Risk management has reduced vulnerability to floods and droughts globally^1,2, yet their impacts are still increasing³. An improved understanding of the causes of changing impacts is therefore needed, but has been hampered by a lack of empirical data^4,5. On the basis of a global dataset of 45 pairs of events that occurred within the same area, we show that risk management generally reduces the impacts of floods and droughts but faces difficulties in reducing the impacts of unprecedented events of a magnitude not previously experienced. If the second event was much more hazardous than the first, its impact was almost always higher. This is because management was not designed to deal with such extreme events: for example, they exceeded the design levels of levees and reservoirs. In two success stories, the impact of the second, more hazardous, event was lower, as a result of improved risk management governance and high investment in integrated management. The observed difficulty of managing unprecedented events is alarming, given that more extreme hydrological events are projected owing to climate change³.

Concerning the significant increase in the negative effects of flash-floods worldwide, the main goal of this research is to evaluate the power of the Analytical Hierarchy Process (AHP), fi (kNN), K-Star (KS) algorithms and their ensembles in flash-flood susceptibility mapping. To train the two stand-alone models and their ensembles, for the first stage, the areas affected in the past by torrential phenomena are identified using remote sensing techniques. Approximately 70% of these areas are used as a training data set along with 10 flash-flood predictors. It should be remarked that the remote sensing techniques play a crucial role in obtaining eight out of 10 flash-flood conditioning factors. The predictive capability of predictors is evaluated through the Information Gain Ratio (IGR) method. As expected, the slope angle results in the factor with the highest predictive capability. The application of the AHP model implies the construction of ten pair-wise comparison matrices for calculating the normalized weights of each flash-flood predictor. The computed weights are used as input data in kNN–AHP and KS–AHP ensemble models for calculating the Flash-Flood Potential Index (FFPI). The FFPI also is determined through kNN and KS stand-alone models. The performance of the models is evaluated using statistical metrics (i.e., sensitivity, specificity and accuracy) while the validation of the results is done by constructing the Receiver Operating Characteristics (ROC) Curve and Area Under Curve (AUC) values and by calculating the density of torrential pixels within FFPI classes. Overall, the best performance is obtained by the kNN–AHP ensemble model.

For effective management of water quantity and quality, it is absolutely essential to estimate the pollution level of the existing surface water. This case study aims to evaluate the performance of twelve machine learning (ML) models, including five boosting-based algorithms (adaptive boosting, gradient boosting, histogram-based gradient boosting, light gradient boosting, and extreme gradient boosting), three decision tree-based algorithms (decision tree, extra trees, and random forest), and four ANN-based algorithms (multilayer perceptron, radial basis function, deep feed-forward neural network, and convolutional neural network), in estimating the surface water quality of the La Buong River in Vietnam. Water quality data at four monitoring stations alongside the La Buong River for the period 2010–2017 were utilized to calculate the water quality index (WQI). Prediction performance of the ML models was evaluated by using two efficiency statistics (i.e., R² and RMSE). The results indicated that all twelve ML models have good performance in predicting the WQI but that extreme gradient boosting (XGBoost) has the best performance with the highest accuracy (R² = 0.989 and RMSE = 0.107). The findings strengthen the argument that ML models, especially XGBoost, may be employed for WQI prediction with a high level of accuracy, which will further improve water quality management.

The impact of climate and land-use changes on hydrological processes and sediment yield is investigated in the Be River catchment, Vietnam, using the Soil and Water Assessment Tool (SWAT) hydrological model. The sensitivity analysis, model calibration and validation indicated that the SWAT model could reasonably simulate the hydrology and sediment yield in the catchment. From this, the responses of the hydrology and sediment to climate change and land-use changes were considered. The results indicate that deforestation had increased the annual flow (by 1.2%) and sediment load (by 11.3%), and that climate change had also significantly increased the annual streamflow (by 26.3%) and sediment load (by 31.7%). Under the impact of coupled climate and land-use changes, the annual streamflow and sediment load increased by 28.0% and 46.4%, respectively. In general, during the 1978–2000 period, climate change influenced the hydrological processes in the Be River catchment more strongly than the land-use change.

Hydrological models play vital roles in management of water resources. However, the calibration of the hydrological models is a large challenge because of the uncertainty involved in the large number of parameters. In this study, four uncertainty analysis methods, including Generalized Likelihood Uncertainty Estimation (GLUE), Parameter Solution (ParaSol), Particle Swarm Optimization (PSO), and Sequential Uncertainty Fitting (SUFI-2), were employed to perform parameter uncertainty analysis of streamflow simulation in the Srepok River Catchment by using the Soil and Water Assessment Tool (SWAT) model. The four methods were compared in terms of the model prediction uncertainty, the model performance, and the computational efficiency. The results showed that the SUFI-2 method has the advantages in the model calibration and uncertainty analysis. This technique could be run with the smallest of simulation runs to achieve good prediction uncertainty bands and model performance. This technique could be run with the smallest of simulation runs to achieve good prediction uncertainty bands and model performance