Towards deep learning based flood forecasting for ungauged basins

Floods are among the most destructive natural hazards in the world. To reduce flood induced damages and casualties, streamflow forecasts should be as accurate as possible.As of today, streamflow forecasts are usually made with either conceptual or process-based hydrological models. The problem these models usually have is that they perform best when calibrated for a specific basin, and performance degrades drastically if the models are used in places without historic streamflow measurements. To make things worse, some of the most devastating floods occur in developing and low-income countries, where historic records of streamflow measurements are scarce. Therefore, a central task for enhancing flood forecasts and helping local authorities to manage these areas is to provide high-quality streamflow forecasts in ungauged rivers. Although the IAHS dedicated an entire decade (2003-2012) to advance the problem of Prediction in Ungauged Basins the central goal remains largely a challenge.In this talk, we will present a novel approach for tackling the problem of prediction in ungauged basins using a data-driven approach. More concretely, we show that the Long Short-Term Memory network (LSTM), which is a special type of a deep learning model, can serve as a generalizable rainfall-runoff simulation model. We will present recent results indicating that the LSTM gives on average better out-of-sample predictions (ungauged prediction) than e.g. the SAC-SMA in-sample (gauged) or the US National Water Model (Kratzert et al., 2019).One place where these research results are already finding their way into operation is Google&#8217;s Flood Forecasting Initiative. The goal of this initiative is to provide (enhanced) flood warnings, where needed, starting with a pilot project in India. And as mentioned above, historic streamflow records in those regions are scarce, which motivates new and innovative approaches for enhanced streamflow forecasting.References:Kratzert, F., Klotz, D., Herrnegger, M., Sampson, A. K., Hochreiter, S., & Nearing, G. S.: Toward improved predictions in ungauged basins: Exploiting the power of machine learning. Water Resources Research, 55. https://doi.org/10.1029/2019WR026065, 2019.

Download Full-text

A novel artificial neural network for flood forecasting based on deep learning encoder-decoder architecture

10.5194/egusphere-egu2020-6277 ◽

2020 ◽

Author(s):

Kangling Lin ◽

Hua Chen ◽

Chong-Yu Xu ◽

Yanlai Zhou ◽

Shenglian Guo

Keyword(s):

Neural Network ◽

Deep Learning ◽

Goodness Of Fit ◽

Short Term Memory ◽

Flood Forecasting ◽

Learning Ability ◽

Convolutional Network ◽

Study Results ◽

Decoder Architecture ◽

Artificial Neural

With the rapid growth of deep learning recently, artificial neural networks have been propelled to the forefront in flood forecasting via their end-to-end learning ability. Encoder-decoder architecture, as a novel deep feature extraction, which captures the inherent relationship of the data involved, has emerged in time sequence forecasting nowadays. As the advance of encoder-decoder architecture in sequence to sequence learning, it has been applied in many fields, such as machine translation, energy and environment. However, it is seldom used in hydrological modelling. In this study, a new neural network is developed to forecast flood based on the encoder-decoder architecture. There are two deep learning methods, including the Long Short-Term Memory (LSTM) network and Temporal Convolutional Network (TCN), selected as encoders respectively, while the LSTM was also chosen as the decoder, whose results are compared with those from the standard LSTM without using encoder-decoder architecture.These models were trained and tested by using the hourly flood events data from 2009 to 2015 in Jianxi basin, China. The results indicated that the new neural flood forecasting networks based encoder-decoder architectures generally perform better than the standard LSTM, since they have better goodness-of-fit between forecasted and observed flood and produce the promising performance in multi-index assessment. The TCN as an encoder has better model stability and accuracy than LSTM as an encoder, especially in longer forecast periods and larger flood. The study results also show that the encoder-decoder architecture can be used as an effective deep learning solution in flood forecasting.

Download Full-text

Deep Learning with a Long Short-Term Memory Networks Approach for Rainfall-Runoff Simulation

Water ◽

10.3390/w10111543 ◽

2018 ◽

Vol 10 (11) ◽

pp. 1543 ◽

Cited By ~ 59

Author(s):

Caihong Hu ◽

Qiang Wu ◽

Hui Li ◽

Shengqi Jian ◽

Nan Li ◽

...

Keyword(s):

Deep Learning ◽

Short Term Memory ◽

Network Models ◽

Machine Learning Techniques ◽

Rainfall Runoff ◽

Ann Model ◽

Runoff Simulation ◽

Simulation Performance ◽

Artificial Neural Network Ann ◽

Runoff Events

Considering the high random and non-static property of the rainfall-runoff process, lots of models are being developed in order to learn about such a complex phenomenon. Recently, Machine learning techniques such as the Artificial Neural Network (ANN) and other networks have been extensively used by hydrologists for rainfall-runoff modelling as well as for other fields of hydrology. However, deep learning methods such as the state-of-the-art for LSTM networks are little studied in hydrological sequence time-series predictions. We deployed ANN and LSTM network models for simulating the rainfall-runoff process based on flood events from 1971 to 2013 in Fen River basin monitored through 14 rainfall stations and one hydrologic station in the catchment. The experimental data were from 98 rainfall-runoff events in this period. In between 86 rainfall-runoff events were used as training set, and the rest were used as test set. The results show that the two networks are all suitable for rainfall-runoff models and better than conceptual and physical based models. LSTM models outperform the ANN models with the values of R 2 and N S E beyond 0.9, respectively. Considering different lead time modelling the LSTM model is also more stable than ANN model holding better simulation performance. The special units of forget gate makes LSTM model better simulation and more intelligent than ANN model. In this study, we want to propose new data-driven methods for flood forecasting.

Download Full-text

Using MODIS estimates of fractional snow cover extent to improve streamflow forecasts in Interior Alaska

10.5194/hess-2018-96 ◽

2018 ◽

Author(s):

Katrina E. Bennett ◽

Jessica E. Cherry ◽

Ben Balk ◽

Scott Lindsey

Keyword(s):

Snow Cover ◽

Remotely Sensed ◽

Poor Quality ◽

Water Model ◽

Streamflow Forecasting ◽

Snow Cover Area ◽

Snow Cover Extent ◽

The Us ◽

Physically Based ◽

Streamflow Forecasts

Abstract. Remotely sensed snow cover observations provide an opportunity to improve operational snowmelt and streamflow forecasting in remote regions. This is particularly true in Alaska, where remote basins and a spatially and temporally sparse gaging network plague efforts to understand and forecast the hydrology of subarctic boreal watersheds and where climate change is leading to rapid shifts in watershed function. In this study, the operational framework employed by the US National Weather Service, including the Alaska Pacific River Forecast Center, is adapted to integrate Moderate Resolution Imaging Spectroradiometer (MODIS) remotely sensed observations of snow cover extent (SCE) to determine if these data improve streamflow forecasts in Interior Alaskan river basins. Two versions of MODIS fractional SCE are tested in this study: the MODIS 10A1 (MOD10A1), and the MODIS Snow Cover Area and Grain size (MODSCAG) product. Observed runoff is compared to simulated runoff to calibrate both iterations of the model. MODIS-forced runs have improved snow depletion timing compared with snow telemetry sites in the basins, with discernable increases in skill for the streamflow simulations. The MODSCAG SCE version provides moderate increases in skill, but is similar to the MOD10A1 results in these watersheds. The basins with the greatest improvement in streamflow simulations have the sparsest streamflow observations. Considering the numerous low-quality gages (discontinuous, short, or unreliable) and ungaged systems throughout the high latitude regions of the globe, this result is of great value and indicates the utility of the MODIS SCE data in these regions. Additionally, while improvements in predicted discharge values are subtle, the snow model better represents the physical conditions of the snow pack and therefore provides more robust simulations, which are consistent with the US National Weather Service's move toward a physically-based National Water Model. Physically-based models may be more capable of adapting to changing climates than statistical models tuned to past regimes. This work provides direction for both the Alaska Pacific River Forecast Center and other forecast centers across the US to implement remote sensing observations within their operational framework, to refine the representation of snow, and to improve streamflow forecasting skill in basins with few or poor-quality observations.

Download Full-text

Flood Stage Forecasting at the Gurye-Gyo Station in Sumjin River Using LSTM-Based Deep Learning Models

Korean Society of Hazard Mitigation ◽

10.9798/kosham.2021.21.3.193 ◽

2021 ◽

Vol 21 (3) ◽

pp. 193-201

Author(s):

Jaewon Jung ◽

Hyelim Mo ◽

Junhyeong Lee ◽

Younghoon Yoo ◽

Hung Soo Kim

Keyword(s):

Deep Learning ◽

Short Term Memory ◽

Warning System ◽

Flood Forecasting ◽

Flood Damage ◽

Support Vector ◽

Forecasting Models ◽

Physically Based ◽

Forecasting And Warning ◽

Storm Rainfall

Instances of flood damage caused by extreme storm rainfall due to climate change and variability have been showing an increasing trend. Particularly, a flood forecasting and warning system has been recognized as an important nonstructural measure for flood damage reduction, including loss of life. Flood forecasting and warning have been performed by the forecasts of flood discharge and flood stage using the physically based rainfall-runoff models. However, recently, studies involving the application of a machine learning-based flood forecasting models, which addresses the limitations of extant physically based flood stage forecasting models, have been performed. We may require various case studies to determine more accurate methods. Therefore, this study performed the real-time forecasting of the river water level or stage at the Gurye station of the Sumjin river with lead times of 1, 3, and 6 h by applying a long short-term memory (LSTM)-based deep learning model. In addition, the applicability of the LSTM model was evaluated by comparing the results with those from widely used models based on support vector machine and multilayer perceptron. Consequently, we noted that the LSTM model exhibited a relatively better forecasting performance. Therefore, the applicability of the LSTM model should be extensively studied for flood forecasting applications.

Download Full-text

Is the Deep-Learning Technique a Completely Alternative for the Hydrological Model?: A Case Study on Hyeongsan River Basin, Korea

10.21203/rs.3.rs-669773/v1 ◽

2021 ◽

Author(s):

Jaewon Kwak ◽

Heechan Han ◽

Soojun Kim ◽

Hung Soo Kim

Keyword(s):

Deep Learning ◽

River Basin ◽

Hydrological Model ◽

Short Term Memory ◽

Hydrologic Model ◽

Hydrologic Models ◽

Runoff Simulation ◽

Runoff Modeling ◽

Learning Technique ◽

Relationship Of

Abstract It is no doubt that the reliable runoff simulation for proper water resources management is essential. In the past, the runoff was generally modeled from hydrologic models that analyze the rainfall-runoff relationship of the basin. However, since techniques have developed rapidly, it has been attempted to apply especially deep-learning technique for hydrological studies as an alternative to the hydrologic model. The objective of the study is to examine whether the deep-learning technique can completely replace the hydrologic model and show how to improve the performance of runoff simulation using deep-learning technique. The runoff in the Hyeongsan River basin, South Korea from 2013 to 2020 were simulated using two models, 1) Long Short-Term Memory model that is a deep learning technique widely used in the hydrological study and 2) TANK model, and then we compared the runoff modeling results from both models. The results suggested that it is hard to completely replace the hydrological model with the deep-learning technique due to its simulating behavior and discussed how to improve the reliability of runoff simulation results. Also, a method to improve the efficiency of runoff simulation through a hybrid model which is a combination of two approaches, deep-learning technique and hydrologic model was presented.

Download Full-text

WaterBench: A Large-scale Benchmark Dataset for Data-Driven Streamflow Forecasting

10.31223/x57w6t ◽

2021 ◽

Author(s):

Ibrahim Demir ◽

Zhongrun Xiang ◽

Bekir Zahit Demiray ◽

Muhammed Sit

Keyword(s):

Machine Learning ◽

Deep Learning ◽

Large Scale ◽

Short Term Memory ◽

Earth Science ◽

Science Research ◽

Benchmark Dataset ◽

Data Driven ◽

Streamflow Forecasting ◽

Learning Research

This study proposes a comprehensive benchmark dataset for streamflow forecasting, WaterBench, that follows FAIR data principles that is prepared with a focus on convenience for utilizing in data-driven and machine learning studies, and provides benchmark performance for state-of-art deep learning architectures on the dataset for comparative analysis. By aggregating the datasets of streamflow, precipitation, watershed area, slope, soil types, and evapotranspiration from federal agencies and state organizations (i.e., NASA, NOAA, USGS, and Iowa Flood Center), we provided the WaterBench for hourly streamflow forecast studies. This dataset has a high temporal and spatial resolution with rich metadata and relational information, which can be used for varieties of deep learning and machine learning research. We defined a sample streamflow forecasting task for the next 120 hours and provided performance benchmarks on this task with sample linear regression and deep learning models, including Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and S2S (Sequence-to-sequence). To some extent, WaterBench makes up for the lack of a unified benchmark in earth science research. We highly encourage researchers to use the WaterBench for deep learning research in hydrology.

Download Full-text

Deep learning rainfall-runoff predictions of extreme events

10.5194/hess-2021-423 ◽

2021 ◽

Author(s):

Jonathan Frame ◽

Frederik Kratzert ◽

Daniel Klotz ◽

Martin Gauch ◽

Guy Shelev ◽

...

Keyword(s):

Deep Learning ◽

Extreme Events ◽

Short Term Memory ◽

Data Driven ◽

Water Model ◽

Rainfall Runoff ◽

Short Term ◽

Term Memory ◽

High Return ◽

Long Short Term Memory

Abstract. The most accurate rainfall-runoff predictions are currently based on deep learning. There is a concern among hydrologists that data-driven models based on deep learning may not be reliable in extrapolation or for predicting extreme events. This study tests that hypothesis using Long Short-Term Memory networks (LSTMs) and an LSTM variant that is architecturally constrained to conserve mass. The LSTM (and the mass-conserving LSTM variant) remained relatively accurate in predicting extreme (high return-period) events compared to both a conceptual model (the Sacramento Model) and a process-based model (US National Water Model), even when extreme events were not included in the training period. Adding mass balance constraints to the data-driven model (LSTM) reduced model skill during extreme events.

Download Full-text

Three Cases of Joubert Syndrome in a Consanguineous Syrian Family and a Interesting Case of Multinational Collaboration

Journal of Pediatric Genetics ◽

10.1055/s-0040-1721826 ◽

2021 ◽

Author(s):

Davor Petrović ◽

Vida Čulić ◽

Zofia Swinderek-Alsayed

Keyword(s):

Low Income ◽

Autosomal Recessive ◽

Autosomal Recessive Disorder ◽

Joubert Syndrome ◽

Interesting Case ◽

Low Income Countries ◽

Ocular Motor ◽

Quality Health Care ◽

Quality Health ◽

Motor Apraxia

AbstractJoubert syndrome (JS) is a rare congenital, autosomal recessive disorder characterized by a distinctive brain malformation, developmental delay, ocular motor apraxia, breathing abnormalities, and high clinical and genetic heterogeneity. We are reporting three siblings with JS from consanguineous parents in Syria. Two of them had the same homozygous c.2172delA (p.Trp725Glyfs*) AHI1 mutation and the third was diagnosed prenatally with magnetic resonance imaging. This pathogenic variant is very rare and described in only a few cases in the literature. Multinational collaboration could be of benefit for the patients from undeveloped, low-income countries that have a low-quality health care system, especially for the diagnosis of rare diseases.

Download Full-text

Reviewing the Development of Rural Finance in Vietnam

Journal of Economics and Development ◽

10.33301/2013.15.01.06 ◽

2013 ◽

pp. 121-136

Author(s):

Duong Pham Bao

Keyword(s):

Low Income ◽

Financial System ◽

Low Income Countries ◽

Supply Side ◽

New Paradigm ◽

Rural Finance ◽

Market Mechanisms ◽

Rural Credit ◽

Rural Organizations ◽

Doi Moi

The objective of this article is to review the development of the rural financial system in Vietnam in recent years, especially, after Doi moi. There are two opposite schools of thought in the literature on rural credit policies in developing countries. One is the conventional supply-side (government-led) approach while the other is called “a new paradigm” that emphasizes the importance of the viability of financial providers and the well functioning of rural credit markets. Conventional theories of rural finance contend that rural finance in low-income countries is generally accompanied by many failures. Contrary to these theories, rural finance in Vietnam does not encounter the above-mentioned failures so far. Up to the present time, it is progressing well. Using a supply-side approach, methodologically, this study reviews the development of the rural financial system in Vietnam. The significance of this study is to challenge the extreme view of dichotomizing between the old and the new credit paradigms. Analysis in this study contends that a rural financial market that, (1) is initiated and spurred by government; (2) operates principally under market mechanisms; and (3) is strongly supported by rural organizations (semi-formal/informal institutions) can progress stably and well. Therefore, the extremely dichotomizing approach must be avoided.

Download Full-text

Abdominal Textiloma Mimicking as Left Colic Tumour: A Postoperative Complication Still Common in Low-Income Countries

EMJ Radiology ◽

10.33590/emjradiol/19-00208 ◽

2020 ◽

Keyword(s):

Low Income ◽

Early Stage ◽

Low Income Countries ◽

Fistula Formation ◽

Emergency Laparotomy ◽

Intermittent Fever ◽

Gastrointestinal Complications ◽

Surgical Safety ◽

Surgical Teams ◽

Safety Rules

Retained foreign bodies have become very rare in countries where the safety rules in the operating theatre are very rigorous and follow precise guidelines. There are low-income countries where hospital structures are precarious, in which the implementation of surgical safety rules has only been effective recently. Surgical teams in these countries are not yet well trained in the observance of the guidelines concerning swab count, meaning that textilomas are not uncommon. Abdominal textiloma may be asymptomatic, or present serious gastrointestinal complications such as bowel obstruction, perforation, or fistula formation because of misdiagnosis. It may mimic abscess formation in the early stage or soft tissue masses in the chronic stage. This case report presents a 27-year-old female who underwent an emergency laparotomy in a rural surgical centre for an ectopic pregnancy. Two months later, a swelling had appeared on the left side of her abdomen, gradually increasing in size, which was not very painful but caused digestive discomfort and asthenia. Intermittent fever was described and treated with antibiotics. The patient was referred to a better equipped centre to benefit from a CT scan. A textiloma was strongly suspected on the CT but a left colic mass was not excluded. Laparotomy confirmed the diagnosis of textiloma and the postoperative course was uneventful. Prevention rules must be strengthened in these countries where patients can hardly bear the costs of iterative surgeries for complications that are avoidable.

Download Full-text