density, soil texture, stream power index (SPI), terrain ruggedness index (TRI), land use/land cover (LULC) and topographic wetness index (TWI). Thus, six susceptibility maps are produced by applying the RNN, SVR, XGBoost, RNN-GWO, SVR-GWO, and XGBoost-GWO models. All six models exhibited outstanding (AUC above 0.90) performance, and the performance ranks in the following order: RNN-GWO (AUC: 0.968) > XGBoost-GWO (AUC: 0.961) > SVR-GWO (AUC: 0.960) > RNN (AUC: 0.956) > XGBoost (AUC: 0.953) > SVR (AUC: 0.948). It was discovered that the hybrid GWO optimization algorithm improved the performance of three models. The RNN-GWO-based flood susceptibility map shows that 8.05% of the MRB is very susceptible to floods. The modeling found that the SPI, geomorphology, LULC, stream density, and TWI are the top five influential conditioning factors.

Malaysia, particularly Pahang, experiences devastating floods annually, causing significant damage. The objective of the research was to create a flood susceptibility map for the designated area by employing an Ensemble Machine Learning (EML) algorithm based on geographic information system (GIS). By analyzing nine key factors from a geospatial database, flood susceptibility map was created with the ArcGIS software (ESRI ArcGIS Pro v3.0.1 x64). The Random Forest (RF) model was employed in this study to categorize the study area into distinct flood susceptibility classes. The Feature selection (FS) method was used to ranking the flood influencing factors. To validate the flood susceptibility models, standard statistical measures and the Area Under the Curve (AUC) were employed. The FS ranking demonstrated that the primary attributes to flooding in the study region are rainfall and elevation, with slope, geology, curvature, flow accumulation, flow direction, distance from the river, and land use/land cover (LULC) patterns ranking subsequently. The categories of ’very high’ and ’high’ class collectively made up 37.1% and 26.3% of the total area, respectively. The flood vulnerability assessment of Pahang found that the Eastern, Southern, and central regions were at high risk of flooding due to intense precipitation, low-lying topography with steep inclines, proximity to the shoreline and rivers, and abundant flooded vegetation, crops, urban areas, bare ground, and rangeland. Conversely, areas with dense tree canopies or forests were less susceptible to flooding in this research area. The ROC analysis demonstrated strong performance on the validation datasets, with an AUC value of >0.73 and accuracy scores exceeding 0.71. Research on flood susceptibility mapping can enhance risk reduction strategies and improve flood management in vulnerable areas. Technological advancements and expertise provide opportunities for more sophisticated methods, leading to better prepared and resilient communities.

Spatial prediction of flood-prone areas using geographically weighted regression

In recent decades, floods have been the most common, complex, and destructive natural calamities worldwide. Hence, for inclusive flood risk assessment, creating flood susceptibility mapping to demarcate flood-vulnerable zones is fundamental for decision makers. To assess flood-prone locations in the Akre, Iraqi Kurdistan Region, fundamental for susceptibility mapping was undertaken using geographic information systems, remote sensing, and an analytic hierarchy process model. To assess flood susceptibility, the geographic information systems framework used 15 ideal causative factors for flooding: altitude, slope, distance to streams, flow accumulation, drainage density, rainfall, soil type, lithology, curvature, topographic wetness index (TWI), topographic roughness index stream power index, stream transport index, land use/land cover, and normalized difference vegetation index. The factors contributing to flooding were optimally weighted with respect to the proposed model. The final flood susceptibility map was reclassified into five different classes of susceptibility to flooding: very low (16.64% of the study area); low (19.53%); moderate (38.92%); high (17.83%); and very high (7.08%). The area under curve values for the predicted rate and success rate of the AHP model were 0.956 and 0.971, respectively. Therefore, the results were accurate and reliable. The AHP model is a powerful method for fundamental for susceptibility mapping to mitigate the serious impacts of flooding and assist scholars, local governments and policymakers in future master planning.

There has been a rapid evolution of tree-based ensemble algorithms which have outperformed deep learning in several studies, thus emerging as a competitive solution for many applications. In this study, ten tree-based ensemble algorithms (random forest, bagging meta-estimator, adaptive boosting (AdaBoost), gradient boosting machine (GBM), extreme gradient boosting (XGBoost), light gradient boosting (LightGBM), histogram-based GBM, categorical boosting (CatBoost), natural gradient boosting (NGBoost), and the regularised greedy forest (RGF)) were comparatively evaluated for the enhancement of Copernicus digital elevation model (DEM) in an agricultural landscape. The enhancement methodology combines elevation and terrain parameters alignment, with feature-level fusion into a DEM enhancement workflow. The training dataset is comprised of eight DEM-derived predictor variables, and the target variable (elevation error). In terms of root mean square error (RMSE) reduction, the best enhancements were achieved by GBM, random forest and the regularised greedy forest at the first, second and third implementation sites respectively. The computational time for training LightGBM was nearly five-hundred times faster than NGBoost, and the speed of LightGBM was closely matched by the histogram-based GBM. Our results provide a knowledge base for other researchers to focus their optimisation strategies on the most promising algorithms.

Floods, which are common than other natural disasters like earthquakes, heavy precipitations, and droughts, are one the primary effects of global climate change and have major effects on human safety, sustainable development, and economic growth. As climate warming and intensifying hydrologic cycle worsen, global flooding risks may increase, potentially impacting Afghanistan as well. Severe flooding being caused by rising temperatures, erratic rainfall patterns, and extreme weather in Afghanistan, especially in the region of Kandahar. Despite the significance of identifying and mapping flood – prone areas, this province has not participated in any previous studies done on the topic at hand. Therefore, the aim of this research was to develop a flood susceptibility map for Kandahar province and identified flood – prone areas with high levels of occurrence by integrating Geographic Information System (GIS) and Multi – Criteria Decision – Making (MCDM) method, with Analytic Hierarchy Process (AHP). To achieve the study’s goal, 11 Flood Causative Factors (FCFs), such as runoff potential, slope, rainfall, flow accumulation, distance from rivers, topographic wetness index (TWI), drainage density, lithology, Digital elevation model (DEM), sediment transport index (STI), and curvature, were weighted and overlayed. The resulting map depicted five different levels of susceptibility to flooding: least, low, moderate, high, and very high. The model’s final map of flood susceptibility was found to be in line with past flood occurrences in the study area, demonstrating the successful outcome of the methodology utilized to locate and map flood – prone areas.

This study responds to the growing concern about flooding and its consequences, particularly in areas prone to severe meteorological events, by employing an innovative approach to identify flood susceptibility in Kayseri. The methodology combines machine learning (ML) algorithms, namely extreme gradient boosting (XGB), categorical boosting (CB), and gradient boosting (GB), with hyperparameter optimization strategies through a hybridization process utilizing the energy valley optimizer technique. A total of 6000 data points were designated for the purposes of training, testing, and validation. In order to create these models, a total of nine variables, which have been identified as influential factors in the occurrence of floods, were selected based on data availability and a review of the relevant literature. It is noteworthy that elevation and rainfall were identified as pivotal predictors across all models. The CB model demonstrated robust predictive accuracy, with a substantial majority of instances correctly classified. The AUC values for the XGB and GB models remain notably high at 0.98, indicating robust predictive power and generalization capabilities. In the test phase, the AUC values underscore the superior performance of the XGB (0.9763) and GB (0.9739) models, with the CB model also demonstrating robust results at 0.9677. This study introduces a novel approach to flood susceptibility mapping by utilizing a range of ML methods. Its key innovations lie in the superior performance of these algorithms compared to traditional methods, as well as their inherent flexibility and heuristic capabilities. The generated flood susceptibility maps offer a detailed insight into the spatial distribution of flood susceptibility, with significant implications for urban planning and disaster preparedness.

Urban flooding is prevalent in low-lying, coastal regions, where subtle topographic variation, shallow groundwater, and impervious surfaces govern inundation dynamics. This study evaluates urban flood susceptibility across Miami-Dade County by integrating flood-conditioning factors, including elevation, slope, rainfall, land use/land cover, distance to roads and open water, stream power index (SPI), topographic wetness index (TWI), groundwater depth, and flow accumulation within an Analytical Hierarchy Process (AHP)-based weighted overlay framework. The AHP-derived weights demonstrated strong consistency (consistency ratio = 0.022) and were applied to reclassify each conditioning factor into five flood susceptibility classes—very low to very high. The model performance was evaluated using the Federal Emergency Management Agency (FEMA) flood zone, and the findings demonstrated that the AHP-based framework effectively differentiates flood susceptibility at a fine urban scale, achieving strong predictive performance; area under the Curve (AUC) = 0.85. The results also reveal pronounced spatial variability in flood susceptibility, with northeastern urbanized areas, particularly in Hialeah, Miami Gardens, Miami Lakes, and Downtown Miami, exhibiting higher susceptibility compared to the northwestern Everglades region. Overall, this study presents a robust urban flood susceptibility framework that supports improved flood risk assessment and decision-making in complex urban coastal environments.

This article presents a geographic information system (GIS)-based artificial neural network (GANN) model for flood susceptibility assessment of Keelung City, Taiwan. Various factors, including elevation, slope angle, slope aspect, flow accumulation, flow direction, topographic wetness index (TWI), drainage density, rainfall, and normalized difference vegetation index, were generated using a digital elevation model and LANDSAT 8 imagery. Historical flood data from 2015 to 2019, including 307 flood events, were adopted for a comparison of flood susceptibility. Using these factors, the GANN model, based on the back-propagation neural network (BPNN), was employed to provide flood susceptibility. The validation results indicate that a satisfactory result, with a correlation coefficient of 0.814, was obtained. A comparison of the GANN model with those from the SOBEK model was conducted. The comparative results demonstrated that the proposed method can provide good accuracy in predicting flood susceptibility. The results of flood susceptibility are categorized into five classes: Very low, low, moderate, high, and very high, with coverage areas of 60.5%, 27.4%, 8.6%, 2.5%, and 1%, respectively. The results demonstrate that nearly 3.5% of the study area, including the core district of the city and an exceedingly populated area including the financial center of the city, can be categorized as high to very high flood susceptibility zones.

The mapping and assessment of flood susceptibility is an integral component of flood mitigation and prevention programs, by determining the most vulnerable regions and the associated characteristics that influence the flood susceptibility. Hence, the aim of the present study is to identify flood-prone areas in the Fez-Meknes region (Morocco) for the first time using a multicriteria approach, in particular the Analytical Hierarchy Process (AHP) technique and Geographic Information Systems (GIS). A total of fifteen conditioning factors for flooding were selected: distance to rivers, river network density, precipitation, flow accumulation, elevation, slope, plane curvature, TWI, aspect, NDVI, LULC, TRI, geology, soil type, and SPI. All factors were defined as raster data sets with a resolution of 30 x 30 m. The results showed that, the efficiency tests of the flood susceptibility map show a good accuracy using an area under the curve (AUC) by remarkably good number (0.90).in addition, LULC was recognized as the most significant factor, which is followed by the stream power index that affect the flood map.

Floods represent the most frequent natural hazard, generating significant impacts on people as well as considerable economic and environmental losses worldwide. These events are particularly exacerbated by extreme climatic phenomena, such as the 2017 Coastal El Niño, the most intense in the past century, with the Piura region of Peru being the most affected. Flood susceptibility mapping (FSM) are essential for mitigating the negative impacts of floods through land-use planning, policy and plan formulation, and fostering community resilience for the sustainable occupation and use of floodplains. This study aimed to develop FSM in northern Peru, particularly in the Piura region, using a hybrid methodology integrating optical and radar remote sensing (RS), GIS, and machine learning (ML) techniques. Sentinel-1 data were used to map flood extent using the Normalized Difference Flood Index (NDFI), while flood susceptibility was modeled using ten topographic variables (derived from a DEM), the Normalized Difference Vegetation Index (NDVI), geology, and geomorphology; issues related to correlation and multicollinearity among topographic variables were addressed through Principal Component Analysis (PCA), selecting four principal components that explained 75.4% of the variance. Six FSMs were generated using Support Vector Machine (SVM) and Random Forest (RF), combined with different methods to estimate the quantitative relationship between variables and flood occurrence: Quantiles (q), Frequency Ratio (FR), and Weights of Evidence (WoE) (SVM-q, SVM-FR, SVM-WoE, RF-q, RF-FR, and RF-WoE). Model validation was performed using metrics such as the Area Under the ROC Curve (AUC), F1-score, and Accuracy, along with a cross-validation analysis. The results revealed that the RF ensemble model with WoE (RF-WoE) exhibited the best performance (AUC = 0.988 in training and >0.907 in validation), outperforming the SVM-based models; the SHAP analysis confirmed the significance of geology, geomorphology, and aspect in flood prediction. The resulting susceptibility maps identified the lower Piura River basin as the most vulnerable area, particularly during the 2017 Coastal El Niño event, due to morphological factors and inadequate land occupation. This study contributes to the field by demonstrating the effectiveness of a hybrid methodology that combines PCA, machine learning, and SHAP analysis, providing a more robust and interpretable approach to flood susceptibility mapping. Finally, the findings provide valuable inputs for local authorities, decision-makers, and organized communities to strengthen resilience, reduce vulnerability, and enhance preparedness against future floods, while also supporting the formulation of public policies and the integration of flood susceptibility into land-use planning for sustainable territorial management.

Floods often cause significant damage to transportation infrastructure such as roads, railways, and bridges. This study identifies several topographic, environmental, and hydrological factors (slope, elevation, rainfall, land use and cover, distance from rivers, geology, topographic wetness index, and drainage density) influencing the safety of the railway infrastructure and uses multi-criteria analysis (MCA) alongside an analytical hierarchy process (AHP) to produce flood susceptibility maps within a geographic information system (GIS). The proposed methodology was applied to the catchment area of a railway track in southern Italy that was heavily affected by a destructive flood that occurred in the autumn of 2015. Two susceptibility maps were obtained, one based on static geophysical factors and another including triggering rainfall (dynamic). The results showed that large portions of the railway line are in a very highly susceptible zone. The flood susceptibility maps were found to be in good agreement with the post-disaster flood-induced infrastructural damage recorded along the railway, whilst the official inundation maps from competent authorities fail to supply information about flooding occurring along secondary tributaries and from direct rainfall. The reliable identification of sites susceptible to floods and damage may provide railway and environmental authorities with useful information for preparing disaster management action plans, risk analysis, and targeted infrastructure maintenance/monitoring programs, improving the resilience capacity of the railway network. The proposed approach may offer railway authorities a cost-effective strategy for rapidly screening flood susceptibility at regional/national levels and could also be applied to other types of linear transport infrastructures.

Flooding is the costliest disaster in Canada, yet traditional flood susceptibility modelling is computationally expensive for large-scale applications and often relies on static geospatial features while excluding temporal antecedent conditions. This study uses the Canadian Flood Archive maintained by Natural Resources Canada (NRCan) to develop a large-scale flood susceptibility model for Canada's major drainage areas. The model integrates static terrain derivatives, dynamic climate variables, and semi-static geospatial variables using a hybrid Convolutional Neural Network-Convolutional Long Short-Term Memory (CNN-ConvLSTM) framework. The Canadian Medium Resolution Digital Elevation Model (MRDEM) was used to derive geospatial features, including height above nearest drainage (HAND), Euclidean distance to rivers (EUC), slope, aspect, topographic position index (TPI), and terrain ruggedness index (TRI). Semi-static geospatial variables include land cover (available every 5 years) and the annual normalized difference vegetation index (NDVI), which were temporally matched to each historical flood event. The static and semi-static features were coupled with Daymet meteorological data (precipitation, temperature extremes, snow water equivalent) spanning 1–3-month antecedent windows. The performance of the 2D hybrid CNN-ConvLSTM model will be compared with an Extreme Gradient Boosting (XGBoost) baseline. While XGBoost has performed well in prior research, the hybrid CNN-ConvLSTM is hypothesized to offer superior interpretability of flooding mechanisms. By leveraging the temporal sequence of meteorological drivers, the model captures complex spatiotemporal dependencies that traditional machine learning methods cannot. A preliminary sensitivity analysis of temporal sequence lengths (1-3 months) and resampling ratios (0.1-0.7) showed that the CNN-ConvLSTM architecture achieved the highest predictive accuracy (F1 = 0.89) with a 3-month sequence length and a resampling ratio of 0.5. These initial findings suggest that capturing the full spring snowmelt-to-rainfall cycle is critical for flood susceptibility mapping in Canadian watersheds.

Urbanization processes greatly change urban landscape patterns and the urban thermal environment. Significant multi-scale correlation exists between the land surface temperature (LST) and landscape pattern. Compared with traditional linear regression methods, the regression model based on random forest has the advantages of higher accuracy and better learning ability, and can remove the linear correlation between regression features. Taking Beijing’s metropolitan area as an example, this paper conducted multi-scale relationship analysis between 3D landscape patterns and LST using Pearson Correlation Coefficient (PCC), Multiple Linear Regression and Random Forest Regression (RFR). The results indicated that LST was relatively high in the central area of Beijing, and decreased from the center to the surrounding areas. The interpretation effect of 3D landscape metrics on LST was more obvious than that of the 2D landscape metrics, and 3D landscape diversity and evenness played more important roles than the other metrics in the change of LST. The multi-scale relationship between LST and the landscape pattern was discovered in the fourth ring road of Beijing, the effect of the extent of change on the landscape pattern is greater than that of the grain size change, and the interpretation effect and correlation of landscape metrics on LST increase with the increase in the rectangle size. Impervious surfaces significantly increased the LST, while the impervious surfaces located at low building areas were more likely to increase LST than those located at tall building areas. It seems that increasing the distance between buildings to improve the rate of energy exchange between urban and rural areas can effectively decrease LST. Vegetation and water can effectively reduce LST, but large, clustered and irregularly shaped patches have a better effect on land surface cooling than small and discrete patches. The Coefficients of Rectangle Variation (CORV) power function fitting results of landscape metrics showed that the optimal rectangle size for studying the relationship between the 3D landscape pattern and LST is about 700 m. Our study is useful for future urban planning and provides references to mitigate the daytime urban heat island (UHI) effect.

An integrated strategy for evaluating flood susceptibility combining deep neural networks with biologically inspired meta-heuristic algorithms