Search

Snow Water Equivalent (SWE) is one of the most critical variables in mountainous watersheds and needs to be considered in water resources management plans. As direct measurement of SWE is difficult and empirical equations are highly uncertain, the present study aimed to obtain accurate predictions of SWE using machine learning methods. Five standalone algorithms of tree-based [M5P and random tree (RT)], rule-based [M5Rules (M5R)] and lazy-based learner (IBK and Kstar) and five novel hybrid bagging-based algorithms (BA) with standalone models (i.e., BA-M5P, BA-RT, BA-IBK, BA-Kstar and BA-M5R) were developed. A total of 2550 snow measurements were collected from 62 snow and rain-gauge stations located in 13 mountainous provinces in Iran. Data including ice beneath the snow (IBS), fresh snow depth (FSD), length of snow sample (LSS), snow density (SDN), snow depth (SD) and time of falling (TS) were measured. Based on the Pearson correlation between inputs (IBS, FSD, LSS, SDN, SD and TS) and output (SWE), six different input combinations were constructed. The dataset was separated into two groups (70% and 30% of the data) by a cross-validation technique for model construction (training dataset) and model evaluation (testing dataset), respectively. Different visual and quantitative metrics (e.g., Nash–Sutcliffe efficiency (NSE)) were used for evaluating model accuracy. It was found that SD had the highest correlation with SWE in Iran (r = 0.73). In general, the bootstrap aggregation (i.e., bagging) hybrid machine learning methods (BA-M5P, BA-RT, BA-IBK, BA-Kstar and BA-M5R) increased prediction accuracy when compared to each standalone method. While BA-M5R had the highest prediction accuracy (NSE = 0.83) (considering all six input variables), BA-IBK could predict SWE with high accuracy (NSE = 0.71) using only two input variables (SD and LSS). Our findings demonstrate that SWE can be accurately predicted through a variety of machine learning methods using easily measurable variables and may be useful for applications in other mountainous regions across the globe. [ABSTRACT FROM AUTHOR]

In the present study, three widely used modeling approaches: (1) sediment rating curve (SRC) and optimized OSRC, (2) machine learning models (ML) (random forest (RF) and Dagging-RF (DARF)) and (3) the semi-physically based soil and water assessment tool (SWAT) are applied to predict suspended sediment load (Qs) at the Talar watershed in Iran. Various graphical and quantitative methods were used to evaluate the goodness of fit. Results indicated that the RF model had the best prediction power in the training phase, while the dagging-RF hybrid algorithm outperformed all other models in the validation phase. The OSRC, RF and DA-RF had ‘very good’ performances based on the NSE in the validation phase, SRC showed ‘good’ performance, while the predicted values using SWAT were ‘satisfactory’. Our results suggest that the OSRC and ML models are more suitable for prediction of Qs in study catchments with poor data availability. [ABSTRACT FROM AUTHOR]

Land subsidence (LS), which mainly results from poor watershed management, is a complex and nonlinear phenomenon. In the present study, LS at a country-wide assessment of Iran was mapped by using several geo-environmental conditioning factors (namely, altitude, slope degree and aspect, plan and profile curvature, distance from a river, road or fault, rainfall, geology and land use) into a machine learning algorithm-based artificial neural network (ANN), and a powerful group method of data handling (GMDH). The total dataset includes historical LS and non-LS locations, identified by the interferometric synthetic aperture radar (InSAR). The whole dataset was divided into two subsets at a ratio of 70:30 for training and validating the model, respectively. ANNand GMDH-based LS maps were evaluated using receiver-operator characteristic (ROC) curves. The information gain ratio (IGR) was calculated to determine the relative importance of the conditioning factors. The results showed that all of the considered factors contributed significantly to the LS mapping in Iran, with geology having the strongest impact. According to the ROC curve analysis, both ANN and GMDH-based LS maps were accurate, but the map obtained by the GMDH approach had a higher accuracy than that of ANN. Southwestern, northeastern and some parts of the central region of Iran were shown to be susceptible to LS in the future. According to the GMDH susceptibility map, 10% of Iran exhibits high or very high susceptibility to LS in the future. The provinces of Hamedan and Khouzestan had the highest percentage of areas at risk of LS. According to the InSAR deformation map, 39%, 20%, 25%, 13% and 3% of the investigated areas are subject to a yearly LS of -1 to -2.5, -2.5 to -5, -5 to -7.5, -7.5 to -10 and -10 to -20 cm, respectively. The province of Razavi Khorasan in the northeast of Iran had the largest area (about 3500 km² ) vulnerable to LS occurrence. Based on the LS susceptibility map, the provinces of Ardebil, Kurdistan, West and East Azerbaijan, Sistan and Baluchistan and Kermanshah, although not currently undergoing a high rate of LS, will be at high risk of severe LS in the future. [ABSTRACT FROM AUTHOR]

Labyrinth weirs are utilized to transport a greater discharge during floods in contrast to conventional weirs due to their increased weir crest length. Nevertheless, due to the increased geometric complexity of labyrinth weirs, determination of accurate discharge coefficients and accordingly, head-discharge ratings are quite essential issues in practical application. Hence, as a first step the present study proposes the following eight standalone algorithms: decision table (DT), Kstar, least median square (LMS), M5 prime (M5P), M5 rule (M5R), pace regression (PR), random forest (RF) and sequential minimal optimization (SMO). Then, applying the stacking (ST) algorithm, these standalone models were hybridized to predict the discharge coefficient (Cd) for sharp-crested labyrinth weirs. Potential/effective variables were constructed in the form of several independent dimensionless parameters (i.e., θ, h/W, L/B, L/h, Froude number (Fr), B/W and L/W) to predict Cd as an output. The accuracy of the developed models was examined in terms of different statistical visually based and quantitative-based error measurement criteria. The results illustrate that h/W and B/W parameters have the highest and lowest effect on the Cd prediction, respectively. According to NSE, all developed algorithms provided accurate performances, while ST-Kstar had the highest prediction power. [ABSTRACT FROM AUTHOR]

The free overfall is a simple and widely used device for measuring discharge in open irrigation channels and agricultural research projects. However, the direct measurement of discharge can be difficult and time-consuming with care needed to minimize potential inaccuracies of empirical equations applied to site-specific conditions. Thus, in the present study four standalone algorithms of Isotonic Regression (ISO), Least Median of Square Regression (LMS), M5Prime (M5P) and REPT and four novel hybrid algorithms of rotation forest (ROF) combined with those four standalone models (i.e., ROF-ISO, ROF-LMS, ROF-M5P and ROF-REPT) were applied for the intelligent prediction of discharge per unit width for the free overfall condition in rectangular channels. This was accomplished via six data sets (355 data) collected from the published literature including end depth, Manning's roughness coefficient, channel width, bed slope and unit discharge. The dataset was partitioned in a 70:30 ratio randomly, 70% (248 data) of data used for model development while 30% (107 data) applied for model validation. Also, four different input combinations were constructed to identify the most effective prediction method. Furthermore, results were validated using several visually based (line graph, scatter plot, violin plot and Taylor diagram) and quantitative-based [root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), Willmott's index of agreement, Legates and McCabe coefficient of efficiency (LM)] approaches. Results of the sensitivity analysis revealed that end depth had the highest effect on the results, while channel width was least influential. Results also showed that the best input combination incorporated all four input parameters. According to the results, ROF-REPT had the best performance with RMSE of 0.0035 (m3/s/m), NSE of 0.990, WI of 0.997% and LM of 0.905% followed by ROF-M5P REPT, M5P, ROF-LMS, ISO and LMS. [ABSTRACT FROM AUTHOR]

In the current paper, the efficiency of three new standalone data-mining algorithms [M5 Prime (M5P), Random Forest (RF), M5Rule (M5R)] and six novel hybrid algorithms of bagging (BA-M5P, BA-RF and BA-M5R) and Attribute Selected Classifier (ASC-M5P, ASC-RF and ASC-M5R) for streamflow prediction were assessed and compared with an autoregressive integrated moving average (ARIMA) model as a benchmark. The models used precipitation (P) and streamflow (Q) data from the period 1979–2012 for training and validation (70% and 30% of data, respectively). Different input combinations were prepared using both P and Q with different lag times. The best input combination proved to be that in which all of the the data were used (i.e. R and Q – with lag times). Overall, employing Q with different lag times proved to be more effective than using only P as input for streamflow prediction. Although all models showed very good predictive power, BA-M5P outperformed the other models. [ABSTRACT FROM AUTHOR]

Due to climate change and urban growth, the demand for new freshwater sources, especially groundwater, is increasing in water-deficient countries like Iran. Therefore, this study aimed at groundwater potential mapping (GPM) of the Nahavand Plain, Iran, using an optimized adaptive neuro fuzzy inference system (ANFIS) in a geographic information system, with three metaheuristic optimization algorithms: differential evolution (DE), particle swarm optimization (PSO) and ant colony optimization (ACO). A spatial database was constructed using 273 spring locations and 14 groundwater conditioning factors. The optimization algorithms were evaluated using the receiver operating characteristic (ROC) technique. The ANFIS-DE, ANFIS-PSO and ANFIS-ACO models resulted in accuracy of 0.816, 0.809 and 0.758, respectively; the high and very high potential for groundwater springs covered 26% of the Nahavand Plain. The Root Mean Square Error (RMSE) for the training and validation datasets was lowest for the ANFIS-DE model compared to the other two models; and the ANFIS-PSO model had a higher convergence speed. These results may play an important role in sustainable groundwater management in the Nahavand Plain. [ABSTRACT FROM AUTHOR]

The predictive capability of a new artificial intelligence method, random subspace (RS), for the prediction of suspended sediment load in rivers was compared with commonly used methods: random forest (RF) and two support vector machine (SVM) models using a radial basis function kernel (SVM-RBF) and a normalized polynomial kernel (SVM-NPK). Using river discharge, rainfall and river stage data from the Haraz River, Iran, the results revealed: (a) the RS model provided a superior predictive accuracy (NSE = 0.83) to SVM-RBF (NSE = 0.80), SVM-NPK (NSE = 0.78) and RF (NSE = 0.68), corresponding to very good, good, satisfactory and unsatisfactory accuracies in load prediction; (b) the RBF kernel outperformed the NPK kernel; (c) the predictive capability was most sensitive to gamma and epsilon in SVM models, maximum depth of a tree and the number of features in RF models, classifier type, number of trees and subspace size in RS models; and (d) suspended sediment loads were most closely correlated with river discharge (PCC = 0.76). Overall, the results show that RS models have great potential in data poor watersheds, such as that studied here, to produce strong predictions of suspended load based on monthly records of river discharge, rainfall depth and river stage alone. [ABSTRACT FROM AUTHOR]

While it remains the primary source of safe drinking and irrigation water in northwest Iran's Maku Plain, the region's groundwater is prone to fluoride contamination. Accordingly, modeling techniques to accurately predict groundwater fluoride concentration are required. The current paper advances several novel data mining algorithms including Lazy learners [instance‐based K‐nearest neighbors (IBK); locally weighted learning (LWL); and KStar], a tree‐based algorithm (M5P), and a meta classifier algorithm [regression by discretization (RBD)] to predict groundwater fluoride concentration. Drawing on several groundwater quality variables (e.g., Ca2+,Mg2+,Na+,K+,HCO3−,CO32−,SO42−,andCl− concentrations), measured in each of 143 samples collected between 2004 and 2008, several models predicting groundwater fluoride concentrations were developed. The full dataset was divided into two subsets: 70% for model training (calibration) and 30% for model evaluation (validation). Models were validated using several statistical evaluation criteria and three visual evaluation approaches (i.e., scatter plots, Taylor and Violin diagrams). Although Na+ and Ca2+ showed the greatest positive and negative correlations with fluoride (r = 0.59 and −0.39, respectively), they were insufficient to reliably predict fluoride levels; therefore, other water quality variables, including those weakly correlated with fluoride, should be considered as inputs for fluoride prediction. The IBK model outperformed other models in fluoride contamination prediction, followed by KStar, RBD, M5P, and LWL. The RBD and M5P models were the least accurate in terms of predicting peaks in fluoride concentration values. Results of the current study can be used to support practical and sustainable management of water and groundwater resources. Article impact statement: This work relates unstructured grid generation to the dynamics of finite‐difference groundwater flow models for the first time. [ABSTRACT FROM AUTHOR]

The objective of this study was to experimentally evaluate the difference in the transport of uniform (5.17,10.35,14, 20.7 mm) and graded sediment (mixture of these rounded particles with equal weight proportions) under different unsteady flow hydrographs in a 12 m long, 0.5 m wide and deep glass-walled flume. There was a lag time between fractions and uniform particles, such that peaks of coarser and finer fraction particles occurred before and after the peak of uniform sediment with the same size, respectively. Comparison between uniform particles and fractions in graded sediment showed that the sediment transport rate of fine and coarse fractions was lower and higher than their counterpart uniform particles, respectively. Overall, the uniform particles demonstrated a clockwise hysteresis loop and graded sediment had a counterclockwise hysteresis loop. The mobility of coarser fractions increased during the rising limb of hydrograph, whereas the mobility of finer fractions increased during the falling limb. In general, the mobility of coarse fractions increased and that of fine fractions reduced. Result of transported sediment showed that average particle size collected in traps (Db50) was coarser than bed material (Ds50) on both limbs. The relative transport ratio for uniform and graded sediment is higher and lower than 1, respectively. [ABSTRACT FROM AUTHOR]

The main goal of this study was to optimize an adaptive neuro-fuzzy inference system (ANFIS) using three meta-heuristic optimization algorithms—genetic algorithm (GA), biogeography-based optimization (BBO) and simulated annealing (SA)—to prepare groundwater potential maps. The methodology was applied to the Booshehr plain, Iran. The results of optimized models were compared with ANFIS individually and three bivariate models: frequency ratio (FR), evidential belief function (EBF), and the entropy model. First, 339 wells with groundwater yield higher than 11 m3/h were selected and randomly divided into two groups. In all, 238 wells (70%) were used for training the models and 101 wells (30%) were used for testing and validating the models. Fifteen conditioning factors were selected as input parameters for the modeling. The accuracy of the groundwater potential maps for the study area was determined using root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and standard deviation of error (SD), as well as the area under the receiver operating characteristic (ROC) curve (AUC). Overall, the results demonstrated that ANFIS-GA had the highest prediction capability (AUC = 0.915) for groundwater potential mapping followed by ANFIS-BBO (0.903), entropy (0.862), FR (0.86), ANFIS-SA (0.83), ANFIS (0.82) and EBF (0.80). According to the entropy model, land-use, soil order and rainfall factors had the highest impact on groundwater potential in the study area. The results of this research show that the ANFIS models combined with meta-heuristic optimization algorithms can be a useful decision-making tool for assessment and management of groundwater resources. [ABSTRACT FROM AUTHOR]

• MCDM and machine learning models were compared for flood modelling. • Results show that machine learning is more potential for flood modelling. • Out of machine learning methods, NB outperforms NBT. • Out of MCDM methods, SAW and VIKOR outperform TOPSIS. Floods around the world are having devastating effects on human life and property. In this paper, three Multi-Criteria Decision-Making (MCDM) analysis techniques (VIKOR, TOPSIS and SAW), along with two machine learning methods (NBT and NB), were tested for their ability to model flood susceptibility in one of China's most flood-prone areas, the Ningdu Catchment. Twelve flood conditioning factors were used as input parameters: Normalized Difference Vegetation Index (NDVI), lithology, land use, distance from river, curvature, altitude, Stream Transport Index (STI), Topographic Wetness Index (TWI), Stream Power Index (SPI), soil type, slope and rainfall. The predictive capacity of the models was evaluated and validated using the Area Under the Receiver Operating Characteristic curve (AUC). While all models showed a strong flood prediction capability (AUC > 0.95), the NBT model performed best (AUC = 0.98), suggesting that, among the models studied, the NBT model is a promising tool for the assessment of flood-prone areas and can allow for proper planning and management of flood hazards. [ABSTRACT FROM AUTHOR]

Highlights • Novel data mining models are developed for hourly river sediment prediction. • Various hydrological processes are used as predictors for the sediments load. • SSL resulting from glacierized melting at Andes Catchment is selected as a case study. • The proposed individual and hybrid models are validated against each other. • Bagging-M5P is exhibited a robust and reliable predictive model for SSL prediction. Abstract Suspended sediment has significant effects on reservoir storage capacity, the operation of hydraulic structures and river morphology. Hence, modeling suspended sediment loads (SSL) in rivers contributes to various water resource management and river engineering. An evaluation of stand-alone data mining models (i.e., reduced error pruning tree (REPT), M5P and instance-based learning (IBK)) and hybrid models, (i.e., bagging-M5P, random committee-REPT (RC-REPT) and random subspace-REPT (RS-REPT)) for predicting SSL resulting from glacial melting at an Andean catchment in Chile has been conducted in this study. The best input combinations are constructed based on Pearson correlation coefficient (PCC) of hourly SSL time series data with water discharge (Q), water temperature (T) and electrical conductivity (C) for different time lags. Seventy percent of the available data (one year of hourly data) is used to calibrate the models (dataset training) and the remaining 30% is used for model evaluation (dataset testing). The performances of the models are evaluated using several quantitative and graphical criteria, including coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NSE), percentage of bias (PBIAS), the ratio of RMSE to the standard deviation of observation (RSR), a Taylor diagram and a boxplot. All the models performed well in predicting SSL. However, the Friedman and Wilcoxon signed rank tests revealed that predicted SSL significantly differed for different models except between IBK (or M5P) and REPT. The hybrid models performed better than individual models. The bagging-M5P had the best predictive capability while the REPT had the poorest. [ABSTRACT FROM AUTHOR]

Groundwater vulnerability assessment is a measure of potential groundwater contamination for areas of interest. The main objective of this study is to modify original DRASTIC model using four objective methods, Weights-of-Evidence (WOE), Shannon Entropy (SE), Logistic Model Tree (LMT), and Bootstrap Aggregating (BA) to create a map of groundwater vulnerability for the Sari-Behshahr plain, Iran. The study also investigated impact of addition of eight additional factors (distance to fault, fault density, distance to river, river density, land-use, soil order, geological time scale, and altitude) to improve groundwater vulnerability assessment. A total of 109 nitrate concentration data points were used for modeling and validation purposes. The efficacy of the four methods was evaluated quantitatively using the Area Under the Receiver Operating Characteristic (ROC) Curve (AUC). AUC value for original DRASTIC model without any modification of weights and rates was 0.50. Modification of weights and rates resulted in better performance with AUC values of 0.64, 0.65, 0.75, and 0.81 for BA, SE, LMT, and WOE methods, respectively. This indicates that performance of WOE is the best in assessing groundwater vulnerability for DRASTIC model with 7 factors. The results also show more improvement in predictability of the WOE model by introducing 8 additional factors to the DRASTIC as AUC value increased to 0.91. The most effective contributing factor for ground water vulnerability in the study area is the net recharge. The least effective factors are the impact of vadose zone and hydraulic conductivity. [ABSTRACT FROM AUTHOR]

Floods are one of the most damaging natural hazards causing huge loss of property, infrastructure and lives. Prediction of occurrence of flash flood locations is very difficult due to sudden change in climatic condition and manmade factors. However, prior identification of flood susceptible areas can be done with the help of machine learning techniques for proper timely management of flood hazards. In this study, we tested four decision trees based machine learning models namely Logistic Model Trees (LMT), Reduced Error Pruning Trees (REPT), Naïve Bayes Trees (NBT), and Alternating Decision Trees (ADT) for flash flood susceptibility mapping at the Haraz Watershed in the northern part of Iran. For this, a spatial database was constructed with 201 present and past flood locations and eleven flood-influencing factors namely ground slope, altitude, curvature, Stream Power Index (SPI), Topographic Wetness Index (TWI), land use, rainfall, river density, distance from river, lithology, and Normalized Difference Vegetation Index (NDVI) . Statistical evaluation measures, the Receiver Operating Characteristic (ROC) curve, and Freidman and Wilcoxon signed-rank tests were used to validate and compare the prediction capability of the models. Results show that the ADT model has the highest prediction capability for flash flood susceptibility assessment, followed by the NBT, the LMT, and the REPT, respectively. These techniques have proven successful in quickly determining flood susceptible areas. [ABSTRACT FROM AUTHOR]

• Soil water erosion (SWE) is predicted though three kinds of deep learning algorithms. • Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Long-Short Term Memory (LSTM) are investigated. • All three models had good prediction performance, with RNN being marginally the most superior. • Elevation was the most effective variable on soil water erosion susceptibility. • Maps of SWE susceptibility revealed that almost 40% of the catchment was highly or very highly susceptible to SWE. Accurate assessment of soil water erosion (SWE) susceptibility is critical for reducing land degradation and soil loss, and for mitigating the negative impacts of erosion on ecosystem services, water quality, flooding and infrastructure. Deep learning algorithms have been gaining attention in geoscience due to their high performance and flexibility. However, an understanding of the potential for these algorithms to provide fast, cheap, and accurate predictions of soil erosion susceptibility is lacking. This study provides the first quantification of this potential. Spatial predictions of susceptibility are made using three deep learning algorithms – Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Long-Short Term Memory (LSTM) – for an Iranian catchment that has historically experienced severe water erosion. Through a comparison of their predictive performance and an analysis of the driving geo-environmental factors, the results reveal: (1) elevation was the most effective variable on SWE susceptibility; (2) all three developed models had good prediction performance, with RNN being marginally the most superior; (3) maps of SWE susceptibility revealed that almost 40 % of the catchment was highly or very highly susceptible to SWE and 20 % moderately susceptible, indicating the critical need for soil erosion control in this catchment. Through these algorithms, the soil erosion susceptibility of catchments can potentially be predicted accurately and with ease using readily available data. Thus, the results reveal that these models have great potential for use in data poor catchments, such as the one studied here, especially in developing nations where technical modeling skills and understanding of the erosion processes occurring in the catchment may be lacking. [ABSTRACT FROM AUTHOR]

• New advanced machine learning (ML) models applied for suspended sediment load prediction. • Results compared with traditional ML models as a benchmark. • Manually and feature selection methods applied to explore the best input scenario. • Advanced hybrid Elastic Network model outperform of all developed models. • The outcomes have practical implications in watershed management applications. The distribution and transportation of suspended sediment load (Q ssl) in rivers have a significant effect on the design of hydraulic structures, river morphology, water quality, and aquatic ecosystems. As direct measurement of Q ssl can be costly and time-consuming, reliable estimates are vital for watershed management. In the present study, four standalone models including an Elastic Network (EN), Alternating Model Tree (AM Tree), Reduced Error Pruning Tree (REP Tree) and the Dual Perturb and Combine Tree (DPC Tree), including four hybridized models that combine a standalone model with the Rotation Forest (RF), were developed and evaluated for Q ssl prediction at Talar Watershed, in the northern Iran. Multiple scenarios comprised of antecedent flow discharge (Q w), rainfall (R) and Q ssl , with all constructed manually and automatically using a Wrapper Feature Selection (WFS) to predict the Q ssl at Shirgah hydrometric station from January 1, 2004 to September 22, 2019. The optimal model fitted results were evaluated using multiple graphical and quantitative metrics with the results revealing that the flow discharge is perhaps the best predictor of Q ssl. Based on the Nash-Sutcliffe Efficiency (NSE) metric, the RF-EN (NSE = 0.85), EN (NSE = 0.83), AM Tree (NSE = 0.79), RF-AM Tree (NSE = 0.81) and the RF-REP Tree (NSE = 0.79) models seemed to perform very well, with the REP Tree (NSE = 0.65) and RF-DPC (NSE = 0.71) performed well, while the DPC Tree model (NSE = 0.35) were notably unsatisfactory. The hybridized models, however, captured extreme values more accurately compared with the standalone models. Finally, the model outputs were compared to the well-known optimized ANFIS models with a metaheuristic approach (imperialist competitive algorithm (ICA) and BAT algorithms), and all these results revealed that most the of newly developed models outperformed the ANFIS-ICA and ANFIS-BAT algorithms. The new modelling approaches developed and testing using advanced hybrid Rotation Forest based Elastic Networks in this study have important practical implications for suspended sediment load modeling and applications. [ABSTRACT FROM AUTHOR]

Current study presents the boosting of two base models, including a Heuristic Search Algorithm for finding the k shortest paths (K*) and an alternating model tree (AM Tree) through combining bagging (BA), dagging (DA), and random subspace (RS) hybridize models, to predict monthly suspended sediment load (SSL) in subtropical monsoon climatic regions. In this study, the rainfall (R), streamflow (Q), water level (W), and SSL with various lag-times were fed into the model as inputs. The study found that each input scenario has a unique ability to predict SSL. Different input scenario both manually and automatically derived to improve the modeling performance and reveal impact of the potential variable on the SSL modeling. Model fit outcomes finally were evaluated using graphical and statistical performance metrics. The most effective input scenario is the one that includes all possible input variables. The study also revealed that the hybridized ensemble models outperformed standalone model. The hybrid model of BAK* performed superior to other models. While the standalone model of K* had the lowest prediction ability. Hybridized ensemble models assist in capturing extreme SSL values, especially the maximum SSL value. Results suggest that BA-K* model is superior to other models with R² = 0.80, RMSE = 772.94 Kg/s, PBIAS = 0.06, and NSE = 0.79. The reliability analysis also confirmed that the BA-K* model provided the optimal performance. [ABSTRACT FROM AUTHOR]

Scour depth prediction and its prevention is one of the most important issues in channel and waterway design. However the potential for advanced machine learning (ML) algorithms to provide models of scour depth has yet to be explored. This study provides the first quantification of the predictive power of a range of standalone and hybrid machine learning models. Using previously collected scour depth data from laboratory flume experiments, the performance of five types of recently developed standalone machine learning techniques - the Isotonic Regression (ISOR), Sequential Minimal Optimization (SMO), Iterative Classifier Optimizer (ICO), Locally Weighted learning (LWL) and Least Median of Squares Regression (LMS) - are assessed, along with their hybrid versions with Dagging (DA) and Random Subspace (RS) algorithms. The main findings are five-fold. First, the DA-ICO model had the highest prediction power. Second, the hybrid models had a higher prediction power than standalone models. Third, all algorithms underestimated the maximum scour depth, except DA-ICO which predicted scour depth almost perfectly. Fourth, scour depth was most sensitive to densimetric particle Froude number followed by the non-dimensionalized contraction width, flow depth within the contraction, sediment geometric standard deviation, approach flow velocity and median grain size. Fifth, most of the algorithms performed best when all the input parameters were involved in the building of the model. An important exception was the best performing model that required only four input parameters: densimetric particle Froude number, non-dimensionalized contraction width, flow depth within the contraction and sediment geometric standard deviation. Overall the results revealed that hybrid machine learning algorithms provide more accurate predictions of scour depth than empirical equations and traditional ML-algorithms. In particular, the DA-ICO model not only created the most accurate predictions but also used the fewest easily and readily measured input parameters. Thus this type of model could be of real benefit to practicing engineers required to estimate maximum scour depth when designing in-channel structures. • Hybrid machine learning algorithms were developed for clear-water scour depth modeling in long channel contractions. • Standalone models were hybridized through Dagging (DA) and Random Subspace (RS) algorithms. • Dagging (DA)-Iterative Classifier Optimizer (ICO) hybrid model outperformed its traditional counterparts. • Dimensionless particle Froude number was found as the most effective parameter in scour depth modeling. [ABSTRACT FROM AUTHOR]

Trace element (TE) pollution in groundwater resources is one of the major concerns in both developing and developed countries as it can directly affect human health. Arsenic (As), Barium (Ba), and Rubidium (Rb) can be considered as TEs naturally present in groundwater due to water-rock interactions in Campania Plain (CP) aquifers, in South Italy. Their concentration could be predicted via some readily available input variables using an algorithm like the iterative classifier optimizer (ICO) for regression, and novel hybrid algorithms with additive regression (AR-ICO), attribute selected classifier (ASC-ICO) and bagging (BA-ICO). In this regard, 244 groundwater samples were collected from water wells within the CP and analyzed with respect to the electrical conductivity, pH, major ions and selected TEs. To develop the models, the available dataset was divided randomly into two subsets for model training (70% of the dataset) and evaluation (30% of the dataset), respectively. Based on the correlation coefficient (r), different input variables combinations were constructed to find the most effective one. Each model's performance was evaluated using common statistical and visual metrics. Results indicated that the prediction of As and Ba concentrations strongly depends on HCO 3 −, while Na+ is the most effective variable on Rb prediction. Also, the findings showed that the most powerful predictive models were those that used all the available input variables. According to models' performance evaluation metrics, the hybrid ASC-ICO outperformed other hybrid (BA- and AR-ICO) and standalone (ICO) algorithms to predict As and Ba concentrations, while both hybrid ASC- and BA-ICO models had higher accuracy and lower error than other algorithms for Rb prediction. [Display omitted] • Groundwater trace elements concentration has been predicted using novel machine learning algorithms. • The ensemble ASC-ICO algorithm has proven to give the most reliable results in predicting As, Ba and Rb concentrations. • All the ensemble algorithm showed enhanced performance than canonical machine learning algorithm. • The high concentrations of As, Rb and Ba stated the importance of the volcanic (tuff) water-rock interaction in the area. [ABSTRACT FROM AUTHOR]

Accurate prediction of stable alluvial hydraulic geometry, in which erosion and sedimentation are in equilibrium, is one of the most difficult but critical topics in the field of river engineering. Data mining algorithms have been gaining more attention in this field due to their high performance and flexibility. However, an understanding of the potential for these algorithms to provide fast, cheap, and accurate predictions of hydraulic geometry is lacking. This study provides the first quantification of this potential. Using at-a-station field data, predictions of flow depth, water-surface width and longitudinal water surface slope are made using three standalone data mining techniques -, Instance-based Learning (IBK), KStar, Locally Weighted Learning (LWL) - along with four types of novel hybrid algorithms in which the standalone models are trained with Vote, Attribute Selected Classifier (ASC), Regression by Discretization (RBD), and Cross-validation Parameter Selection (CVPS) algorithms (Vote-IBK, Vote-Kstar, Vote-LWL, ASC-IBK, ASC-Kstar, ASC-LWL, RBD-IBK, RBD-Kstar, RBD-LWL, CVPS-IBK, CVPS-Kstar, CVPS-LWL). Through a comparison of their predictive performance and a sensitivity analysis of the driving variables, the results reveal: (1) Shield stress was the most effective parameter in the prediction of all geometry dimensions; (2) hybrid models had a higher prediction power than standalone data mining models, empirical equations and traditional machine learning algorithms; (3) Vote-Kstar model had the highest performance in predicting depth and width, and ASC-Kstar in estimating slope, each providing very good prediction performance. Through these algorithms, the hydraulic geometry of any river can potentially be predicted accurately and with ease using just a few, readily available flow and channel parameters. Thus, the results reveal that these models have great potential for use in stable channel design in data poor catchments, especially in developing nations where technical modelling skills and understanding of the hydraulic and sediment processes occurring in the river system may be lacking. • The data mining algorithms is provided to predict stable channel geometry. • Predictions were made using standalone and novel hybrid datamining models. • A comparison was made of the predictive power of these data-driven models, and a sensitivity analysis of three driving variables was performed. • The developed hybrid models had a higher performance than empirical and traditional models extended by other researchers. [ABSTRACT FROM AUTHOR]

• 5 new hybrid machine learning models applied to abutment scour depth prediction. • A standalone machine learning model and 2 empirical methods served as benchmarks. • 4 different input scenarios are considered for the machine learning models. • Same input variable has a different effectiveness on different abutment shape. • The machine learning models substantially outperform empirical methods. Complex vortex flow patterns around bridge piers, especially during floods, cause scour process that can result in the failure of foundations. Abutment scour is a complex three-dimensional phenomenon that is difficult to predict especially with traditional formulas obtained using empirical approaches such as regressions. This paper presents a test of a standalone Kstar model with five novel hybrid algorithm of bagging (BA-Kstar), dagging (DA-Kstar), random committee (RC-Kstar), random subspace (RS-Kstar), and weighted instance handler wrapper (WIHW-Kstar) to predict scour depth (d s) for clear water condition. The dataset consists of 99 scour depth data from flume experiments (Dey and Barbhuiya, 2005) using abutment shapes such as vertical, semicircular and 45° wing. Four dimensionless parameter of relative flow depth (h / l), excess abutment Froude number (Fe), relative sediment size (d 50 / l) and relative submergence (d 50 / h) were considered for the prediction of relative scour depth (d s / l). A portion of the dataset was used for the calibration (70%), and the remaining used for model validation. Pearson correlation coefficients helped deciding relevance of the input parameters combination and finally four different combinations of input parameters were used. The performance of the models was assessed visually and with quantitative metrics. Overall, the best input combination for vertical abutment shape is the combination of F e , d 50 / l and h / l , while for semicircular and 45° wing the combination of the Fe and d 50 / l is the most effective input parameter combination. Our results show that incorporating Fe , d 50 / l and h/l lead to higher performance while involving d 50 / h reduced the models prediction power for vertical abutment shape and for semicircular and 45° wing involving h / l and d 50 / h lead to more error. The WIHW-Kstar provided the highest performance in scour depth prediction around vertical abutment shape while RC-Kstar model outperform of other models for scour depth prediction around semicircular and 45° wing. [ABSTRACT FROM AUTHOR]

Due to excessive exploitation, groundwater resources of coastal regions are exposed to seawater intrusion. Therefore, vulnerability assessments are essential for the quantitative and qualitative management of these resources. The GALDIT model is the most widely used approach for coastal aquifer vulnerability assessment, but suffers from subjectivity of the identification of rates and weights. This study aimes at developing a new hybrid framework for improving the accuracy of coastal aquifer vulnerability assessment using various statistical, metaheuristic, and Multi-Attribute Decision Making (MADM) methods to improve the GALDIT model. The Gharesoo-Gorgan Rood coastal aquifer in northern Iran is used as study site. In order to meet this aim, the Differential Evolution (DE) and Biogeography-Based Optimization (BBO) metaheuristic algorithms were employed to optimize the GALDIT weights. In addition, a novel MADM method, named Step-wise Weight Assessment Ratio Analysis (SWARA), and the bivariate statistical method called statistical index (SI) were used to modify the GALDIT ratings. Finally, correlation coefficients between the maps obtained from each method and Total Dissolved Solid (TDS) as an indicator of seawater intrusion were computed to evaluate the models' prediction power. Correlation coefficients of 0.72, 0.75, 0.76 and 0.78 were obtained for the GALDIT SWARA-BBO , GALDIT SI-BBO , GALDIT SWARA-DE and GALDIT SI-DE models, respectively. The results from the GALDIT SI-DE model outperformed all other models at improving the accuracy of the vulnerability assessment. Moreover, the statistical-metaheuristic method yielded more accurate results than SWARA-metaheuristic hybrid models. The vulnerability map of the studied region indicates that the northwestern and western areas are very highly vulnerable. According to GALDIT SI-DE model, 42%, 17%, 18% and 22% of the aquifer areas respectively have a low, medium, high and very high vulnerability to seawater intrusion. The research findings could be applied by regional authorities to manage and protect groundwater resources. Unlabelled Image • A new method for groundwater vulnerability assessment to seawater intrusion is proposed. • The method validation is based on various statistical, metaheuristic, and Multi-Attribute Decision Making methods. • The weights of the method were optimized by using the BBO and DE algorithms. • The GALDIT SI-DE model outperformed among the produced models. [ABSTRACT FROM AUTHOR]

Suspended sediment load modeling through advanced computational algorithms is of major importance and a challenging topic for developing highly accurate hydrological models. To model the suspended sediment load in the Rampur watershed station in the Mahanadi River Basin, Chhattisgarh State, India, unique integrated computational intelligence regression models with an optimizer are proposed in this study. For the first time in the literature, the isotonic regression (ISO) and sequential minimal optimization regression (SMOR) models and their hybrid versions with an iterative classifier optimizer (ICO) are applied for suspended sediment load modeling. The research is based on daily discharge and suspended sediment data collected over a 38-year period (1976–2014). Root mean square error (RMSE), relative root mean square error (RRMSE), coefficient of determination (R2), and Nash–Sutcliffe efficiency (NSE) were employed to evaluate the performance of the standalone ISO and SMOR, as well as the proposed ICO–ISO and ICO–SMOR hybrid models. Ten different scenarios were considered for modeling to investigate the performance of the models using different input combinations. The proposed new models were found to be more reliable than standalone ISO and SMOR models. Results revealed that the performance of the hybrid model was mostly attributable to the basic algorithm for the model development, where both SMOR and ICO–SMOR models were superior to their ISO and ICO–ISO counterparts in terms of accurate computation. Overall, the ICO–SMOR models outperformed the other models in terms of accuracy, with RMSE, RRMSE, R2, and NSE of 5495.1 tons/day, 2.77, 0.90, and 0.86, respectively. The current study's findings support the applicability of the proposed methodology for modeling of suspended sediment load and encourage the use of these methods in alternative hydrological modeling. [ABSTRACT FROM AUTHOR]

• Flood susceptibility map at a scale of Iran derived based on deep learning convolutional neural networks. • 15% of the entire country is highly to very highly susceptible to future flooding events. • 29% and 49% of Iran's cities are located in areas with high and very high susceptibility to future flooding hazards. • Land use factor has the highest effect on the flood occurrences across the country. Iran experiences frequent destructive floods with significant socioeconomic consequences. Quantifying the accurate impacts of such natural hazards, however, is a complicated task. The present study uses a deep learning convolutional neural networks (CNN) algorithm, which is among the newer and most powerful algorithms in big data sets, to develop a flood susceptibility map for Iran. A total of 2769 records were collected from flood locations across the entire country; we divided this data set into two groups using a cross-validation technique. The first group, used as a training data set, was constructed from 70% of the data set and was used for model building. The second group, used as a testing data set, was constructed from the remaining 30% of the records and used for validation. Ten flood conditioning factors, slope, altitude, aspect, plan curvature, profile curvature, rainfall, geology, land use, distance from roads, and distance from rivers, were identified and used in the modeling process. The area under the prediction-rate curve was used for model evaluation, with results showing that the flood susceptibility map has an acceptable accuracy of 75%. The results also indicated that approximately 12% and 3% of the country are highly and very highly susceptible to future flooding events, respectively. Moreover, 29% and 49% of Iran's cities are located in areas with high and very high susceptibility to future flooding hazards. The most effective approaches to flood mitigation are preventing urban expansion and new construction in highly to very highly flood-prone areas as well as watershed management plans and constructing flood control structures according to the topographical characteristics of the catchment. [ABSTRACT FROM AUTHOR]

River water quality assessment is one of the most important tasks to enhance water resources management plans. A water quality index (WQI) considers several water quality variables simultaneously. Traditionally WQI calculations consume time and are often fraught with errors during derivations of sub-indices. In this study, 4 standalone (random forest (RF), M5P, random tree (RT), and reduced error pruning tree (REPT)) and 12 hybrid data-mining algorithms (combinations of standalones with bagging (BA), CV parameter selection (CVPS) and randomizable filtered classification (RFC)) were used to create Iran WQI (IRWQI sc) predictions. Six years (2012 to 2018) of monthly data from two water quality monitoring stations within the Talar catchment were compiled. Using Pearson correlation coefficients, 10 different input combinations were constructed. The data were divided into two groups (ratio 70:30) for model building (training dataset) and model validation (testing dataset) using a 10-fold cross-validation technique. The models were evaluated using several statistical and visual evaluation metrics. Result show that fecal coliform (FC) and total solids (TS) had the greatest and least effect on the prediction of IRWQI sc. The best input combinations varied among the algorithms; generally variables with very low correlations displayed weaker performance. Hybrid algorithms improved the prediction power of several of the standalone models, but not all. Hybrid BA-RT outperformed the other models (R2 = 0.941, RMSE = 2.71, MAE = 1.87, NSE = 0.941, PBIAS = 0.500). PBIAS indicated that all algorithms, with the exceptions of RT, BA-RT and CVPS-REPT, overestimated WQI values. Unlabelled Image • 16 novel hybrid data mining algorithm applied for WQI prediction • BA-RT algorithm outperformed while RFC-RT has the lowest prediction power. • Fecal coliform was the most effective predictor on WQI estimation. • The best input combination is not the same for all models. [ABSTRACT FROM AUTHOR]

• Eight standalone and hybrid data mining algorithms were used to predict bedload transport rate in a laboratory flume. • Flow velocity had the largest impact on the bed load transport rate. • BA-M5P hybrid algorithm had the highest prediction performance. • All data mining models displayed 'very good' prediction performance except the BA-REPT and REPT models. The accurate prediction of bedload transport in gravel-bed rivers remains a significant challenge in river science. However the potential for data mining algorithms to provide models of bedload transport have yet to be explored. This study provides the first quantification of the predictive power of a range of standalone and hybrid data mining models. Using bedload transport data collected in laboratory flume experiments, the performance of four types of recently developed standalone data mining techniques - the M5P, random tree (RT), random forest (RF) and the reduced error pruning tree (REPT) - are assessed, along with four types of hybrid algorithms trained with a Bagging (BA) data mining algorithm (BA-M5P, BA-RF, BA-RT and BA-REPT). The main findings are four-fold. First, the BA-M5P model had the highest prediction power (R 2 = 0.943; RMSE = 0.061 kg m−1 s−1; MAE = 0.040 kg m−1 s−1; NSE = 0.945; PBIAS = −1.60) followed by M5P, BA-RT, RT, BA-RF, RF, BA-REPT, and REPT. All models displayed 'very good' performance except the BA-REPT and REPT model, which were 'satisfactory'. Second, the M5P, BA-RT, and RT models underestimated, and the BA-M5P, BA-RF, RF, BA-REPT and REPT models overestimated, bedload transport rates. Third, flow velocity had the most significant impact on bedload transport rate (PCC = 0.760) followed by shear stress (PCC = 0.709), discharge (PCC = 0.668), bed shear velocity (PCC = 0.663), bed slope (PCC = 0.490), flow depth (PCC = 0.303), median sediment diameter (PCC = 0.247), and relative roughness (PCC = 0.003). Fourth, the maximum depth of tree was the most sensitive operator in decision tree-based algorithms, and batch size, number of execution slots and number of decimal places did not have any impact on model' prediction power. Overall the results revealed that hybrid data mining techniques provide more accurate predictions of bedload transport rate than standalone data mining models. In particular, M5P models, trained with a Bagging data mining algorithm, have great potential to produce robust predictions of bedload transport in gravel-bed rivers. [ABSTRACT FROM AUTHOR]

Groundwater resources constitute the main source of clean fresh water for domestic use and it is essential for food production in the agricultural sector. Groundwater has a vital role for water supply in the Campanian Plain in Italy and hence a future sustainability of the resource is essential for the region. In the current paper novel data mining algorithms including Gaussian Process (GP) were used in a large groundwater quality database to predict nitrate (contaminant) and strontium (potential future increasing) concentrations in groundwater. The results were compared with M5P, random forest (RF) and random tree (RT) algorithms as a benchmark to test the robustness of the modeling process. The dataset includes 246 groundwater quality samples originating from different wells, municipals and agricultural. It was divided for the modeling process into two subgroups by using the 10-fold cross validation technique including 173 samples for model building (training dataset) and 73 samples for model validation (testing dataset). Different water quality variables including T, pH, EC, HCO 3 −, F−, Cl−, SO 4 2−, Na+, K+, Mg2+, and Ca2+ have been used as an input to the models. At first stage, different input combinations have been constructed based on correlation coefficient and thus the optimal combination was chosen for the modeling phase. Different quantitative criteria alongside with visual comparison approach have been used for evaluating the modeling capability. Results revealed that to obtain reliable results also variables with low correlation should be considered as an input to the models together with those variables showing high correlation coefficients. According to the model evaluation criteria, GP algorithm outperforms all the other models in predicting both nitrate and strontium concentrations followed by RF, M5P and RT, respectively. Result also revealed that model's structure together with the accuracy and structure of the data can have a relevant impact on the model's results. Unlabelled Image • Novel Gaussian Process (GP) algorithm was used for NO 3 − and Sr2+ concentration prediction. • GP algorithm had a higher performance followed by RF, M5P and RT. • Different input variables had a different impact on the result. • HCO 3 − and Mg2+ have a significant effect on the prediction of NO 3 − and Sr2+. • Both variables with high and low correlation coefficient should be used as an input. [ABSTRACT FROM AUTHOR]

• Evaporation process is predicted using newly developed soft computing models. • The models are developed using climatic variables-based correlation analysis. • Arid and semi-arid environments are investigated for the prediction process. • A comprehensive comparative analysis is conducted for the inspected case study. • Overall, the ANFIS-GA model demonstrated the excellent and robust prediction model. To model an agriculture process for any region, it is significantly essential to accurately simulate the reference evaporation (ET o) from the available regional meteorological parameters. Nine models, including five data mining algorithms and four adaptive neuro-fuzzy inference systems (ANFISs), were tested for their ability to predict ET o at meteorological stations in Baghdad and Mosul (Iraq). Various weather parameters (e.g., wind speed, sunshine hours, rainfall, maximum and minimum temperature and relative humidity) were recorded and employed as explanatory variables in the models. Pearson correlation analysis showed ET o to have the closest correlation with sunshine hours, maximum and minimum temperatures, and relative humidity. The modeling performance was assessed using the statistical measures of coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NSE), percentage of bias (PBIAS), and the ratio of RMSE to the standard deviation of observations (RSR). Investigations on the modeling accuracy with different input parameter combinations showed that, despite the different structures of the models, no single input combination showed a consistent modeling outcome. Fewer variables were necessary to achieve the same high predictive power for the models developed for the Baghdad station than for those developed for the Mosul station. For both stations, the ANFIS-GA model generally showed the greatest predictive power whereas the random tree algorithm showed the poorest. Moreover, hybrid models showed a higher predictive power than the individual models. [ABSTRACT FROM AUTHOR]

Floods are some of the most dangerous and most frequent natural disasters occurring in the northern region of Iran. Flooding in this area frequently leads to major urban, financial, anthropogenic, and environmental impacts. Therefore, the development of flood susceptibility maps used to identify flood zones in the catchment is necessary for improved flood management and decision making. The main objective of this study was to evaluate the performance of an Evidential Belief Function (EBF) model, both as an individual model and in combination with Logistic Regression (LR) methods, in preparing flood susceptibility maps for the Haraz Catchment in the Mazandaran Province, Iran. The spatial database created consisted of a flood inventory, altitude, slope angle, plan curvature, Topographic Wetness Index (TWI), Stream Power Index (SPI), distance from river, rainfall, geology, land use, and Normalized Difference Vegetation Index (NDVI) for the region. After obtaining the required information from various sources, 151 of 211 recorded flooding points were used for model training and preparation of the flood susceptibility maps. For validation, the results of the models were compared to the 60 remaining flooding points. The Receiver Operating Characteristic (ROC) curve was drawn, and the Area Under the Curve (AUC) was calculated to obtain the accuracy of the flood susceptibility maps prepared through success rates (using training data) and prediction rates (using validation data). The AUC results indicated that the EBF, EBF from LR, EBF-LR (enter), and EBF-LR (stepwise) success rates were 94.61%, 67.94%, 86.45%, and 56.31%, respectively, and the prediction rates were 94.55%, 66.41%, 83.19%, and 52.98%, respectively. The results showed that the EBF model had the highest accuracy in predicting flood susceptibility within the catchment, in which 15% of the total areas were located in high and very high susceptibility classes, and 62% were located in low and very low susceptibility classes. These results can be used for the planning and management of areas vulnerable to floods in order to prevent flood-induced damage; the results may also be useful for natural disaster assessment. [ABSTRACT FROM AUTHOR]

Suspended sediment load (SSL) is one of the essential hydrological processes that affects river engineering sustainability. Sediment has a major influence on the operation of dams and reservoir capacity. This investigation is aimed at exploring a new version of machine learning models (i.e. data mining), including M5P, attribute selected classifier (AS M5P), M5Rule (M5R), and K Star (KS) models for SSL prediction at the Trenton meteorological station on the Delaware River, USA. Different input scenarios were examined based on the river flow discharge and sediment load database. The performance of the applied data mining models was evaluated using various statistical metrics and graphical presentation. Among the applied data mining models, the M5P model gave a superior prediction result. The current and one-day lead time river flow and sediment load were the influential predictors for one-day-ahead SSL prediction. Overall, the applied data mining models achieved excellent predictions of the SSL process. [ABSTRACT FROM AUTHOR]

In this study, the main goal is to compare the predictive capability of Support Vector Machines (SVM) with four Bayesian algorithms namely Naïve Bayes Tree (NBT), Bayes network (BN), Naïve Bayes (NB), Decision Table Naïve Bayes (DTNB) for identifying landslide susceptibility zones in Pauri Garhwal district (India). First, landslide inventory map was built using 1295 historical landslide data, then in total sixteen influencing factors were selected and tested for landslide susceptibility modelling. Performance of the model was evaluated and compared using Statistical based index methods, Area under the Receiver Operating Characteristic (ROC) curve named AUC, and Chi-square method. Analysis results show that that the SVM has the highest prediction capability, followed by the NBT, DTNBT, BN and NB, respectively. Thus, this study confirms that the SVM is one of the benchmark models for the assessment of susceptibility of landslides. [ABSTRACT FROM AUTHOR]

• Groundwater spatial modeling using spring data set. • Presenting new hybrid models of ANFIS-TLBO and ANFIS-BBO for groundwater potential mapping and its zoning. • Predictive capability of spring affecting factors using correlation attribute evaluation method. • Correlation between springs and affecting factors using the PCF algorithm. • ANFIS-TLBO shows the better results in term of AUC value. Groundwater is one of the most beneficial natural resources worldwide. The main aim of the present study is the mapping of groundwater potential of the Zhangjiamao area in China using two novel hybrid data mining techniques that involve the adaptive neuro-fuzzy inference system (ANFIS) ensembled with teaching-learning-based optimization (TLBO) and biogeography-based optimization (BBO). First, 93 spring locations were identified in the study area, which were randomly divided in the ratio of 70:30 to be used for building models and validation, respectively. A total of 16 spring affecting factors viz., slope aspect, altitude, slope angle, plan curvature, profile curvature, curvature, sediment transport index, stream power index, topographic wetness index, distance to roads, distance to rivers, rainfall, lithology, soil, NDVI, and land use/land cover, were used as input variables. The probability certainty factor (PCF) method was applied for the correlation analysis between spring occurrences and the affecting factors. Subsequently, the ANFIS-TLBO and ANFIS-BBO models were generated to produce groundwater spring potential maps (GSPMs). Finally, the GSPMs were evaluated by using the area under the receiver operating characteristic (AUROC). The results showed that the AUROC values for the ANFIS-TLBO model based on the training and validation datasets were 0.866 and 0.905, respectively. In contrast, the AUROC values for the ANFIS-BBO model based on the training and validation datasets were 0.861 and 0.887, respectively. The results of ANFIS-TLBO showed that 25.66% of the study area could be classified into the high and very high groundwater spring potential classes as compared to the ANFIS-BBO model (30.52%). The results of the present study can be useful for groundwater management. [ABSTRACT FROM AUTHOR]

Infiltration is the process by which water enters the soil, and it plays a significant role in the hydrologic cycle. Direct measurement of infiltration is time consuming; however, empirical and physical models are inaccurate. In this study, we compared the results of a deep learning-based convolutional neural network (CNN) algorithm with those of models based on standalone support vector regression (SVR) and group method of data handling (GMDH) algorithms. We also tested a hybridized SVR and GMDH-based model enhanced by three metaheuristic algorithms: gray wolf optimization (GWO), bat algorithm (BA), and particle swarm optimization (PSO). Measured variables including measurement time; sand, clay and silt contents; bulk density; and soil moisture content were used as model inputs to predict cumulative infiltration as an output. Finally, models were evaluated using the Pearson correlation coefficient, root mean square error, mean absolute error, Nash–Sutcliffe efficiency, percentage of bias, and relative error. Weighting of the model input parameters/variables demonstrated that time was the most effective variable for predicting cumulative infiltration, whereas the most powerful parameter input combination required all seven variables. The prediction model evaluation results showed that the hybridized models improved the predictive capability of the standalone models. [ABSTRACT FROM AUTHOR]

• Novel machine learning techniques are recommended for sediment transport modeling. • M5P, REPT, RF and RT are hybridized through WIHW and ROF techniques. • Linked to the effective parameters involved different scenarios are considered. • WIHW-RT is found superior to its counterparts. • ML results are in consistent with hydrological consideration of the problem. Sediment transport modeling has been known as an essential issue and challenging task in water resources and environmental engineering. In order to minimize the adverse impacts of the continues sediment deposition that is known as a main source of pollution in the urban area, the self-cleansing method is widely utilized for designing the sewer pipes to create a condition to keep the bottom of channel clean from sedimentation. In the present study, an extensive data range is utilized for modeling the sediment transport in non-deposition with clean bed condition. Regarding the effective parameters involved, four different scenarios are considered for the modeling. To this end, four standalone methods including the M5P, reduced error pruning tree (REPT), random forest (RF) and random tree (RT) and two hybrid models based on rotation forest (ROF) and weighted instances handler wrapper (WIHW) techniques are developed and result compared with three empirical equations. Based on the results, the hybrid WIHW-RT and WIHW-RF models provide better performance in particle Froude number estimation in comparison to other standalone and hybrid models. Performances of the most of the models are found accurate except RT and REPT standalone models. The outcomes revealed that the empirical models have considerable overestimation. Generally, hybrid data mining methods yield more precise estimations of sediment transport in contrast to the regression equations and standalone models. Particularly, both WIHW-RT and WIHW-RF models provide almost the same performances however, as WIHW-RT can better capture the extreme particle Froude number values, it slightly outperforms WIHW-RF. Promising findings of the current study may encourage the implementation of the recommended approaches in alternative hydrological problems. [ABSTRACT FROM AUTHOR]

Flooding is one of the most significant environmental challenges and can easily cause fatal incidents and economic losses. Flood reduction is costly and time-consuming task; so it is necessary to accurately detect flood susceptible areas. This work presents an effective flood susceptibility mapping framework by involving an adaptive neuro-fuzzy inference system (ANFIS) with two metaheuristic methods of biogeography based optimization (BBO) and imperialistic competitive algorithm (ICA). A total of 13 flood influencing factors, including slope, altitude, aspect, curvature, topographic wetness index, stream power index, sediment transport index, distance to river, landuse, normalized difference vegetation index, lithology, rainfall and soil type, were used in the proposed framework for spatial modeling and Dingnan County in China was selected for the application of the proposed methods due to data availability. There are 115 flood occurrences in the study area which were randomly separated into training (70% of the total) and verification (30%) sets. To perform the proposed framework, the step-wise weight assessment ratio analysis algorithm is first used to evaluate the correlation between influencing factors and floods. Then, two ensemble methods of ANFIS-BBO and ANFIS-ICA are constructed for spatial prediction and producing flood susceptibility maps. Finally, these resultant maps are assessed in terms of several statistical and error measures, including receiver operating characteristic (ROC) curve and area under the ROC curve (AUC), root-mean-square error (RMSE). The experimental results demonstrated that the two ensemble methods were more effective than ANFIS in the study area. For instance, the predictive AUC values of 0.8407, 0.9045 and 0.9044 were achieved by the methods of ANFIS, ANFIS-BBO and ANFIS-ICA, respectively. Moreover, the RMSE values for ANFIS, ANFIS-BBO and ANFIS-ICA using the verification set were 0.3100, 0.2730 and 0.2700, respectively. In addition, as regards ANFIS-BBO and ANFIS-ICA, a total areas of 39.30% and 35.39% were classified as highly susceptible to flooding. Therefore, the proposed ensemble framework can be used for flood susceptibility mapping in other sites with similar geo-environmental characteristics for taking measures to manage and prevent flood damages. Image 1 • Prediction power of two novel ensemble methods for flood susceptibility mapping. • The proposed ensemble methods can improve the prediction performance of ANFIS. • The proposed methods can accurately produce flood susceptibility maps for mitigation and management. [ABSTRACT FROM AUTHOR]

In this study, we introduced novel hybrid of evidence believe function (EBF) with logistic regression (EBF-LR) and logistic model tree (EBF-LMT) for landslide susceptibility modelling. Fourteen conditioning factors were selected, including slope aspect, elevation, slope angle, profile curvature, plan curvature, topographic wetness index (TWI), stream sediment transport index (STI), stream power index (SPI), distance to rivers, distance to faults, distance to roads, lithology, normalized difference vegetation index (NDVI), and land use. The importance of factors was assessed using correlation attribute evaluation method. Finally, the performance of three models was evaluated using the area under the curve (AUC). The validation process indicated that the EBF-LMT model acquired the highest AUC for the training (84.7%) and validation (76.5%) datasets, followed by EBF-LR and EBF models. Our result also confirmed that combination of a decision tree-logistic regression-based algorithm with a bivariate statistical model lead to enhance the prediction power of individual landslide models. [ABSTRACT FROM AUTHOR]

This study presented the prediction capability of three methods including the frequency ratio (FR), fuzzy gamma (FG) and landslide index method (LIM) to produce landslide-prone areas in the Sari-Kiasar watershed, Mazandaran Province of Iran. In the first step, 105 landslide locations were selected and were randomly divided into two groups of 75% (78 locations) and 25% (27 locations) as training and testing datasets. Then the 17 landslide conditioning factors including land use/land cover, Differential Vegetation Index (DVI), lithology and distance from faults, elevation, slope aspect, slope angle, tangential curvature, profile curvature and plane curvature, distance from drainage, rainfall, Stream Power Index, Sediment Transport Index and temperature, and distance from road, density of settlement were considered for the proposed modelling approach. Finally, by applying the training dataset, three landslide susceptibility maps were constructed by using the FR, FG and LIM methods. The prediction capability of the performed model was evaluated by the area under the receiver operating curve or AUC for both training (success rate) and testing (prediction rate) datasets. The results showed that the AUC for success rate of FR, FG and LIM models was 82.04%, 81.08% and 73.61% and for prediction rate was 82.72%, 79.09% and 65.45%, respectively. The results showed that the FR model has a higher prediction accuracy than the FG and LIM methods. This study revealed that the most important factors in landslide occurrence are rainfall, slope and vegetation. The result of the present study can be possibly useful for land use planning and watershed management. [ABSTRACT FROM AUTHOR]

Abstract Landslides represent a part of the cascade of geological hazards in a wide range of geo-environments. In this study, we aim to investigate and compare the performance of two state-of-the-art machine learning models, i.e., decision tree (DT) and random forest (RF) approaches to model the massive rainfall-triggered landslide occurrences in the Izu-Oshima Volcanic Island, Japan at a regional scale. At first, a landslide inventory map is prepared consisting of 44 landslide polygons (10,444 pixels) from aerial photo-interpretation and field surveys. To estimate the robustness of the models, we randomly adapted two different samples (S1 and S2), comprising of both positive and negative cells (70% of total landslides - 7293 pixels) for training and remaining (30%–3151 pixels) for validation. Twelve causative factors including altitude, slope angle, slope aspect, plan curvature, total curvature, compound topographic index, stream power index, distance to drainage network, drainage density, distance to geological boundaries, lithology and cumulative rainfall were selected as predictors to implement the landslide susceptibility model. The area under the receiver operating characteristics (ROC) curves (AUC) and other statistical signifiers were used to verify the model accuracies. The result shows that the DT and RF models achieved remarkable predictive performance (AUC > 0.9), producing near accurate susceptibility maps. The overall efficiency of RF (AUC = 0.956) is found significantly higher than the DT (AUC = 0.928) results. Additionally, we noticed that the performance of RF for modeling landslide susceptibility is very robust even though the training and validation samples are altered. Considering the performances, we suggest that both RF and DT models can be used in other similar non-eruption-related landslide studies in the tephra-deposited rich volcanoes, as they are capable of rapidly generating accurate and stable LSM maps for risk mitigation, management practices, and decision-making. Moreover, the RF-based model is promising and enough to be recommended as a method to map regional landslide susceptibility. Graphical abstract Unlabelled Image Highlights • Decision tree and random forest models applied to map landslide-prone areas in a volcanic Island. • Two sample set (S1, and S2) for computing the robustness of the model • LSM maps were compared using different assessment principles. • Random forest performs better on both samples with AUC > 0.9. [ABSTRACT FROM AUTHOR]

A new artificial intelligence (AI) model, called Bagging-LMT - a combination of bagging ensemble and Logistic Model Tree (LMT) - is introduced for mapping flood susceptibility. A spatial database was generated for the Haraz watershed, northern Iran, that included a flood inventory map and eleven flood conditioning factors based on the Information Gain Ratio (IGR). The model was evaluated using precision, sensitivity, specificity, accuracy, Root Mean Square Error, Mean Absolute Error, Kappa and area under the receiver operating characteristic curve criteria. The model was also compared with four state-of-the-art benchmark soft computing models, including LMT, logistic regression, Bayesian logistic regression, and random forest. Results revealed that the proposed model outperformed all these models and indicate that the proposed model can be used for sustainable management of flood-prone areas. [ABSTRACT FROM AUTHOR]

Landslides are a geological hazard that can pose a serious threat to human health and the environment of highlands or mountain slopes. Landslide susceptibility mapping is an essential tool for predicting and mitigating landslides. This study aimed to investigate the application of deep learning algorithms based on convolutional neural networks (CNNs) with metaheuristic optimization algorithms, namely the grey wolf optimizer (GWO) and imperialist competitive algorithm (ICA), to landslide susceptibility mapping. The study area was Icheon City, South Korea, for which an accurate landslide inventory dataset was available. The landslide inventory map was prepared and randomly divided into datasets of 70% for training and 30% for validation. Additionally, 18 landslide-related factors, including geo-environmental and topo-hydrological factors, were considered as predictive variables. The models were compared using area under the curve (AUC) values in receiver operating characteristic (ROC) curve analysis. The validation results showed that optimized models based on CNN-GWO (AUC = 0.876, RMSE = 0.08) and CNN-ICA (AUC = 0.852, RMSE = 0.09) outperformed the standalone CNN model (AUC = 0.847, RMSE = 0.12). Nevertheless, the CNN model outperformed previous research that used a machine learning algorithm alone. Thus, the deep learning algorithm with optimization algorithms proposed in this study can generate more suitable models for landslide susceptibility mapping in the study area due to its improved accuracy. • LS maps are generated using CNN method with metaheuristic optimization algorithms. • The CNN with metaheuristic optimization algorithms outperformed the standalone CNN. • The CNN-GWO shows better accuracy than the CNN-ICA in term of AUC value. [ABSTRACT FROM AUTHOR]

• Flood susceptibility maps generated using different models. • Prediction power of standalone and optimized SVR models compared. • GOA and PSO algorithms applied for optimization of SVR model. • Elevation, lithology and aspect were the most effective flood conditioning factors. Flood spatial susceptibility prediction is the first essential step in developing flood mitigation strategies and reducing flood damage. Flood occurrence is a complex process that is not easily predicted through simple methods. This study describes optimization of support vector regression (SVR) using meta-optimization algorithms including the grasshopper optimization algorithm (GOA) and particle swarm optimization (PSO) for flood modeling at Qazvin Plain, Iran. Geospatial data including nine readily available geo-environmental flood conditioning factors (i.e., ground slope, aspect, elevation, planform curvature, profile curvature, proximity to a river, land use, lithology and rainfall) were derived. The information gain ratio (IGR) method was used to determine the relative importance of input variables. A historical flood inventory map for 43 locations was created from existing reports. The geospatial data and historical flood levels were used to construct the training and testing datasets. Then, the training dataset was used to generate flood-susceptibility maps using the optimized SVR model with the GOA and PSO algorithms. Finally, the predictive accuracy of the models was quantified using the statistical measures of root mean square error (RMSE) , mean absolute error (MAE) , and area under the receiver operating characteristic (ROC) curve (AUC). Although both the GOA and PSO algorithms improved SVR performance, the SVR-GOA model performed best (AUC = 0.959, RMSE = 0.31 and MSE = 0.098), followed by the SVR-PSO model (AUC = 0.959, RMSE = 0.33 and MSE = 0.11) and standalone SVR model (AUC = 0.87, RMSE = 0.35 and MSE = 0.12). Elevation, lithology and aspect had the highest IGR values and were identified as the most effective predictors of flood susceptibility. [ABSTRACT FROM AUTHOR]

Reliable flash flood susceptibility maps are a vital tool for land planners and emergency management officials for early flood warning and mitigation. We have developed a new ensemble learning model that predicts flash flood susceptibility at Haraz, Iran. The new model couples a Bayesian Belief Network (BBN) model with an extreme learning machine (ELM) and backpropagation (BP) structure optimized by a genetic algorithm (GA) named GA-BN-NN model. We applied the support vector machine (SVM) technique to a database of 194 flood locations with ten conditioning factors. An artificial neural network (ANN) algorithm with a multi-layer perceptron function, MLP-BP, optimized by a genetic algorithm, GA-MLP, and a shuffled frog-leaping algorithm, SFLA-MLP, were used as benchmark models for assessing the power prediction of the proposed model. Statistical measures, including sensitivity, specificity, accuracy, F1-measure and Jaccard coefficient, and root mean square error, were used to evaluate the goodness-of-fit and prediction accuracy, respectively, of the training and testing datasets. We found that all ten factors are positively correlated with flood occurrence, but slope angle has the highest average merit (AM = 9.7) and thus contributes most to the occurrence of flooding. Results indicate that the GA-BN-NN model has the highest goodness-of-fit and prediction accuracy (AUC = 0.966) and hence outperforms other ensemble learning models that we tested — the SFLA-MLP, MLP-BP, and GA-MLP models. We thus conclude that the proposed model is a promising technique for managing risk in flood-prone areas around the world. • A new deep learning model, E(T)-ELM-DBN-BP, was explored for flash flood susceptibility mapping. • The LSSVM technique was applied to determine the most important flash flood conditioning factors. • The MLP was optimized by BP, GA and SFLA algorithms as benchmark models to validity of the new model. • The E(T)-ELM-DBN-BP model outperformed the E(T)-MLP-BP, E(W)-MLP-GA and E(W)-MLP-SFLA models. • The new proposed model was offered as a promising and trustworthy tool for flash flood mitigation. [ABSTRACT FROM AUTHOR]

Mapping flood-prone areas is a key activity in flood disaster management. In this paper, we propose a new flood susceptibility mapping technique. We employ new ensemble models based on bagging as a meta-classifier and K-Nearest Neighbor (KNN) coarse, cosine, cubic, and weighted base classifiers to spatially forecast flooding in the Haraz watershed in northern Iran. We identified flood-prone areas using data from Sentinel-1 sensor. We then selected 10 conditioning factors to spatially predict floods and assess their predictive power using the Relief Attribute Evaluation (RFAE) method. Model validation was performed using two statistical error indices and the area under the curve (AUC). Our results show that the Bagging–Cubic–KNN ensemble model outperformed other ensemble models. It decreased the overfitting and variance problems in the training dataset and enhanced the prediction accuracy of the Cubic–KNN model (AUC=0.660). We therefore recommend that the Bagging–Cubic–KNN model be more widely applied for the sustainable management of flood-prone areas. [ABSTRACT FROM AUTHOR]

We used a novel hybrid functional machine learning algorithm to predict the spatial distribution of landslides in the Sarkhoon watershed, Iran. We developed a new ensemble model which is a combination of a functional algorithm, stochastic gradient descent (SGD) and an AdaBoost (AB) Meta classifier namely ABSGD model to predict the landslides. The model incorporates 20 landslide conditioning factors, which we ranked using the least-square support vector machine (LSSVM) technique. For the modeling, we considered 98 landslide locations, of which 70% (79) were used for training and 30% (19) for validation processes. Model validation was performed using sensitivity, specificity, accuracy, the root mean square error (RMSE) and the area under the receiver operatic characteristic (AUC) curve. We also used soft computing benchmark models, including SGD, logistic regression (LR), logistic model tree (LMT) and functional tree (FT) algorithms for model validation and comparison. The selected conditioning factors were significant in landslide occurrence but distance to road was found to be the most important factor. The ABSGD model (AUC= 0.860) outperformed the LR (0.797), SGD (0.776), LMT (0.740) and FT (0.734) models. Our results confirm that the combined use of a functional algorithm and a Meta classifier prevents over-fitting, reduces noise and enhances the power prediction of the individual SGD algorithm for the spatial prediction of landslides. [ABSTRACT FROM AUTHOR]

The main objective of this research was to introduce a novel machine learning algorithm of alternating decision tree (ADTree) based on the multiboost (MB), bagging (BA), rotation forest (RF) and random subspace (RS) ensemble algorithms under two scenarios of different sample sizes and raster resolutions for spatial prediction of shallow landslides around Bijar City, Kurdistan Province, Iran. The evaluation of modeling process was checked by some statistical measures and area under the receiver operating characteristic curve (AUROC). Results show that, for combination of sample sizes of 60%/40% and 70%/30% with a raster resolution of 10 m, the RS model, while, for 80%/20% and 90%/10% with a raster resolution of 20 m, the MB model obtained a high goodness-of-fit and prediction accuracy. The RS-ADTree and MB-ADTree ensemble models outperformed the ADTree model in two scenarios. Overall, MB-ADTree in sample size of 80%/20% with a resolution of 20 m (area under the curve (AUC) = 0.942) and sample size of 60%/40% with a resolution of 10 m (AUC = 0.845) had the highest and lowest prediction accuracy, respectively. The findings confirm that the newly proposed models are very promising alternative tools to assist planners and decision makers in the task of managing landslide prone areas. [ABSTRACT FROM AUTHOR]

Adaptive neuro-fuzzy inference system (ANFIS) includes two novel GIS-based ensemble artificial intelligence approaches called imperialistic competitive algorithm (ICA) and firefly algorithm (FA). This combination could result in ANFIS-ICA and ANFIS-FA models, which were applied to flood spatial modelling and its mapping in the Haraz watershed in Northern Province of Mazandaran, Iran. Ten influential factors including slope angle, elevation, stream power index (SPI), curvature, topographic wetness index (TWI), lithology, rainfall, land use, stream density, and the distance to river were selected for flood modelling. The validity of the models was assessed using statistical error-indices (RMSE and MSE), statistical tests (Friedman and Wilcoxon signed-rank tests), and the area under the curve (AUC) of success. The prediction accuracy of the models was compared to some new state-of-the-art sophisticated machine learning techniques that had previously been successfully tested in the study area. The results confirmed the goodness of fit and appropriate prediction accuracy of the two ensemble models. However, the ANFIS-ICA model (AUC = 0.947) had a better performance in comparison to the Bagging-LMT (AUC = 0.940), BLR (AUC = 0.936), LMT (AUC = 0.934), ANFIS-FA (AUC = 0.917), LR (AUC = 0.885) and RF (AUC = 0.806) models. Therefore, the ANFIS-ICA model can be introduced as a promising method for the sustainable management of flood-prone areas. [ABSTRACT FROM AUTHOR]