Temporal and Spatial Satellite Data Augmentation for Deep Learning-Based Rainfall Nowcasting DOI Creative Commons
Özlem Baydaroğlu, İbrahim Demir

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

Опубликована: Окт. 3, 2023

Climate change has been associated with alterations in precipitation patterns and increased vulnerability to floods droughts. The need for improvements forecasting monitoring approaches become imperative due flash severe flooding. Rainfall prediction is a challenging but critical issue owing the complexity of atmospheric processes, spatial temporal variability rainfall, dependency this on several nonlinear factors. Because excessive rainfall cause natural disasters such as landslides, accurate real-time nowcast necessary precautions, control, planning. In study, nowcasting studied utilizing NASA Giovanni satellite-derived products convolutional long short-term memory (ConvLSTM) approach, which variation LSTM. Due data requirements deep learning-based methods, augmentation performed using interpolation techniques. study utilized three types data, including spatial, temporal, spatio-temporal interpolated conduct comparative analysis results obtained through rainfall. This research examines two catastrophic that transpired Türkiye Marmara Region 2009 Central Black Sea 2021, are selected focal case studies. It also explores suitability model various flood events, while examining impact nowcast.

Язык: Английский

MA-SARNet: A one-shot nowcasting framework for SAR image prediction with physical driving forces DOI
Zhouyayan Li, Zhongrun Xiang, Bekir Zahit Demiray

и другие.

ISPRS Journal of Photogrammetry and Remote Sensing, Год журнала: 2023, Номер 205, С. 176 - 190

Опубликована: Окт. 12, 2023

Язык: Английский

Процитировано

15
A Systematic Review of Deep Learning Applications in Streamflow Data Augmentation and Forecasting DOI Creative Commons
Muhammed Sit, Bekir Zahit Demiray, İbrahim Demir

и другие.

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

Опубликована: Сен. 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.

Язык: Английский

Процитировано

9
MA-SARNet: A one-shot nowcasting framework for SAR image prediction with physical driving forces DOI Creative Commons
Zhouyayan Li, Zhongrun Xiang, Bekir Zahit Demiray

и другие.

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

Опубликована: Дек. 19, 2022

Remote sensing imagery is one of the most widely used data sources for large-scale Earth observations with consistent spatial and temporal quality. However, current usage scenarios Earth’s surface remote images, such as those generated from Landsat, Sentinel 2, SAR, are largely limited to retrospective tasks, they often reconstruct existing phenomena, land use change, flood inundation, wildfire. This study proposes MA-SARNet, a one-shot nowcasting framework built modified MA-Net structure ResNet50 backbone, predict backscatter values Synthetic-Aperture Radar (SAR) images using previous SAR observations, precipitation, soil moisture, geomorphic layers input. The model was trained, validated, tested collected during catastrophic 2019 Midwest U.S. Floods that affected several states on Missouri Mississippi River tributaries. Compared benchmark performance, predictions show an increase 31.9% 17.8% mean median AAI (Assemble Accuracy Index) 37.9% 15.1% NSE (Nash-Sutcliffe Efficiency) test set. Results showed extent derived has less misclassifications caused by pixel-level noise compared map real backscatters. robustness tests demonstrate sufficient generalization potential does not require further fine-tuning work new data, therefore proves its usefulness in real-time prediction tasks aimed at fast response mitigation upcoming floods tight time schedule.

Язык: Английский

Процитировано

4
Democratizing Deep Learning Applications in Earth and Climate Sciences on the Web: EarthAIHub DOI Creative Commons
Muhammed Sit, İbrahim Demir

Applied Sciences, Год журнала: 2023, Номер 13(5), С. 3185 - 3185

Опубликована: Март 2, 2023

Most deep learning application studies have limited accessibility and reproducibility for researchers students in many domains, especially earth climate sciences. In order to provide a step towards improving the of models such disciplines, this study presents community-driven framework repository, EarthAIHub, that is powered by TensorFlow.js, where can be tested run without extensive technical knowledge. achieve this, we present configuration data specification form middleware, an abstraction layer, between models. Once easy-to-create file generated model user, EarthAIHub seamlessly makes publicly available testing access using web platform. The platform community-enabled repository will benefit who are new domain enabling them test existing community with their datasets, share novel community. help before adapting research learn about model’s details performance.

Язык: Английский

Процитировано

2
Temporal and Spatial Satellite Data Augmentation for Deep Learning-Based Rainfall Nowcasting DOI Creative Commons
Özlem Baydaroğlu, İbrahim Demir

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

Опубликована: Окт. 3, 2023

Climate change has been associated with alterations in precipitation patterns and increased vulnerability to floods droughts. The need for improvements forecasting monitoring approaches become imperative due flash severe flooding. Rainfall prediction is a challenging but critical issue owing the complexity of atmospheric processes, spatial temporal variability rainfall, dependency this on several nonlinear factors. Because excessive rainfall cause natural disasters such as landslides, accurate real-time nowcast necessary precautions, control, planning. In study, nowcasting studied utilizing NASA Giovanni satellite-derived products convolutional long short-term memory (ConvLSTM) approach, which variation LSTM. Due data requirements deep learning-based methods, augmentation performed using interpolation techniques. study utilized three types data, including spatial, temporal, spatio-temporal interpolated conduct comparative analysis results obtained through rainfall. This research examines two catastrophic that transpired Türkiye Marmara Region 2009 Central Black Sea 2021, are selected focal case studies. It also explores suitability model various flood events, while examining impact nowcast.

Язык: Английский

Процитировано

2