Integrated dynamic framework for predicting urban flooding and providing early warning

Streamflow forecasting is a vital task for hydrology and water resources engineering, and the different artificial intelligence (AI) approaches have been employed for this purposes until now. Additionally, the forecasting accuracy and uncertainty estimation are the meaningful assignments that need to be recognized. The addressed research investigates the potential of novel ensemble approach, Bayesian model averaging (BMA), in streamflow forecasting using daily time series data from two stations (i.e., Hongcheon and Jucheon), South Korea. Six categories (i.e., M1–M6) of input combination using different antecedent times were employed for streamflow forecasting. The outcomes of BMA model were compared with those of multivariate adaptive regression spline (MARS), M5 model tree (M5Tree), and Kernel extreme learning machines (KELM) models considering four assessment indexes, root mean square error (RMSE), Nash-Sutcliffe efficiency (NSE), correlation coefficient (R), and mean absolute error (MAE). The results revealed the superior accuracy of BMA model over three machine learning models in daily streamflow forecasting. Considering RMSE values among the best models during testing phase, the best BMA model (i.e., BMA2) enhanced the forecasting accuracy of MARS1, M5Tree4, and KELM3 models by 5.2%, 5.8%, and 3.4% in Hongcheon station. Additionally, the best BMA model (i.e., BMA1) improved the forecasting accuracy of MARS1, M5Tree1, and KELM1 models by 6.7%, 9.5%, and 3.7% in Jucheon station. In addition, the best BMA models in both stations allowed the uncertainty estimation, and produced higher uncertainty of peak flows compared to that of low flows. As one of the most robust and effective tools, therefore, the BMA model can be successfully employed for streamflow forecasting with different antecedent times.

A multiple crop model ensemble for improving broad-scale yield prediction using Bayesian model averaging

Purpose Low-dose radiation risks are generally extrapolated from groups exposed at high levels of dose. Measurement error substantially alters dose-response shape and hence extrapolated risk. Much attention has been paid to methods of dealing with shared errors, common in many datasets, in particular using Bayesian model averaging (BMA) methods. Materials and methods We test two types of BMA model using simulated data. The first we term the quasi-two-dimensional Monte Carlo with BMA (quasi-2DMC + BMA) method, similar to the BMA method proposed by Hoeting et al. but distinct from the 2DMC + BMA method of Kwon et al. The second we term the marginal-quasi-2DMC + BMA method, which uses a more complicated marginal calculation, and may be closer to the 2DMC + BMA method of Kwon et al. Results Assuming a true linear model of dose response, the coverage probabilities for the linear coefficient are 90–95%, for the quasi-2DMC + BMA method, but lower than this, 52–60%, for the marginal-quasi-2DMC + BMA method. Assuming a true linear-quadratic model the coverage probabilities of both linear/quadratic excess relative risk (ERR) coefficients for quasi-2DMC + BMA are too low, and when shared Berkson error is large (50%) the probabilities do not exceed 5%. By comparison, the coverage probabilities for both linear and quadratic coefficients for the marginal-quasi-2DMC + BMA method are generally too high, about 100%. For the linear model quasi-2DMC + BMA yields good estimates of the ERR coefficient, but for marginal-quasi-2DMC + BMA these are upwardly biased. For the linear-quadratic model both quasi-2DMC + BMA and marginal-quasi-2DMC + BMA methods yield substantially biased estimates. Conclusions The performance of both quasi-2DMC + BMA and marginal-quasi-2DMC + BMA methods is bad, with bias and poor coverage.

Electroencephalography (EEG) source localization in epileptology continues to be a challenge for neuroscientists. A number of inverse solution (IS) methodologies have been proposed to solve this problem, and their advantages and limitations have been described. In the present work, a previously developed IS approach called Bayesian model averaging (BMA) is introduced in clinical practice in order to improve the localization accuracy of epileptic discharge sources. For this study, 31 patients with the diagnosis of partial epilepsies were studied: 14 had benign childhood epilepsy with centrotemporal spikes and 17 had temporal lobe epilepsy (TLE). The underlying epileptic sources were localized using the BMA approach, and the results were compared with those expected from the clinical diagnosis. Additional comparisons with results obtained from 3 of the most commonly used distributed IS methods for these purposes (minimum norm [MN], weighted minimum norm [WMN], and low-resolution electromagnetic tomography [LORETA]) were carried out in terms of source localization accuracy and spatial resolutions. The BMA approach estimated discharge sources that were consistent with the clinical diagnosis, and this method outperformed LORETA, MN, and WMN in terms of both localization accuracy and spatial resolution. The BMA was able to localize deeper generators with high accuracy. In conclusion, the BMA methodology has a great potential for the noninvasive accurate localization of epileptic sources, even those located in deeper structures. Therefore, it could be a promising tool for clinical practice in epileptology, although additional studies in other types of epileptic syndromes are necessary.

Application of Bayesian model averaging in modeling long-term wind speed distributions

Correcting mean areal precipitation forecasts to improve urban flooding predictions by using long short-term memory network

Study regionAkaki is a headwater catchment of Awash River Basin that hosts the capital city of Ethiopia, Addis Ababa. The area encompasses several agglomerated towns, water supply, and hydropower reservoirs and is characterized by a chain of mountains and floodplains. Due to basin rainfall, and the expansion of urbanized areas, the catchment is frequently affected by flooding. Study focusThis study evaluates dynamic Bayesian Model Averaging (BMA) approach to improve rainfall estimation over the catchment by blending four high-resolution satellite rainfall estimate (SRE) products. Using daily data (2003–2019) observed at thirteen stations as a reference, seven statistical metrics served to assess the point and spatial scale accuracy of the rainfall estimates. New hydrological insightsMain findings from this study are: (i) the blended product outperformed the individual SRE products by notably improving correlation with in-situ observed rainfall, and reducing the error of the estimated rainfall, (ii) the blended and individual SRE products performed better in the highlands than the lowlands of the catchment, and (iii) the amount of daily rainfall during the main-rainy season was mostly overestimated by the individual SRE products but was fairly estimated by the blended product. This study showed the nonexistence of surpassing individual SRE products and emphasized the blending of several products for gaining optimal results from each product.

To study the impact of training sample heterogeneity on the performance of Bayesian model averaging (BMA), two BMA experiments were performed on probabilistic quantitative precipitation forecasts (PQPFs) in the northern China region in July and August of 2010 generated from an 11-member short-range ensemble forecasting system. One experiment, as in many conventional BMA studies, used an overall training sample that consisted of all available cases in the training period, while the second experiment used stratified sampling BMA by first dividing all available training cases into subsamples according to their ensemble spread, and then performing BMA on each subsample. The results showed that ensemble spread is a good criterion to divide ensemble precipitation cases into subsamples, and that the subsamples have different statistical properties. Pooling the subsamples together forms a heterogeneous overall sample. Conventional BMA is incapable of interpreting heterogeneous samples, and produces unreliable PQPF. It underestimates the forecast probability at high-threshold PQPF and local rainfall maxima in BMA percentile forecasts. BMA with stratified sampling according to ensemble spread overcomes the problem reasonably well, producing sharper predictive probability density functions and BMA percentile forecasts, and more reliable PQPF than the conventional BMA approach. The continuous ranked probability scores, Brier skill scores, and reliability diagrams of the two BMA experiments were examined for all available forecast days, along with a logistic regression experiment. Stratified sampling BMA outperformed the raw ensemble and conventional BMA in all verifications, and also showed better skill than logistic regression in low-threshold forecasts.

The development of high resolution mapping models of forest attributes based on employing machine or deep learning techniques has increasingly accelerated in the last couple of years. The consequence of this is the widespread availability of multiple sources of information, which can either lead to a potential confusion, or to a possibility to get an "extended” insight into the state of our forests by interpreting these sources jointly. This contribution aims at addressing the latter, by relying on the Bayesian model averaging (BMA) approach.BMA is a method that can be used in building a consensus from an ensemble of different model predictions. It can be seen as weighted mean of different predictions with weights reflecting the predictive performances of different models, or as a finite mixture model which estimates the probability that each observation from the independent validation dataset has been generated by one of the models belonging to the ensemble. BMA can thus be used to diagnose and understand the difference in the predictions and to possibly interpret them.The predictions in our case are the forest canopy height estimations for the metropolitan France coming from 5 different AI models [1-5], while the independent validation dataset comes from the French National Forest Inventory (NFI) disposing with some 6000 plots per year, distributed across the territory of interest. For every plot we have several measurements/estimations of the forest canopy height out of which the following two are considered in this study: h_m – the maximum total height (from the tree's base level to the terminal bud of the tree's main stem) measured within the plot, and h_dom – the average height of the seven largest dominant trees per hectare.In this contribution we present for every considered plot the dominant model with respect to both references i.e. the model having the highest probability to be the one generating measurements/estimations at NFI plot (h_m and h_dom). We present as well as the respective inter-model and the intra-model variance estimations, allowing us to propose a series of hypotheses concerning the established differences between predictions of individual models in function of their specificities.[1] Schwartz, M., et al.: FORMS: Forest Multiple Source height, wood volume, and biomass maps in France at 10 to 30 m resolution based on Sentinel-1, Sentinel-2, and Global Ecosystem Dynamics Investigation (GEDI) data with a deep learning approach, Earth Syst. Sci. Data, 15, 4927–4945, 2023, https://doi.org/10.5194/essd-15-4927-2023[2] Lang, N., et al.: A high-resolution canopy height model of the Earth, Nat Ecol Evol 7, 1778–1789, 2023. https://doi.org/10.1038/s41559-023-02206-6[3] Morin, D. et al.: Improving Heterogeneous Forest Height Maps by Integrating GEDI-Based Forest Height Information in a Multi-Sensor Mapping Process, Remote Sens., 14, 2079. 2022, https://doi.org/10.3390/rs14092079[4] Potapov, P., et al.: Mapping global forest canopy height through integration of GEDI and Landsat data, Remote Sensing of Environment, 253, 2021, https://doi.org/10.1016/j.rse.2020.112165.[5] Liu, S. et al.: The overlooked contribution of trees outside forests to tree cover and woody biomass across Europe, Sci. Adv. 9, eadh4097, 2023, 10.1126/sciadv.adh4097.

Evaluation of crop model prediction and uncertainty using Bayesian parameter estimation and Bayesian model averaging

On a Quest for Robustness: About Model Risk, Randomness and Discretion in Credit Risk Stress Tests

Characterization of aquifer heterogeneity is inherently difficult because of the insufficiency of data, the inflexibility of parameterization methods, and non-uniqueness of parameterization methods. Groundwater predictions are greatly affected by multiple interpretations of aquifer properties and the uncertainties of model parameters. This study introduces a Bayesian model averaging (BMA) method along with multiple generalized parameterization (GP) methods to identify hydraulic conductivity and along with multiple simulation models to predict groundwater head and quantify the prediction uncertainty. Two major issues about BMA are discussed. The first problem is with using Occam’s window in usual BMA applications. Occam’s window only accepts models in a very narrow range, tending to single out the best method and discard other good methods. A variance window is proposed to replace Occam’s window to cope with this problem. The second problem is with using the Kashyap information criterion (KIC) in the approximation of posterior model probabilities, which tends to prefer highly uncertain model by considering the Fisher information matrix. The Bayesian information criterion (BIC) is recommended because it is able to avoid controversial results and it is computationally efficient. Numerical examples are designed to test the Bayesian model averaging method on hydraulic conductivity identification and groundwater head prediction. The proposed methodologies are then applied to the hydraulic conductivity identification of the Alamitos Gap area, and the hydraulic conductivity estimation and groundwater head prediction of the “1,500-foot” sand in East Baton Rouge Parish, Louisiana. The results show that the GP method provides great flexibility in parameterization with small conditional variance. The use of the variance window is necessary to avoid a dominant model when many models perform equally well. Compared to KIC, BIC is able to give an unbiased posterior model probability. It is also concluded that the uncertainty increases by including multiple models under the BMA framework, but risks are reduced by avoiding overconfidence in the solution from one model.

Research on improving an overland flow model is presented for urban pluvial flooding under the dual-drainage concept where sewer flow dynamically interacts with overland flow. This occurs during heavy storms when the sewer system is surcharged. The system becomes pressurised and overland flow increases by the additional volume flowing out from the sewer. To represent the overland flow realistically, a new methodology was developed to automatically create the overland flow network which can interact with the drainage system. Use is made of high-resolution, accurate Digital Elevation Model data collected by the LiDAR technique. This approach updates the current urban drainage models to urban flood models with detailed representation of overland flow processes such as pond forming, flow through preferential surface pathways and surface drainage capacity. This work advances new areas of urban flood management including improvement in real-time control and of links with rainfall now-casting, and short term urban flood forecasting. The dual-drainage approach is appropriate for real-time applications.

Urban flooding is emerging as an increasingly severe global challenge due to climate change and urbanization. Although machine learning offers numerous solutions for urban flood forecasting, its application remains constrained. Existing research remains constrained by the scarcity of traditional hydrological monitoring data, and the absence of systematic comparisons across multiple models creates uncertainty when selecting the most suitable algorithms and features, making the decision-making mechanisms for selecting the most suitable algorithms and features remains unclear. To address these challenges, social media data was adopted as the sole basis in this study to evaluate and compare the performance of seven typical machine learning algorithms in urban flood forecasting. The Shapley Additive exPlanations (SHAP) framework was established, investigating the adaptability of the selected algorithms based on a multidimensional feature system while elucidating the decision-making mechanisms for selecting the most suitable algorithms and features. The results suggest that: (1) Social media data can serve as the sole source for precise urban flood identification, overcoming the real-time and spatial coverage limitations of traditional methods. (2) Different machine learning models show significant performance heterogeneity; reliance on a single model risks systematic bias, whereas ensemble tree models demonstrate superior predictive performance. (3) Feature importance is highly model-dependent, exhibiting contextual sensitivity and interactive influence mechanisms. Therefore, feature engineering should be based on multi-model consensus, prioritizing features with significant differences such as natural characteristics and risk exposure.

It is unrealistic to build independent alternative models to constitute the model space of Bayesian model aver­ aging (BMA) in groundwater/surface water modeling. Using uniform prior weights can lead to overweighting models with similar structures, as well as biased posterior model weights and BMA predictions. This study applied a correlation matrix R to measure the correlations among alternative models. And two weighting schemes based on R, namely the cos-square (CS) and capped eigenvalue (CE), were used to dilute models’ prior weights. Additionally, the effective model number (Neff) metric derived from R was proposed to measure the effectiveness of BMA model set. Based on two real-world cases (snowmelt runoff modeling and groundwater modeling), and a synthetical groundwater case, we validated the importance of prior weight dilution and the important value of the R-based methods in improving BMA prediction. The results demonstrated that the prior weight dilution schemes redistribute models’ prior weights by penalizing highly correlated models while rewarding those with relatively independent structures. The BMA predictive performance is improved using the weight dilution schemes, with the CS scheme outperforming the CE scheme. In addition, the correlation matrix provides insight into the rationality of the model structures in the BMA model set. The metric of Neff can serve as an effective tool for quantifying the effectiveness of the model set, which provides an important reference for updating the model set and improving BMA predictions with prior weight dilution schemes.