Discharge Prediction Research Articles

UAV-based DEM augmentation using ConSinGAN for efficient flood parameter prediction with machine learning and 1D hydrodynamic models

Flooding disasters have a significant financial impact on metropolitan regions, highlighting the critical need for accurate flood prediction methods. This study addresses the problem of limited datasets in flood prediction by employing a novel approach that integrates Unmanned Aerial Vehicle (UAV)-based Digital Elevation Models (DEMs) and the ConSinGAN generative model to enhance flood prediction accuracy. The methodology involves synthesizing DEMs using ConSinGAN to augment small datasets, followed by the application of three machine learning (ML) algorithms—Random Forest (RF), Support Vector Machines (SVM), and K-Nearest Neighbors (KNN)—to predict crucial flood-related parameters such as discharge and elevation. The predictive performance of the models was evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), with units specified as meters for elevation and cubic meters per second for discharge. The RF model exhibited superior predictive capabilities, achieving a training RMSE of 0.68 cubic meters per second and an MAE of 0.30 cubic meters per second for discharge predictions. For elevation predictions, the RF model achieved a training RMSE of 0.0101 m and an MAE of 0.0046 m. Cross-validation results indicated the lowest RMSE values of 2.44 cubic meters per second for discharge and 0.03 m for elevation, with corresponding MAE values of 1.37 cubic meters per second and 0.02 m, respectively. This integrated approach, combining UAV-based DEM augmentation and advanced ML techniques, provides a robust tool for urban planners and decision-makers to accurately forecast flood parameters, thereby facilitating better disaster preparedness and mitigation strategies.

Comparison of survival to hospital discharge predictions in elderly traumatic brain injury using CHID, CIT, and CART algorithms

Comparison of survival to hospital discharge predictions in elderly traumatic brain injury using CHID, CIT, and CART algorithms

Anisotropic turbulence modeling for natural channel flow: A numerical approach with finite element method

In this study, we propose a numerical model aimed at improving the understanding and prediction of flow velocities in natural channels. This model specifically focuses on the challenges presented by anisotropic turbulence and bottom roughness. By incorporating the mixing length concept into an algebraic turbulence model and employing the finite element method with Streamline-Upwind/Petrov–Galerkin (SUPG) stabilization, our model seeks to refine the simulation of axial velocity distribution and secondary motions. Validation was achieved through Acoustic Doppler Current Profiler (ADCP) measurements in three sections of the Canal do Rodeador, showing our model’s predictions of discharge and average velocity to have a deviation of approximately 9.34% from experimental data. The results underline the significance of secondary currents and turbulence anisotropy in shaping channel flow behaviors, offering new insights into the interactions between flow characteristics and channel bed features. This model stands out as a robust tool for hydraulic structure design and hydrokinetic potential evaluation, providing a non-intrusive, cost-effective alternative to traditional methods.

External validation of the SHA2PE score and its comparison to the Oakland score for the prediction of safe discharge in patients with lower gastrointestinal bleeding.

The growing incidence of lower gastrointestinal bleeding (LGIB) is leading to a rise in-hospital admissions even though most LGIB episodes are self-limiting. The Oakland and SHA2PE scores were designed to identify patients best suited to outpatient care. Our aim is explore the validity of the SHA2PE score and compare both of these scores in terms of predictiveness of safe discharge. Retrospective observational study of LGIB patients admitted to a tertiary hospital between June 2014 and June 2019. Safe discharge was defined as the absence of all the following: blood transfusion, haemostatic intervention, re-bleeding, in-hospital death, and re-admission due to LGIB within 28days after discharge. From 595 hospital admissions for LGIB, 398 episodes were included. Fifty-four per cent met safe discharge criteria, with these cases being younger, with a lower score in the Charlson's index and significantly higher haemoglobin concentration upon arrival. The performance of both scores was good, with an AUC for the Oakland score of 0.85 (95% CI 0.82-0.89) and of 0.797 (95% CI 0.75-0.84) for the SHA2PE score. The Oakland score performed better in terms of prediction of safe discharge, with a positive predictive value and specificity of 100% when a cut-off value of ≤ 8 points was used; however, only a minority of patients might benefit from its implementation given its low sensitivity. Almost half of the patients admitted for LGIB met criteria for safe discharge. However, the available indexes only allow for the identification of a small proportion of those patients candidates for outpatient care.

Predicting & Analysing Discharge and Sediment of Sore Catchment, Baro-Akobo Sub Basin Using Saturation Excess Model, Western Ethiopia

Soil erosion, land degradation, and soil loss are major problems in Ethiopia's Baro-Akobo Basin. In the Ethiopian highlands, many hydrological models have been employed to forecast sediment and discharge in watersheds of varying sizes. PED mode was used for this study because it requires fewer data and is well recognized by academics in Ethiopia's Upper Blue Nile. The parameter efficient semi-distributed watershed model (PED) was used to study the prediction of discharge and sediment in the Sore watershed. The objective of this study was to analyze and forecast the sediment and discharge from the Sore catchment in the western Ethiopian region of the Baro-Akobo Sub basin. The selected watershed was Sore (1665.5 Km2) in the Baro-Akobo River Basin. The stream flow and sediment data for the Sore watershed from 2005 to 2013 and 2014 to 2018 were used for the model's calibration and validation by the Ministry of Water and Electricity. For the Sore watershed, the daily time step scale model efficiency of PED-W was found to be NSE= 0.40 and NSE= 0.44 for the calibration of sediment and discharge, respectively. Similarly, NSE values for the Sore watershed's NSE= -0.82 and PED-W's discharge and sediment during the validation period were found. As a result, there was a range of uneven to acceptable agreement between the simulated and observed discharge and sediment at daily time steps. In comparison to the daily time step scale, the models' performance varied and performed better on the monthly time step scale. According to the overall model performance, the PED-W model was shown to be less accurate than the SWAT model for estimating stream flow and sediment output. The PED model, on the other hand, was exactly the same as the previously employed SWAT model and was used for calibration and validation at monthly time steps.

Application of Machine Learning Models for Improving Discharge Prediction in Ungauged Watershed: A Case Study in East DuPage, Illinois

Application of Machine Learning Models for Improving Discharge Prediction in Ungauged Watershed: A Case Study in East DuPage, Illinois

Stage–discharge prediction in the multi-stage ice-covered compound channel

The multi-stage compound channel, which is a common pattern in natural alluvial rivers and the regulation projects of urban rivers, inevitably freezes in winter when it is situated in cold northern areas with high latitudes. Given that ascertaining the stage–discharge relationship for rivers is the foundation for the development of flood control schemes and water resources management, this study concentrates on proposing an analytical model for predicting the stage–discharge curves of multi-stage ice-covered compound channels. In deducing the analytical model, the cross section of the channel is first segmented into several homogeneous subregions that can be grouped into seven categories according to the geometric characteristics. Through analyzing the momentum transfer between adjacent subregions, the force balance equation for each subregion is then established to get the bulk mean velocity for the corresponding subregion, thereby obtaining the discharge by solving a tridiagonal matrix. Subsequently, measurements from two-stage and three-stage ice-covered compound channel experiments and three sets of experimental data from the literature are used to validate the performance of the proposed model. Good agreement between the predictions and the measured data suggests that the deduced model can accurately estimate the discharge for the multi-stage ice-covered compound channels when the flow depth is given. Finally, sensitivity analysis indicates that Manning's roughness coefficient of the channel bed has a more pronounced impact on the stage–discharge relationship than that of the ice cover. Moreover, when compared to the two-stage ice-covered compound channel, the multi-stage ice-covered compound channel offers greater potential for water resource utilization.

Improving the Reliability of Compound Channel Discharge Prediction Using Machine Learning Techniques and Resampling Methods

Improving the Reliability of Compound Channel Discharge Prediction Using Machine Learning Techniques and Resampling Methods

Estimation of Groundwater River Availability in Leang Lonrong Cave Using ARIMA Model and Econophysics Valuation Approach

This quantitative descriptive study uses the ARIMA model approach to predict groundwater river water availability. It assesses its valuation with an econophysics orientation in the Leang Lonrong Cave area of TN Bantimurung Bulusaraung. Secondary data from 2010-2023, including Water Discharge and Water Level data from the River Basin Center of Pompengan Jeneberang (BBWS) and primary data on the Valuation of Willingness to Pay, were utilized. The analysis reveals that water discharge predictions (2024-2030) in the Gua Leang Lonrong area indicate consistent annual water availability, with the highest discharge projected in December 2030 (5.08 m3/s) and the lowest in September 2024 (1.12 m3/s). Valuation of utilizing Leang Lonrong River water through willingness to pay showed residents willing to pay Rp 42,142.85/month for clean water management. At the same time, visitors are willing to contribute Rp 21,596.77/visit to enhance water quality management. Elasticity analysis modeling water discharge with willingness to pay suggests a relatively elastic relationship, with a minor impact, indicating unresponsiveness to changes in water discharge percentages. Statistical and economic methods were employed to analyze daily water discharge patterns and evaluate residents' willingness to pay for underground river water utilization. The ARIMA model (0,1,1) forecasted future water discharge patterns with a mean absolute percentage error (MAPE) of 33.7%. The study surveyed 62 respondents, finding an average willingness to pay approximately IDR 21,596.77 per visit to maintain Leang Lonrong River water quality management.

Utilizing sequential modeling in collaborative method for flood forecasting

Flood forecasting has been a major challenge in hydrology for decades. A variety of approaches have been developed, including numerical models and data-driven models, such as Deep Learning (DL), for disaster early-warning and monitoring systems. This study proposed two combination models, exploring different feature separation techniques to improve the 12-hour multi-step discharge prediction. Specifically, two different approaches were implemented, namely the separation of wet and dry periods (Combination Model No. 1; CM-1) and the separation of flood and non-flood events (Combination Model No. 2; CM-2) for comparison with the baseline model trained on an entire dataset. Multivariate time series discharge data were used, measured at upstream and downstream stations within the same river basins. Experiments were conducted in three case study areas in Thailand: the Lower Loei River Basin, the Upper Nan River Basin, and the Upper Chi River Basin.In the proposed model combination approach, three deep-learning models were investigated: vanilla recurrent neural networks (RNNs) and their variants, long short-term memory networks (LSTMs), and gated recurrent units (GRUs). In general, CM-2 exhibited superior performance compared to the other models in most of the experimented neural networks, while the CM-1 approach outperformed the baseline model. This emphasized the advantages of our proposed feature separation framework, particularly in improving model performance, with respect to flood events. However, some variations were observed in the Upper Chi River Basin because the more complex models, such as LSTMs and GRUs, possibly overfitted the training set with less fluctuating data. Our study highlighted the significant contribution of combining two models through feature separation, tailored to specific periods, thereby enhancing the accuracy of flood predictions.

Physics-informed neural networks for solving flow problems modeled by the 2D Shallow Water Equations without labeled data

This paper investigates the application of physics-informed neural networks (PINNs) to solve free-surface flow problems governed by the 2D shallow water equations (SWEs). Two types of PINNs are developed and analyzed: a physics-informed fully connected neural network (PIFCN) and a physics-informed convolutional neural network (PICN). The PINNs eliminate the need for labeled data for training by employing the SWEs, initial and boundary conditions as components of the loss function to be minimized. Results from a set of idealized and real-world tests showed that the prediction accuracy and computation time (i.e., training time) of both PINNs may be less affected by the resolution of the domain discretization when compared against solutions by a Finite Volume (FV) model. Overall, the PICN shows a better trade-off between computational speed and accuracy than the PIFCN. Also, our results for the idealized problems indicated that PINNs can provide more than 5 times higher prediction accuracy than the FV model, while the FV simulation with coarse resolution (e.g., 10 m) can provide sub-centimeter accurate (RMSE) solutions at least one order of magnitude faster than the PINNs. Results from a river flood simulation showed that PINNs delivered better speed-accuracy trade-off than the FV model in terms of predicting the water depth, while FV models outperformed the PINNs for predictions of total flow discharge.

Prediction of Peak Discharge Using the SCS Curve Number Method in the Manikin Watershed

Surface runoff is a crucial hydrological variable in the analysis of water infrastructure planning. A reliable method for predicting surface runoff resulting from rainfall in an ungauged watershed is the SCS CN method. This research aims to represent the effectiveness of the Curve Number (CN) method in calculating peak discharge in the Manikin Watershed. The data used for this analysis includes rainfall data from three rain stations, each with a 25-year dataset, water level data of 11 years, digital elevation model (DEM) data, land use maps, and hydrogeological maps. The SCS curve number method is the most commonly used method for the estimation of peak discharge in a watershed. The calculated flood discharge values for the Manikin River Basin, with return periods of 5, 10, 20, 25, 50, 100, 500, and 1000 years, are as follows: 60.32 m3/s, 84.74 m3/s, 111.98 m3/s, 121.41 m3/s, 153.25 m3/s, 188.79 m3/s, 287.24 m3/s, and 337.30 m3/s, respectively. The reliability testing of the method in the Manikin Watershed was determined by a Nash–Sutcliffe efficiencies (NSE) value of 0.93 and root mean square error (RMSE) value of 19.70. As a result, the Curve Number Method proves to be highly reliable in representing the peak discharge in the Manikin Watershed.

Stage and discharge prediction from documentary time-lapse imagery

Imagery from fixed, ground-based cameras is rich in qualitative and quantitative information that can improve stream discharge monitoring. For instance, time-lapse imagery may be valuable for filling data gaps when sensors fail and/or during lapses in funding for monitoring programs. In this study, we used a large image archive (>40,000 images from 2012 to 2019) from a fixed, ground-based camera that is part of a documentary watershed imaging project (https://plattebasintimelapse.com/). Scalar image features were extracted from daylight images taken at one-hour intervals. The image features were fused with United States Geological Survey stage and discharge data as response variables from the site. Predictions of stage and discharge for simulated year-long data gaps (2015, 2016, and 2017 water years) were generated from Multi-layer Perceptron, Random Forest Regression, and Support Vector Regression models. A Kalman filter was applied to the predictions to remove noise. Error metrics were calculated, including Nash-Sutcliffe Efficiency (NSE) and an alternative threshold-based performance metric that accounted for seasonal runoff. NSE for the year-long gap predictions ranged from 0.63 to 0.90 for discharge and 0.47 to 0.90 for stage, with greater errors in 2016 when stream discharge during the gap period greatly exceeded discharge during the training periods. Importantly, and in contrast to gap-filling methods that do not use imagery, the high discharge conditions in 2016 could be visually (qualitatively) verified from the image data. Half-year test sets were created for 2016 to include higher discharges in the training sets, thus improving model performance. While additional machine learning algorithms and tuning parameters for selected models should be tested further, this study demonstrates the potential value of ground-based time-lapse images for filling large gaps in hydrologic time series data. Cameras dedicated for hydrologic sensing, including nighttime imagery, could further improve results.

Discharge estimation in compound channels with converging and diverging floodplains using an optimised Gradient Boosting Algorithm

ABSTRACT River discharge estimation is vital for effective flood management and infrastructure planning. River systems consist of a main channel and floodplains, collectively forming a compound channel, posing challenges in discharge calculation, particularly when floodplains converge or diverge. In the present study, ML algorithms such as XGBoost, CatBoost, and LightGBM were developed to predict discharge in a compound channel. PSO algorithm is applied for optimization of hyperparameters of gradient boosting models, denoted as PSO-XGBoost, PSO-LightGBM, and PSO-CatBoost. ML model discharge predictions were validated with existing empirical models and feature importance was explored using SHAP and sensitivity analysis. Results show that all three gradient-boosting algorithms effectively predict discharge in compound channels and are further enhanced by application of PSO algorithm. The R2 values for XGBoost, PSO-XGBoost, CatBoost, and PSO-CatBoost exceed 0.95, whereas they are above 0.85 for LightBoost and PSO-LightBoost. PSO-CatBoost performance is better than other models based on findings of statistical performance parameters, uncertainty analysis, reliability index, and resilience index for prediction of discharge in a compound channel with converging and diverging flood plains. The findings of this study validate the suitability of the proposed models especially optimized with PSO is recommended for predicting discharge in a compound channel.

Application and Evaluation of Stage–Storage–Discharge Methodology in Hydrological Study of the Southern Phosphate Mining Model Domain in Southwest Florida

This hydrological study investigated a combined rating methodology tested on a 14,090 km2 area in Southwest Florida. The approach applied the Hydrological Simulation Program-Fortran (HSPF) over a 23-year period and was validated by 28 stream gauging stations. The regional hydrological complexity includes natural and agricultural areas, as well as extensive phosphate mining and urbanizing areas. This application is a novel and efficient methodology for generating stage–storage–discharge relationships using a geographic information system (GIS), empirical equations, and spreadsheets for 148,000 isolated and connected alluvial wetlands within the model domain. The validation metrics used to evaluate the applied methodology for populating the stage–storage–discharge relationship demonstrated the model effectiveness in simulating a range of hydrological events across various regions. For discharge prediction, the Nash–Sutcliffe efficiency values surpassed 0.7 at most stations, with an average of 0.67, and the average R squared was 0.74. This methodology, when applied, achieved a root-mean-square error of 4 m3/s for discharge prediction and 0.47 m for stage prediction. However, limitations emerged in simulating baseflow (low flows), highlighting the need for integrated modeling approaches to accurately capture groundwater–surface water interactions. The research provides an improved means for modeling regional water resources and lays the groundwork for enhanced hydrological modeling in watersheds with complex alluvial and isolated wetland systems.

An approach to hydraulic design of Conical Central Baffle Flumes

Design methodology for providing a Conical Central Baffle flume (CCBF) has not yet been found in the literature, though it is required by the hydraulic engineers considering the increased popularity of the baffled flumes, now-a-days. Some guidance for discharge prediction using CCBF is only found in the literature. The main objective of the present research is to address the issue of hydraulic design of CCBFs. A measuring flume serves as a device for quantifying the flow rate in an open channel. These flumes operate on the basis of principles of critical flow and energy conservation. Conventionally, it is assumed that the critical flow occurs at location of the minimum flow cross-section or throat. In the current investigation, a CCBF is employed to gauge the flow discharge. The CCBF configuration involves placing one conical obstruction in a trapezoidal channel, thus creating a throat. One of the objectives of the study is to scrutinize the prevailing location of the critical section and introduce a refining approach to discharge prediction. Through a series of experiments and numerical simulations, the present study endeavors to ascertain the true location of the critical section in CCBF. Contrary to the traditional belief that the critical section is always at the throat, the investigation revealed that the critical section tends to shift upstream with increased discharge. The theoretical analysis, employing energy expression and critical flow concepts, unveiled that despite this shifting nature, the energy-critical depth ratio of the flume remains constant. Based on these findings, a discharge prediction model is proposed. The model demonstrates a mean relative error of ±1.42 % and a maximum relative error of 3.5 % in discharge prediction. The theoretical investigation also highlighted that the critical section occurring at the throat signifies the minimum discharge value, and induced a threshold choking flow condition in the flume. The present study also proposes the expressions for minimum as well as maximum discharge measuring capacity of a CCBF which leads to its proper design methodology, which has never been presented so far in the literature. Additionally, a design example is presented, elucidating the process of designing a conical CCBF for a trapezoidal channel based on the understanding of the limiting discharge.

Temporal cluster-based local deep learning or signal processing-temporal convolutional transformer for daily runoff prediction?

Water scarcity poses a major obstacle to sustainable development, and precise discharge prediction plays a vital role in enabling effective water resource management. This study investigated improved prediction techniques for nonstationary time series. The study evaluated the effect of signal processing techniques and blending approaches on the performance of deep learning models for daily discharge prediction. It also compared the performance of cluster-based local modeling with hybrid signal processing-deep learning approaches. Two robust deep learning methods, Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN), along with a powerful signal processing approach called discrete wavelet transform, were utilized for prediction of daily discharge. Three blending approaches were assessed: 1) decomposing both inputs and target, 2) decomposing only the target, and 3) decomposing only the inputs and then blending them with deep learning models. Also, a new hybrid deep learning based model namely discrete wavelet transform-Temporal Convolutional Transformer (DWT-TCT) was developed. The results showed that a single-output wavelet transform-deep learning model (3rd blending approach) outperformed multi-output models, demonstrating a relative enhancement of up to 56% for the LSTM model and 51% for the CNN model. Furthermore, temporal cluster-based local modeling displayed promising performance, resulting in an improvement of up to 18% in NRMSE compared to the wavelet transform-deep learning model, while also requiring less computational cost. The successful results of the temporal cluster-based local modeling approach provide a beneficial alternative to hybrid signal processing-deep learning models. In addition the results showed that the proposed DWT-TCT model outperformed all other models with NRMSE ranges from 6.8% to 16.2% in the study areas. The results have implications for hydrology and water resources management, as they can be used to develop more precise and effective models for predicting discharge in view of nonstationarity.

Is it possible to estimate the number of patients with COVID-19 admitted to intensive care units and general wards using clinical and telemedicine data?

Gabaldi et al. utilized telemedicine data, web search trends, hospitalized patient characteristics, and resource usage data to estimate bed occupancy during the COVID-19 pandemic. The results showcase the potential of data-driven strategies to enhance resource allocation decisions for an effective pandemic response. To develop and validate predictive models to estimate the number of COVID-19 patients hospitalized in the intensive care units and general wards of a private not-for-profit hospital in São Paulo, Brazil. Two main models were developed. The first model calculated hospital occupation as the difference between predicted COVID-19 patient admissions, transfers between departments, and discharges, estimating admissions based on their weekly moving averages, segmented by general wards and intensive care units. Patient discharge predictions were based on a length of stay predictive model, assessing the clinical characteristics of patients hospitalized with COVID-19, including age group and usage of mechanical ventilation devices. The second model estimated hospital occupation based on the correlation with the number of telemedicine visits by patients diagnosed with COVID-19, utilizing correlational analysis to define the lag that maximized the correlation between the studied series. Both models were monitored for 365 days, from May 20th, 2021, to May 20th, 2022. The first model predicted the number of hospitalized patients by department within an interval of up to 14 days. The second model estimated the total number of hospitalized patients for the following 8 days, considering calls attended by Hospital Israelita Albert Einstein's telemedicine department. Considering the average daily predicted values for the intensive care unit and general ward across a forecast horizon of 8 days, as limited by the second model, the first and second models obtained R² values of 0.900 and 0.996, respectively and mean absolute errors of 8.885 and 2.524 beds, respectively. The performances of both models were monitored using the mean error, mean absolute error, and root mean squared error as a function of the forecast horizon in days. The model based on telemedicine use was the most accurate in the current analysis and was used to estimate COVID-19 hospital occupancy 8 days in advance, validating predictions of this nature in similar clinical contexts. The results encourage the expansion of this method to other pathologies, aiming to guarantee the standards of hospital care and conscious consumption of resources. Developed models to forecast bed occupancy for up to 14 days and monitored errors for 365 days. Telemedicine calls from COVID-19 patients correlated with the number of patients hospitalized in the next 8 days.

Ambulatory Status at Discharge Predicts Six-Month Mortality in Patients with COVID-19: A Retrospective Cohort Study.

This retrospective cohort study aimed to evaluate the association between ambulatory status at discharge and six-month post-discharge mortality among adults with coronavirus disease (COVID-19). We analyzed data from 398 patients aged over 18 admitted to a tertiary hospital in South Korea between December 2019 and June 2022. Patients were classified into two groups based on their ambulatory status at discharge: ambulatory (able to walk independently, n = 286) and non-ambulatory (unable to walk independently, requiring wheelchair or bed-bound, n = 112). Our analysis revealed that six-month survival rates were significantly higher in the ambulatory group (94.2%) compared to the non-ambulatory group (84.4%). Multivariate analysis identified ambulatory status at discharge (p = 0.047) and pre-existing malignancy (p = 0.007) as significant prognostic factors for post-discharge survival. This study highlights that the ability to walk independently at discharge is a crucial predictor of six-month survival in COVID-19 patients. These findings emphasize the need for interventions to improve the physical performance of non-ambulatory patients, potentially enhancing their survival prospects. This underscores the importance of targeted rehabilitation and physical therapy for the comprehensive care of COVID-19 survivors.

Graphics-processing-unit-accelerated ice flow solver for unstructured meshes using the Shallow-Shelf Approximation (FastIceFlo v1.0.1)

Abstract. Ice-sheet flow models capable of accurately projecting their future mass balance constitute tools to improve flood risk assessment and assist sea-level rise mitigation associated with enhanced ice discharge. Some processes that need to be captured, such as grounding-line migration, require high spatial resolution (under the kilometer scale). Conventional ice flow models mainly execute on central processing units (CPUs), which feature limited parallel processing capabilities and peak memory bandwidth. This may hinder model scalability and result in long run times, requiring significant computational resources. As an alternative, graphics processing units (GPUs) are ideally suited for high spatial resolution, as the calculations can be performed concurrently by thousands of threads, processing most of the computational domain simultaneously. In this study, we combine a GPU-based approach with the pseudo-transient (PT) method, an accelerated iterative and matrix-free solution strategy, and investigate its performance for finite elements and unstructured meshes with application to two-dimensional (2-D) models of real glaciers at a regional scale. For both the Jakobshavn and Pine Island glacier models, the number of nonlinear PT iterations required to converge a given number of vertices (N) scales in the order of 𝒪(N1.2) or better. We further compare the performance of the PT CUDA C implementation with a standard finite-element CPU-based implementation using the price-to-performance metric. The price of a single Tesla V100 GPU is 1.5 times that of two Intel Xeon Gold 6140 CPUs. We expect a minimum speedup of at least 1.5 times to justify the Tesla V100 GPU price to performance. Our developments result in a GPU-based implementation that achieves this goal with a speedup beyond 1.5 times. This study represents a first step toward leveraging GPU processing power, enabling more accurate polar ice discharge predictions. The insights gained will benefit efforts to diminish spatial resolution constraints at higher computing performance. The higher computing performance will allow for ensembles of ice-sheet flow simulations to be run at the continental scale and higher resolution, a previously challenging task. The advances will further enable the quantification of model sensitivity to changes in upcoming climate forcings. These findings will significantly benefit process-oriented sea-level-projection studies over the coming decades.