A comprehensive review of deep learning applications in hydrology and water resources

Sit, Muhammed; Demiray, Bekir Zahit; Xiang, Zhongrun; Ewing, Gregory J.; Sermet, Yusuf; Demir, İbrahim

doi:10.2166/wst.2020.369

Cited by 287 publications

(125 citation statements)

References 181 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Artificial neural networks are algorithms applied to map features into a series of outputs. Through a structure of the input, output and intermediate hidden layers, artificial neural networks can learn data relationships between input and output data [22]. A feedforward neural network is applied in this work for modeling the study area, proceeding and transmitting data in a network structure [23].…”

Section: Data and Structure Of Artificial Neural Networkmentioning

confidence: 99%

Multistep Flood Inundation Forecasts with Resilient Backpropagation Neural Networks: Kulmbach Case Study

Lin

Leandro

Gerber

et al. 2020

Water

View full text Add to dashboard Cite

Flooding, a significant natural disaster, attracts worldwide attention because of its high impact on communities and individuals and increasing trend due to climate change. A flood forecast system can minimize the impacts by predicting the flood hazard before it occurs. Artificial neural networks (ANN) could efficiently process large amounts of data and find relations that enable faster flood predictions. The aim of this study is to perform multistep forecasts for 1–5 h after the flooding event has been triggered by a forecast threshold value. In this work, an ANN developed for the real-time forecast of flood inundation with a high spatial resolution (4 m × 4 m) is extended to allow for multiple forecasts. After trained with 120 synthetic flood events, the ANN was first tested with 60 synthetic events for verifying the forecast performance for 3 h, 6 h, 9 h and 12 h lead time. The model produces good results, as shown by more than 81% of all grids having an RMSE below 0.3 m. The ANN is then applied to the three historical flood events to test the multistep inundation forecast. For the historical flood events, the results show that the ANN outputs have a good forecast accuracy of the water depths for (at least) the 3 h forecast with over 70% accuracy (RMSE within 0.3 m), and a moderate accuracy for the subsequent forecasts with (at least) 60% accuracy.

show abstract

Section: Data and Structure Of Artificial Neural Networkmentioning

confidence: 99%

Multistep Flood Inundation Forecasts with Resilient Backpropagation Neural Networks: Kulmbach Case Study

Lin

Leandro

Gerber

et al. 2020

Water

View full text Add to dashboard Cite

show abstract

“…It shows that the best reconstruction performance can be obtained when the coastal video is used for fine-tuning even in the same model architecture of Raindrop-aware GAN. By creating timestack image and visually assessment it, we can confirm that the performance of the proposed method is the best and it also has high applicability in studying nearshore wave dynamics with video remote sensing, in particular data preparation step, such as breaking wave height estimation from coastal video [31,32], video sensing of nearshore bathymetry evolution [33,34], nearshore wave transform with video imagery [35], shoreline response and resilience through video monitoring [36][37][38][39], wave run-up prediction [40,41], and nearshore wave tracking through coastal video [42,43].…”

Section: Discussionmentioning

confidence: 60%

Raindrop-Aware GAN: Unsupervised Learning for Raindrop-Contaminated Coastal Video Enhancement

et al. 2020

View full text Add to dashboard Cite

We propose an unsupervised network with adversarial learning, the Raindrop-aware GAN, which enhances the quality of coastal video images contaminated by raindrops. Raindrop removal from coastal videos faces two main difficulties: converting the degraded image into a clean one by visually removing the raindrops, and restoring the background coastal wave information in the raindrop regions. The components of the proposed network—a generator and a discriminator for adversarial learning—are trained on unpaired images degraded by raindrops and clean images free from raindrops. By creating raindrop masks and background-restored images, the generator restores the background information in the raindrop regions alone, preserving the input as much as possible. The proposed network was trained and tested on an open-access dataset and directly collected dataset from the coastal area. It was then evaluated by three metrics: the peak signal-to-noise ratio, structural similarity, and a naturalness-quality evaluator. The indices of metrics are 8.2% (+2.012), 0.2% (+0.002), and 1.6% (−0.196) better than the state-of-the-art method, respectively. In the visual assessment of the enhanced video image quality, our method better restored the image patterns of steep wave crests and breaking than the other methods. In both quantitative and qualitative experiments, the proposed method more effectively removed the raindrops in coastal video and recovered the damaged background wave information than state-of-the-art methods.

show abstract

“…Researchers of deep learning in hydrology and water resources have increased significantly in the past several years (Sit et al, 2020), but there are limited considerations on physical features and processes. As shown in this study, deep learning models considering watershed-scale physical features and semi-distributed structures can work on multiple watersheds using regional models with limited sacrifice of accuracy.…”

Section: Resultsmentioning

confidence: 99%

A Regional Semi-Distributed Streamflow Model Using Deep Learning

Xiang¹,

Demir²,

Mantilla³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recent studies have shown that deep learning models in hydrological applications significantly improved streamflow predictions at multiple stations compared to traditional machine learning approaches. However, most studies lack generalization; i.e. researchers are training separate models for each location. The spatial and temporal generalization ability of deep learning models in hydrology that can be gained by training a single model for multiple stations is evaluated in this study. We developed a generalized model with a multi-site structure for hourly streamflow hindcasts on 125 USGS gauges. Considering watershed-scale features including drainage area, time of concentration, slope, and soil types, the proposed models have acceptable performance and slightly higher median NSE value than training individual models for each USGS station. Furthermore, we showed that the trained generalized model can be applied to any new gauge in the state of Iowa that was not used in the training set with acceptable accuracy. This study demonstrates the potential of deep learning studies in hydrology where more domain knowledge and physical features can support further generalization.

show abstract

A comprehensive review of deep learning applications in hydrology and water resources

Cited by 287 publications

References 181 publications

Multistep Flood Inundation Forecasts with Resilient Backpropagation Neural Networks: Kulmbach Case Study

Multistep Flood Inundation Forecasts with Resilient Backpropagation Neural Networks: Kulmbach Case Study

Raindrop-Aware GAN: Unsupervised Learning for Raindrop-Contaminated Coastal Video Enhancement

A Regional Semi-Distributed Streamflow Model Using Deep Learning

Contact Info

Product

Resources

About