Enhancing Hydrological Modeling with Transformers: A Case Study for 24-Hour Streamflow Prediction DOI Creative Commons
Bekir Zahit Demiray, Muhammed Sit, Omer Mermer

и другие.

EarthArXiv (California Digital Library), Год журнала: 2023, Номер unknown

Опубликована: Сен. 12, 2023

In this paper, we address the critical task of 24-hour streamflow forecasting using advanced deep-learning models, with a primary focus on Transformer architecture which has seen limited application in specific task. We compare performance five different including Persistence, LSTM, Seq2Seq, GRU, and Transformer, across four distinct regions. The evaluation is based three metrics: Nash-Sutcliffe Efficiency (NSE), Pearson’s r, Normalized Root Mean Square Error (NRMSE). Additionally, investigate impact two data extension methods: zero-padding persistence, model's predictive capabilities. Our findings highlight Transformer's superiority capturing complex temporal dependencies patterns data, outperforming all other models terms both accuracy reliability. study's insights emphasize significance leveraging deep learning techniques, such as hydrological modeling for effective water resource management flood prediction.

Язык: Английский

Long-term streamflow forecasting in data-scarce regions: Insightful investigation for leveraging satellite-derived data, Informer architecture, and concurrent fine-tuning transfer learning DOI
Fatemeh Ghobadi, Zaher Mundher Yaseen‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬,

Doosun Kang

и другие.

Journal of Hydrology, Год журнала: 2024, Номер 631, С. 130772 - 130772

Опубликована: Фев. 2, 2024

Язык: Английский

Процитировано

17
Predicting Harmful Algal Blooms Using Explainable Deep Learning Models: A Comparative Study DOI Open Access
Bekir Zahit Demiray, Omer Mermer, Özlem Baydaroğlu

и другие.

Water, Год журнала: 2025, Номер 17(5), С. 676 - 676

Опубликована: Фев. 26, 2025

Harmful algal blooms (HABs) have emerged as a significant environmental challenge, impacting aquatic ecosystems, drinking water supply systems, and human health due to the combined effects of activities climate change. This study investigates performance deep learning models, particularly Transformer model, there are limited studies exploring its effectiveness in HAB prediction. The chlorophyll-a (Chl-a) concentration, commonly used indicator phytoplankton biomass proxy for occurrences, is target variable. We consider multiple influencing parameters—including physical, chemical, biological quality monitoring data from stations located west Lake Erie—and employ SHapley Additive exPlanations (SHAP) values an explainable artificial intelligence (XAI) tool identify key input features affecting HABs. Our findings highlight superiority especially Transformer, capturing complex dynamics parameters providing actionable insights ecological management. SHAP analysis identifies Particulate Organic Carbon, Nitrogen, total phosphorus critical factors predictions. contributes development advanced predictive models HABs, aiding early detection proactive management strategies.

Язык: Английский

Процитировано

2
Enhancing hydrological modeling with transformers: a case study for 24-h streamflow prediction DOI Creative Commons
Bekir Zahit Demiray, Muhammed Sit, Omer Mermer

и другие.

Water Science & Technology, Год журнала: 2024, Номер 89(9), С. 2326 - 2341

Опубликована: Апрель 4, 2024

ABSTRACT In this paper, we address the critical task of 24-h streamflow forecasting using advanced deep-learning models, with a primary focus on transformer architecture which has seen limited application in specific task. We compare performance five different including persistence, long short-term memory (LSTM), Seq2Seq, GRU, and transformer, across four distinct regions. The evaluation is based three metrics: Nash–Sutcliffe Efficiency (NSE), Pearson's r, normalized root mean square error (NRMSE). Additionally, investigate impact two data extension methods: zero-padding model's predictive capabilities. Our findings highlight transformer's superiority capturing complex temporal dependencies patterns data, outperforming all other models terms both accuracy reliability. Specifically, model demonstrated substantial improvement NSE scores by up to 20% compared models. study's insights emphasize significance leveraging deep learning techniques, such as hydrological modeling for effective water resource management flood prediction.

Язык: Английский

Процитировано

6
Integrated Explainable Ensemble Machine Learning Prediction of Injury Severity in Agricultural Accidents DOI Creative Commons
Omer Mermer,

Eddie Zhang,

İbrahim Demir

и другие.

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Фев. 7, 2025

Abstract Agricultural injuries remain a significant occupational hazard, causing substantial human and economic losses worldwide. This study investigates the prediction of agricultural injury severity using both linear ensemble machine learning (ML) models applies explainable AI (XAI) techniques to understand contribution input features. Data from AgInjuryNews (2015–2024) was preprocessed extract relevant attributes such as location, time, age, safety measures. The dataset comprised 2,421 incidents categorized fatal or non-fatal. Various ML models, including Naïve Bayes (NB), Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting (GB), were trained evaluated standard performance metrics. Ensemble demonstrated superior accuracy recall compared with XGBoost achieving 100% for injuries. However, all faced challenges in predicting non-fatal due class imbalance. SHAP analysis provided insights into feature importance, gender, time emerging most influential predictors across models. research highlights effectiveness while emphasizing need balanced datasets XAI actionable insights. findings have practical implications enhancing guiding policy interventions. Highlights analyzed (2015– 2024) utilized predict severity, focusing on outcomes. Forest, outperformed recall, especially injuries, although predictions imbalance observed. Key identified through included providing interpretable factors influencing severity. integration enhanced transparency predictions, enabling stakeholders prioritize targeted interventions effectively. potential combining improve practices provides foundation addressing data future studies. Graphical

Язык: Английский

Процитировано

0
Application of hydrological models in climate change framework for a river basin in India DOI Creative Commons
Rishith Kumar Vogeti,

K. Srinivasa Raju,

D. Nagesh Kumar

и другие.

Journal of Water and Climate Change, Год журнала: 2023, Номер 14(9), С. 3150 - 3165

Опубликована: Авг. 28, 2023

Abstract Soil Water Assessment Tool (SWAT), Hydrologic Engineering Center-Hydrologic Modelling System (HEC-HMS), and Simulation Program Fortran (HSPF) are explored for streamflow simulation of Lower Godavari Basin, India. The simulating ability models is evaluated using four indicators. SWAT has shown exceptional in calibration validation compared to the other two. Accordingly, used climate change framework an ensemble 13 Global Climate Models Shared Socioeconomic Pathways (SSPs). Three time segments, near-future (2021–2046), mid-future (2047–2072), far-future (2073–2099), considered analysis. Four SSPs show a substantial increase historical period (1982–2020). These deviations range from 17.14 (in SSP245) 28.35% SSP126) (near-future), 31.32 (SSP370) 43.28% (SSP585) (mid-future), 30.41 (SSP126) 70.8% (far-future). Across all timescales covering 948 months, highest projected streamflows observed SSP126, SSP245, SSP370, SSP585 were 4962.36, 6,108, 6,821, 6,845 m3/s, respectively. Efforts also made appraise influence multi-model combinations on streamflow. present study expected provide platform holistic decision-making, which helps develop efficient basin planning management alternatives.

Язык: Английский

Процитировано

8
Evaluating the impact of improved filter-wrapper input variable selection on Long-term runoff forecasting using local and global climate information DOI
Binlin Yang, Chen Lu, Bin Yi

и другие.

Journal of Hydrology, Год журнала: 2024, Номер unknown, С. 132034 - 132034

Опубликована: Сен. 1, 2024

Язык: Английский

Процитировано

2
A Systematic Review of Deep Learning Applications in Interpolation and Extrapolation of Precipitation Data DOI Creative Commons
Muhammed Sit, Bekir Zahit Demiray, İbrahim Demir

и другие.

EarthArXiv (California Digital Library), Год журнала: 2022, Номер unknown

Опубликована: Ноя. 17, 2022

With technological enhancements, the volume, velocity, and variety (3Vs) of raw digital Earth data have increased in recent years. Due to availability computer resources growing popularity deep learning applications, this has been a crucial source for data-driven studies that transformed fields climate earth science. One critical sources is precipitation supporting science on modeling, forecasting, preparedness extreme events (i.e., floods, droughts, pollution transport). In study, we worked an extensive review manuscripts focusing use methods tackle challenges either improve quality or extrapolate (forecast) rainfall datasets. The purpose study summarize most developments approaches forecasting improving datasets, as well highlighting issues, shortcomings, open questions with insightful recommendations future directions.

Язык: Английский

Процитировано

5
Spatial Downscaling of Streamflow Data with Attention Based Spatio-Temporal Graph Convolutional Networks DOI Creative Commons
Muhammed Sit, Bekir Zahit Demiray, İbrahim Demir

и другие.

EarthArXiv (California Digital Library), Год журнала: 2023, Номер unknown

Опубликована: Март 31, 2023

Accurate streamflow data is vital for various climate modeling applications, including flood forecasting. However, many streams lack sufficient monitoring due to the high operational costs involved. To address this issue and promote enhanced disaster preparedness, management, response, our study introduces a neural network-based method estimating historical hourly in two spatial downscaling scenarios. The targets types of ungauged locations: (1) those without sensors sparsely gauged river networks, (2) that previously had sensor, but gauge no longer available. For both cases, we propose ScaleGNN, graph network based on Attention-Based Spatio-Temporal Graph Convolutional Networks (ASTGCN). We evaluate performance ScaleGNN against Long Short-Term Memory (LSTM) baseline persistence discharge values over 36-hour period. Our findings indicate surpasses first scenario, while approaches demonstrate their effectiveness compared second scenario.

Язык: Английский

Процитировано

2
Enhancing Hydrological Modeling with Transformers: A Case Study for 24-Hour Streamflow Prediction DOI Creative Commons
Bekir Zahit Demiray, Muhammed Sit, Omer Mermer

и другие.

EarthArXiv (California Digital Library), Год журнала: 2023, Номер unknown

Опубликована: Сен. 12, 2023

In this paper, we address the critical task of 24-hour streamflow forecasting using advanced deep-learning models, with a primary focus on Transformer architecture which has seen limited application in specific task. We compare performance five different including Persistence, LSTM, Seq2Seq, GRU, and Transformer, across four distinct regions. The evaluation is based three metrics: Nash-Sutcliffe Efficiency (NSE), Pearson’s r, Normalized Root Mean Square Error (NRMSE). Additionally, investigate impact two data extension methods: zero-padding persistence, model's predictive capabilities. Our findings highlight Transformer's superiority capturing complex temporal dependencies patterns data, outperforming all other models terms both accuracy reliability. study's insights emphasize significance leveraging deep learning techniques, such as hydrological modeling for effective water resource management flood prediction.

Язык: Английский

Процитировано

0