A Systematic Review of Deep Learning Applications in Streamflow Data Augmentation and Forecasting DOI Creative Commons
Muhammed Sit, Bekir Zahit Demiray, İbrahim Demir

et al.

EarthArXiv (California Digital Library), Journal Year: 2022, Volume and Issue: unknown

Published: Sept. 26, 2022

The volume and variety of Earth data have increased as a result growing attention to climate change and, subsequently, the availability large-scale sensor networks remote sensing instruments. This has been an important resource for data-driven studies generate practical knowledge services, support environmental modeling forecasting needs, transform earth science research thanks computational resources popularity novel techniques like deep learning. Timely accurate simulation extreme events are critical planning mitigation in hydrology water resources. There is strong need short-term long-term forecasts streamflow, benefiting from recent developments learning methods. In this study, we review literature that employ tackling tasks either improve quality streamflow or forecast streamflow. study aims serve starting point by covering latest approaches those topics well highlighting problems, limitations, open questions with insights future directions.

Language: Английский

Neural network attribution methods for problems in geoscience: A novel synthetic benchmark dataset DOI Creative Commons
Antonios Mamalakis, Imme Ebert‐Uphoff, Elizabeth A. Barnes

et al.

Environmental Data Science, Journal Year: 2022, Volume and Issue: 1

Published: Jan. 1, 2022

Despite the increasingly successful application of neural networks to many problems in geosciences, their complex and nonlinear structure makes interpretation predictions difficult, which limits model trust does not allow scientists gain physical insights about problem at hand. Many different methods have been introduced emerging field eXplainable Artificial Intelligence (XAI), aim attributing network s prediction specific features input domain. XAI are usually assessed by using benchmark datasets (like MNIST or ImageNet for image classification). However, an objective, theoretically derived ground truth attribution is lacking most these datasets, making assessment cases subjective. Also, specifically designed geosciences rare. Here, we provide a framework, based on use additively separable functions, generate regression known priori. We large dataset train fully connected learn underlying function that was used simulation. then compare estimated heatmaps from order identify examples where perform well poorly. believe benchmarks as ones herein great importance further more objective accurate implementation methods, will increase assist discovering new science.

Language: Английский

Citations

77
Theory-guided hard constraint projection (HCP): A knowledge-based data-driven scientific machine learning method DOI Creative Commons
Yuntian Chen, Dou Huang, Dongxiao Zhang

et al.

Journal of Computational Physics, Journal Year: 2021, Volume and Issue: 445, P. 110624 - 110624

Published: Aug. 10, 2021

Machine learning models have been successfully used in many scientific and engineering fields. However, it remains difficult for a model to simultaneously utilize domain knowledge experimental observation data. The application of knowledge-based symbolic AI represented by an expert system is limited the expressive ability model, data-driven connectionism neural networks prone produce predictions that violate physical mechanisms. In order fully integrate with observations, make full use prior information strong fitting networks, this study proposes theory-guided hard constraint projection (HCP). This converts constraints, such as governing equations, into form easy handle through discretization, then implements optimization projection. Based on rigorous mathematical proofs, HCP can ensure strictly conform mechanisms patch. performance verified experiments based heterogeneous subsurface flow problem. Due compared connected soft models, physics-informed requires fewer data, achieves higher prediction accuracy stronger robustness noisy observations.

Language: Английский

Citations

93
D-SRGAN: DEM Super-Resolution with Generative Adversarial Networks DOI
Bekir Zahit Demiray, Muhammed Sit, İbrahim Demir

et al.

SN Computer Science, Journal Year: 2021, Volume and Issue: 2(1)

Published: Jan. 20, 2021

Language: Английский

Citations

88
A New Benchmark on Machine Learning Methodologies for Hydrological Processes Modelling: A Comprehensive Review for Limitations and Future Research Directions DOI Open Access
Zaher Mundher Yaseen‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬

Knowledge-Based Engineering and Sciences, Journal Year: 2023, Volume and Issue: 4(3), P. 65 - 103

Published: Dec. 31, 2023

The best practice of watershed management is through the understanding hydrological processes. As a matter fact, processes are highly associated with stochastic, non-linear, and non-stationary phenomena. Hydrological simulation modeling challenging issues in domains hydrology, climate environment. Hence, development machine learning (ML) models for solving those complex problems took essential place over past couple decades. It can be observed, data availability has increased remarkably, thus computational resources led to resurgence ML models’ development. been witnessed huge efforts on using facility several review researches have conducted. Literature studies approved capacity field hydrology classical “traditional models” based their forecastability, flexibility, precision, generalization, execution convergence speed. However, although potential merits were observed model’s development, limitations allied such as interpretability black-box models, practicality management, difficulty explain physical In this survey, an exhibition all published articles recognize research gaps direction. ultimate aim current survey establish new milestone interested environment researchers applications models.

Language: Английский

Citations

24
Satellite Video Remote Sensing for Flood Model Validation DOI Creative Commons
Christopher Masafu, Richard Williams

Water Resources Research, Journal Year: 2024, Volume and Issue: 60(1)

Published: Jan. 1, 2024

Abstract Satellite‐based optical video sensors are poised as the next frontier in remote sensing. Satellite offers unique advantage of capturing transient dynamics floods with potential to supply hitherto unavailable data for assessment hydraulic models. A prerequisite successful application models is their proper calibration and validation. In this investigation, we validate 2D flood model predictions using satellite video‐derived extents velocities. Hydraulic simulations a event 5‐year return period (discharge 722 m 3 s −1 ) were conducted Hydrologic Engineering Center—River Analysis System Darling River at Tilpa, Australia. To extract from studied event, use hybrid transformer‐encoder, convolutional neural network (CNN)‐decoder deep network. We evaluate influence test‐time augmentation (TTA)—the transformations on test image ensembles, during inference. employ Large Scale Particle Image Velocimetry (LSPIV) non‐contact‐based river surface velocity estimation sequential frames. When validating segmented extents, critical success index peaked 94% an average relative improvement 9.5% when TTA was implemented. show that significant value network‐based segmentation, compensating aleatoric uncertainties. The correlations between LSPIV velocities reasonable averaged 0.78. Overall, our investigation demonstrates space‐based studying dynamics.

Language: Английский

Citations

11
Comparative Assessment of Artificial Neural Networks (ANNs), Long Short Term Memory Network (LSTM) and Hydrologic Engineering Centre-Hydrologic Modelling System (HEC-HMS) for Runoff Modelling DOI

Aparna M. Deulkar,

Pradnya Dixit, Shreenivas Londhe

et al.

Water Resources Management, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 16, 2025

Language: Английский

Citations

1
Application of Random Forest Algorithm for Merging Multiple Satellite Precipitation Products across South Korea DOI Creative Commons
Giang V. Nguyen, Xuan-Hien Le, Linh Nguyen Van

et al.

Remote Sensing, Journal Year: 2021, Volume and Issue: 13(20), P. 4033 - 4033

Published: Oct. 9, 2021

Precipitation is a crucial component of the water cycle and plays key role in hydrological processes. Recently, satellite-based precipitation products (SPPs) have provided grid-based with spatiotemporal variability. However, SPPs contain lot uncertainty estimated precipitation, spatial resolution these still relatively coarse. To overcome limitations, this study aims to generate new daily based on combination rainfall observation data multiple for period 2003–2017 across South Korea. A Random Forest (RF) machine-learning algorithm model was applied producing merged product. In addition, several statistical linear merging methods been adopted compare results achieved from RF model. investigate efficiency RF, 64 observed Automated Synoptic Observation System (ASOS) installations were collected analyze accuracy through continuous as well categorical indicators. The values produced by procedure generally not only report higher than single satellite product but also indicate that more effective method. Thus, achievements point out might be products, especially sparse region areas.

Language: Английский

Citations

47
Application of Large Language Models in Developing Conversational Agents for Water Quality Education, Communication and Operations DOI Creative Commons
R. Dinesh Jackson Samuel,

Muhammed Sermet,

Jerry Mount

et al.

EarthArXiv (California Digital Library), Journal Year: 2024, Volume and Issue: unknown

Published: May 1, 2024

The rapid advancement of Large Language Models (LLMs), such as ChatGPT, has opened new horizons in the field Artificial Intelligence (AI), revolutionizing way we can engage with and disseminate complex information. This paper presents an innovative application ChatGPT domain Water Quality (WQ) management, through development AI Hub. Hub encompasses a suite conversational agents, each designed to address different aspects water quality including nitrogen pollution, local issues, actionable planning for conservation. These agents utilize advanced natural language processing capabilities complemented quality-related data, provide users accurate, up-to-date, contextually relevant objective is empower communities knowledge necessary understand challenges effectively. Our comprehensive evaluation these demonstrates their proficiency delivering valuable insights, overall performance accuracy exceeding 89%. underscores potential AI-enabled platforms enhancing public understanding engagement environmental conservation efforts. By bridging gap between data awareness, sets precedent sustainable management.

Language: Английский

Citations

7
Long-run forecasting surface and groundwater dynamics from intermittent observation data: An evaluation for 50 years DOI Creative Commons
M.T. Vu, Abderrahim Jardani, Nicolas Masséi

et al.

The Science of The Total Environment, Journal Year: 2023, Volume and Issue: 880, P. 163338 - 163338

Published: April 5, 2023

Language: Английский

Citations

15
Real-time streamflow forecasting: AI vs. Hydrologic insights DOI Creative Commons
Witold F. Krajewski, Ganesh R. Ghimire, İbrahim Demir

et al.

Journal of Hydrology X, Journal Year: 2021, Volume and Issue: 13, P. 100110 - 100110

Published: Nov. 23, 2021

In this paper, we propose a set of simple benchmarks for the evaluation data-based models real-time streamflow forecasting, such as those developed with sophisticated Artificial Intelligence (AI) algorithms. The are also and provide context to judge incremental improvements in performance metrics from more complicated approaches. include temporal spatial persistence, persistence corrected baseflow streamflow, well river distance weighted runoff obtained space-time distributed rainfall. development benchmarks, use basic hydrologic insights flow aggregation by network, scale-dependence basin response, partitioning into quick baseflow, water travel time, rainfall averaging width function. study uses 140 gauges Iowa that cover range scales between 7 37,000 km2. data 17 years. This work demonstrates proposed can good according several commonly used metrics. For example, forecasting at half test locations across years achieves Kling-Gupta Efficiency (KGE) score 0.6 or higher one-day ahead lead 20% cases reach KGE 0.8 higher. easy implement should prove useful developers physics-based assimilation techniques.

Language: Английский

Citations

28