scholarly journals A Novel Stacked Long Short-Term Memory Approach of Deep Learning for Streamflow Simulation

2021 ◽  
Vol 13 (23) ◽  
pp. 13384
Author(s):  
Majid Mirzaei ◽  
Haoxuan Yu ◽  
Adnan Dehghani ◽  
Hadi Galavi ◽  
Vahid Shokri ◽  
...  
Keyword(s):  
Spatial Distribution ◽  
Short Term Memory ◽  
Data Driven ◽  
Rainfall Runoff ◽  
Short Term ◽  
Runoff Simulation ◽  
Physically Based ◽  
Long Short Term Memory ◽  
Using Data ◽  
Modelling Approach

Rainfall-Runoff simulation is the backbone of all hydrological and climate change studies. This study proposes a novel stochastic model for daily rainfall-runoff simulation called Stacked Long Short-Term Memory (SLSTM) relying on machine learning technology. The SLSTM model utilizes only the rainfall-runoff data in its modelling approach and the hydrology system is deemed a blackbox. Conversely, the distributed and physically-based hydrological models, e.g., SWAT (Soil and Water Assessment Tool) preserve the physical aspect of hydrological variables and their inter-relations while taking a wide range of data. The two model types provide specific applications that interest modelers, who can apply them according to their project specification and objectives. However, sparse distribution of point-data may hinder physical models’ performance, which may not be the case in data-driven models. This study proposes a specific SLSTM model and investigates the SLSTM and SWAT models’ data dependency in terms of their spatial distribution. The study was conducted in the two distinct river basins of Samarahan and Trusan, Malaysia, with over 20 years of hydro-climate data. The Trusan basin’s rain gauges are scattered downstream of the basin outlet and Samarahan’s are located around the basin, with one station within each basin’s limits. The SWAT was developed and calibrated following its general modelling approach, however, the SLSTM performance was also tested using data preprocessing with principal component analysis (PCA). Results showed that the SWAT performance for daily streamflow simulation at Samarahan has been superior to that of Trusan. Both the SLSTM and PCA-SLSTM models, however, showed better performance at Trusan with PCA-SLSTM outperforming the SLSTM. This demonstrates that the SWAT model is greatly affected by the spatial distribution of its input data, while data-driven models, irrespective of the spatial distribution of their entry data, can perform well if the data adequacy condition is met. However, considering the structural difference between the two models, each has its specific application in a water resources context. The study of catchments’ response to changes in the hydrology cycle requires a physically-based model like SWAT with proper spatial and temporal distribution of its entry data. However, the study of a specific phenomenon without considering the underlying processes can be done using data-driven models like SLSTM, where improper spatial distribution of data cannot be a restricting factor.

scholarly journals Long short-term memory networks enhance rainfall-runoff modelling at the national scale of Denmark

2022 ◽  
Vol 49 ◽  
Author(s):  
Julian Koch ◽  
Raphael Schneider
Keyword(s):  
Short Term Memory ◽  
Simulated Data ◽  
National Scale ◽  
Short Term ◽  
Ungauged Catchments ◽  
Term Memory ◽  
Convincing Argument ◽  
Physically Based ◽  
Long Short Term Memory ◽  
Using Data

This study explores the application of long short-term memory (LSTM) networks to simulate runoff at the national scale of Denmark using data from 301 catchments. This is the first LSTM application on Danish data. The results were benchmarked against the Danish national water resources model (DK-model), a physically based hydrological model. The median Kling-Gupta Efficiency (KGE), a common metric to assess performance of runoff predictions (optimum of 1), increased from 0.7 (DK-model) to 0.8 (LSTM) when trained against all catchments. Overall, the LSTM outperformed the DK-model in 80% of catchments. Despite the compelling KGE evaluation, the water balance closure was modelled less accurately by the LSTM. The applicability of LSTM networks for modelling ungauged catchments was assessed via a spatial split-sample experiment. A 20% spatial hold-out showed poorer performance of the LSTM with respect to the DK model. However, after pre-training, that is, weight initialisation obtained from training against simulated data from the DK-model, the performance of the LSTM was effectively improved. This formed a convincing argument supporting the knowledge-guided machine learning (ML) paradigm to integrate physically based models and ML to train robust models that generalise well.

scholarly journals Rainfall-Runoff modelling using Long-Short-Term-Memory (LSTM) networks

2018 ◽  
Author(s):  
Frederik Kratzert ◽  
Daniel Klotz ◽  
Claire Brenner ◽  
Karsten Schulz ◽  
Mathew Herrnegger
Keyword(s):  
Short Term Memory ◽  
Regional Scale ◽  
Model Performance ◽  
Data Driven ◽  
Rainfall Runoff ◽  
Short Term ◽  
Data Set ◽  
Term Memory ◽  
Soil Moisture Accounting ◽  
Long Short Term Memory

Abstract. Rainfall-runoff modelling is one of the key challenges in the field of hydrology. Various approaches exist, ranging from physically based over conceptual to fully data driven models. In this paper, we propose a novel data driven approach, using the Long-Short-Term-Memory (LSTM) network, a special type of recurrent neural networks. The advantage of the LSTM is its ability to learn long-term dependencies between the provided input and output of the network, which are essential for modelling storage effects in e.g. catchments with snow influence. We use 241 catchments of the freely available CAMELS data set to test our approach and also compare the results to the well-known Sacramento Soil Moisture Accounting Model (SAC-SMA) coupled with the Snow-17 snow routine. We also show the potential of the LSTM as a regional hydrological model, in which one model predicts the discharge for a variety of catchments. In our last experiment, we show the possibility to transfer process understanding, learned at regional scale, to individual catchments and thereby increasing model performance when compared to a LSTM trained only on the data of single catchments. Using this approach, we were able to achieve better model performance as the SAC-SMA + Snow-17, which underlines the potential of the LSTM for hydrological modelling applications.

scholarly journals Rainfall–runoff modelling using Long Short-Term Memory (LSTM) networks

2018 ◽  
Vol 22 (11) ◽  
pp. 6005-6022 ◽  
Author(s):  
Frederik Kratzert ◽  
Daniel Klotz ◽  
Claire Brenner ◽  
Karsten Schulz ◽  
Mathew Herrnegger
Keyword(s):  
Short Term Memory ◽  
Regional Scale ◽  
Model Performance ◽  
Data Driven ◽  
Rainfall Runoff ◽  
Short Term ◽  
Data Set ◽  
Term Memory ◽  
Soil Moisture Accounting ◽  
Long Short Term Memory

Abstract. Rainfall–runoff modelling is one of the key challenges in the field of hydrology. Various approaches exist, ranging from physically based over conceptual to fully data-driven models. In this paper, we propose a novel data-driven approach, using the Long Short-Term Memory (LSTM) network, a special type of recurrent neural network. The advantage of the LSTM is its ability to learn long-term dependencies between the provided input and output of the network, which are essential for modelling storage effects in e.g. catchments with snow influence. We use 241 catchments of the freely available CAMELS data set to test our approach and also compare the results to the well-known Sacramento Soil Moisture Accounting Model (SAC-SMA) coupled with the Snow-17 snow routine. We also show the potential of the LSTM as a regional hydrological model in which one model predicts the discharge for a variety of catchments. In our last experiment, we show the possibility to transfer process understanding, learned at regional scale, to individual catchments and thereby increasing model performance when compared to a LSTM trained only on the data of single catchments. Using this approach, we were able to achieve better model performance as the SAC-SMA + Snow-17, which underlines the potential of the LSTM for hydrological modelling applications.

scholarly journals Deep learning rainfall-runoff predictions of extreme events

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.

scholarly journals Data-driven predictions of a multiscale Lorenz 96 chaotic system using machine-learning methods: reservoir computing, artificial neural network, and long short-term memory network

2020 ◽  
Vol 27 (3) ◽  
pp. 373-389 ◽  
Author(s):  
Ashesh Chattopadhyay ◽  
Pedram Hassanzadeh ◽  
Devika Subramanian
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Short Term Memory ◽  
Data Driven ◽  
Reservoir Computing ◽  
Short Term ◽  
Term Memory ◽  
Machine Learning Methods ◽  
Long Short Term Memory ◽  
Lorenz 96

Abstract. In this paper, the performance of three machine-learning methods for predicting short-term evolution and for reproducing the long-term statistics of a multiscale spatiotemporal Lorenz 96 system is examined. The methods are an echo state network (ESN, which is a type of reservoir computing; hereafter RC–ESN), a deep feed-forward artificial neural network (ANN), and a recurrent neural network (RNN) with long short-term memory (LSTM; hereafter RNN–LSTM). This Lorenz 96 system has three tiers of nonlinearly interacting variables representing slow/large-scale (X), intermediate (Y), and fast/small-scale (Z) processes. For training or testing, only X is available; Y and Z are never known or used. We show that RC–ESN substantially outperforms ANN and RNN–LSTM for short-term predictions, e.g., accurately forecasting the chaotic trajectories for hundreds of numerical solver's time steps equivalent to several Lyapunov timescales. The RNN–LSTM outperforms ANN, and both methods show some prediction skills too. Furthermore, even after losing the trajectory, data predicted by RC–ESN and RNN–LSTM have probability density functions (pdf's) that closely match the true pdf – even at the tails. The pdf of the data predicted using ANN, however, deviates from the true pdf. Implications, caveats, and applications to data-driven and data-assisted surrogate modeling of complex nonlinear dynamical systems, such as weather and climate, are discussed.

Sign in / Sign up

Export Citation Format

Share Document