scholarly journals Combining Ensemble Kalman Filter and Reservoir Computing to predict spatio-temporal chaotic systems from imperfect observations and models

2020 ◽  
Author(s):  
Futo Tomizawa ◽  
Yohei Sawada
Keyword(s):  
Machine Learning ◽  
Kalman Filter ◽  
Chaotic Systems ◽  
Weather Prediction ◽  
Machine Learning Techniques ◽  
Reservoir Computing ◽  
Learning Techniques ◽  
Spatio Temporal ◽  
Ensemble Transform Kalman Filter ◽  
Lorenz 96

Abstract. Prediction of spatio-temporal chaotic systems is important in various fields, such as Numerical Weather Prediction (NWP). While data assimilation methods have been applied in NWP, machine learning techniques, such as Reservoir Computing (RC), are recently recognized as promising tools to predict spatio-temporal chaotic systems. However, the sensitivity of the skill of the machine learning based prediction to the imperfectness of observations is unclear. In this study, we evaluate the skill of RC with noisy and sparsely distributed observations. We intensively compare the performances of RC and Local Ensemble Transform Kalman Filter (LETKF) by applying them to the prediction of the Lorenz 96 system. Although RC can successfully predict the Lorenz 96 system if the system is perfectly observed, we find that RC is vulnerable to observation sparsity compared with LETKF. To overcome this limitation of RC, we propose to combine LETKF and RC. In our proposed method, the system is predicted by RC that learned the analysis time series estimated by LETKF. Our proposed method can successfully predict the Lorenz 96 system using noisy and sparsely distributed observations. Most importantly, our method can predict better than LETKF when the process-based model is imperfect.

scholarly journals Combining ensemble Kalman filter and reservoir computing to predict spatiotemporal chaotic systems from imperfect observations and models

2021 ◽  
Vol 14 (9) ◽  
pp. 5623-5635 ◽  
Author(s):  
Futo Tomizawa ◽  
Yohei Sawada
Keyword(s):  
Machine Learning ◽  
Kalman Filter ◽  
Chaotic Systems ◽  
Weather Prediction ◽  
Machine Learning Techniques ◽  
Reservoir Computing ◽  
Learning Techniques ◽  
Ensemble Transform Kalman Filter ◽  
Lorenz 96 ◽  
Better Than

Abstract. Prediction of spatiotemporal chaotic systems is important in various fields, such as numerical weather prediction (NWP). While data assimilation methods have been applied in NWP, machine learning techniques, such as reservoir computing (RC), have recently been recognized as promising tools to predict spatiotemporal chaotic systems. However, the sensitivity of the skill of the machine-learning-based prediction to the imperfectness of observations is unclear. In this study, we evaluate the skill of RC with noisy and sparsely distributed observations. We intensively compare the performances of RC and local ensemble transform Kalman filter (LETKF) by applying them to the prediction of the Lorenz 96 system. In order to increase the scalability to larger systems, we applied a parallelized RC framework. Although RC can successfully predict the Lorenz 96 system if the system is perfectly observed, we find that RC is vulnerable to observation sparsity compared with LETKF. To overcome this limitation of RC, we propose to combine LETKF and RC. In our proposed method, the system is predicted by RC that learned the analysis time series estimated by LETKF. Our proposed method can successfully predict the Lorenz 96 system using noisy and sparsely distributed observations. Most importantly, our method can predict better than LETKF when the process-based model is imperfect.

scholarly journals Predicting land deformation by integrating InSAR data and cone penetration testing through machine learning techniques

2020 ◽  
Vol 382 ◽  
pp. 525-529
Author(s):  
Melika Sajadian ◽  
Ana Teixeira ◽  
Faraz S. Tehrani ◽  
Mathias Lemmens
Keyword(s):  
Machine Learning ◽  
Soil Mechanics ◽  
Machine Learning Techniques ◽  
Cone Penetration ◽  
Penetration Testing ◽  
Cone Penetration Testing ◽  
Learning Techniques ◽  
Land Deformation ◽  
Spatio Temporal

Abstract. Built environments developed on compressible soils are susceptible to land deformation. The spatio-temporal monitoring and analysis of these deformations are necessary for sustainable development of cities. Techniques such as Interferometric Synthetic Aperture Radar (InSAR) or predictions based on soil mechanics using in situ characterization, such as Cone Penetration Testing (CPT) can be used for assessing such land deformations. Despite the combined advantages of these two methods, the relationship between them has not yet been investigated. Therefore, the major objective of this study is to reconcile InSAR measurements and CPT measurements using machine learning techniques in an attempt to better predict land deformation.

Exploring multi-modalities in weather prediction using a univariate graph based on machine learning techniques

2021 ◽  
Author(s):  
Natacha Galmiche ◽  
Nello Blaser ◽  
Morten Brun ◽  
Helwig Hauser ◽  
Thomas Spengler ◽  
...  
Keyword(s):  
Machine Learning ◽  
Standard Deviation ◽  
Probability Distributions ◽  
Weather Prediction ◽  
A Priori ◽  
Clustering Algorithms ◽  
Quantitative Information ◽  
Machine Learning Techniques ◽  
Topological Data Analysis ◽  
Learning Techniques

<p>Probability distributions based on ensemble forecasts are commonly used to assess uncertainty in weather prediction. However, interpreting these distributions is not trivial, especially in the case of multimodality with distinct likely outcomes. The conventional summary employs mean and standard deviation across ensemble members, which works well for unimodal, Gaussian-like distributions. In the case of multimodality this misleads, discarding crucial information. </p><p>We aim at combining previously developed clustering algorithms in machine learning and topological data analysis to extract useful information such as the number of clusters in an ensemble. Given the chaotic behaviour of the atmosphere, machine learning techniques can provide relevant results even if no, or very little, a priori information about the data is available. In addition, topological methods that analyse the shape of the data can make results explainable.</p><p>Given an ensemble of univariate time series, a graph is generated whose edges and vertices represent clusters of members, including additional information for each cluster such as the members belonging to them, their uncertainty, and their relevance according to the graph. In the case of multimodality, this approach provides relevant and quantitative information beyond the commonly used mean and standard deviation approach that helps to further characterise the predictability.</p>

scholarly journals Can machine learning improve the model representation of TKE dissipation rate in the boundary layer for complex terrain?

2020 ◽  
Author(s):  
Nicola Bodini ◽  
Julie K. Lundquist ◽  
Mike Optis
Keyword(s):  
Machine Learning ◽  
Numerical Weather Prediction ◽  
Complex Terrain ◽  
Dissipation Rate ◽  
Prediction Models ◽  
Weather Prediction ◽  
Machine Learning Techniques ◽  
Learning Techniques ◽  
Tke Dissipation Rate ◽  
Numerical Weather Prediction Models

Abstract. Current turbulence parameterizations in numerical weather prediction models at the mesoscale assume a local equilibrium between production and dissipation of turbulence. As this assumption does not hold at fine horizontal resolutions, improved ways to represent turbulent kinetic energy (TKE) dissipation rate (ε) are needed. Here, we use a 6-week data set of turbulence measurements from 184 sonic anemometers in complex terrain at the Perdigão field campaign to suggest improved representations of dissipation rate. First, we demonstrate that a widely used Mellor, Yamada, Nakanishi, and Niino (MYNN) parameterization of TKE dissipation rate leads to a large inaccuracy and bias in the representation of ε. Next, we assess the potential of machine-learning techniques to predict TKE dissipation rate from a set of atmospheric and terrain-related features. We train and test several machine-learning algorithms using the data at Perdigão, and we find that multivariate polynomial regressions and random forests can eliminate the bias MYNN currently shows in representing ε, while also reducing the average error by up to 30 %. Of all the variables included in the algorithms, TKE is the variable responsible for most of the variability of ε, and a strong positive correlation exists between the two. These results suggest further consideration of machine-learning techniques to enhance parameterizations of turbulence in numerical weather prediction models.

scholarly journals Can machine learning improve the model representation of turbulent kinetic energy dissipation rate in the boundary layer for complex terrain?

2020 ◽  
Vol 13 (9) ◽  
pp. 4271-4285
Author(s):  
Nicola Bodini ◽  
Julie K. Lundquist ◽  
Mike Optis
Keyword(s):  
Machine Learning ◽  
Kinetic Energy ◽  
Complex Terrain ◽  
Dissipation Rate ◽  
Prediction Models ◽  
Weather Prediction ◽  
Machine Learning Techniques ◽  
Learning Techniques ◽  
Tke Dissipation Rate ◽  
Numerical Weather Prediction Models

Abstract. Current turbulence parameterizations in numerical weather prediction models at the mesoscale assume a local equilibrium between production and dissipation of turbulence. As this assumption does not hold at fine horizontal resolutions, improved ways to represent turbulent kinetic energy (TKE) dissipation rate (ϵ) are needed. Here, we use a 6-week data set of turbulence measurements from 184 sonic anemometers in complex terrain at the Perdigão field campaign to suggest improved representations of dissipation rate. First, we demonstrate that the widely used Mellor, Yamada, Nakanishi, and Niino (MYNN) parameterization of TKE dissipation rate leads to a large inaccuracy and bias in the representation of ϵ. Next, we assess the potential of machine-learning techniques to predict TKE dissipation rate from a set of atmospheric and terrain-related features. We train and test several machine-learning algorithms using the data at Perdigão, and we find that the models eliminate the bias MYNN currently shows in representing ϵ, while also reducing the average error by up to almost 40 %. Of all the variables included in the algorithms, TKE is the variable responsible for most of the variability of ϵ, and a strong positive correlation exists between the two. These results suggest further consideration of machine-learning techniques to enhance parameterizations of turbulence in numerical weather prediction models.

scholarly journals Using Real-Time Data and Unsupervised Machine Learning Techniques to Study Large-Scale Spatio–Temporal Characteristics of Wastewater Discharges and their Influence on Surface Water Quality in the Yangtze River Basin

Water ◽  
2019 ◽  
Vol 11 (6) ◽  
pp. 1268 ◽  
Author(s):  
Zhenzhen Di ◽  
Miao Chang ◽  
Peikun Guo ◽  
Yang Li ◽  
Yin Chang
Keyword(s):  
Machine Learning ◽  
Surface Water ◽  
Real Time ◽  
Yangtze River Basin ◽  
Clustering Algorithms ◽  
Machine Learning Techniques ◽  
Unsupervised Machine Learning ◽  
Learning Techniques ◽  
The Yangtze River Basin ◽  
Spatio Temporal

Most worldwide industrial wastewater, including in China, is still directly discharged to aquatic environments without adequate treatment. Because of a lack of data and few methods, the relationships between pollutants discharged in wastewater and those in surface water have not been fully revealed and unsupervised machine learning techniques, such as clustering algorithms, have been neglected in related research fields. In this study, real-time monitoring data for chemical oxygen demand (COD), ammonia nitrogen (NH3-N), pH, and dissolved oxygen in the wastewater discharged from 2213 factories and in the surface water at 18 monitoring sections (sites) in 7 administrative regions in the Yangtze River Basin from 2016 to 2017 were collected and analyzed by the partitioning around medoids (PAM) and expectation–maximization (EM) clustering algorithms, Welch t-test, Wilcoxon test, and Spearman correlation. The results showed that compared with the spatial cluster comprising unpolluted sites, the spatial cluster comprised heavily polluted sites where more wastewater was discharged had relatively high COD (>100 mg L−1) and NH3-N (>6 mg L−1) concentrations and relatively low pH (<6) from 15 industrial classes that respected the different discharge limits outlined in the pollutant discharge standards. The results also showed that the economic activities generating wastewater and the geographical distribution of the heavily polluted wastewater changed from 2016 to 2017, such that the concentration ranges of pollutants in discharges widened and the contributions from some emerging enterprises became more important. The correlations between the quality of the wastewater and the surface water strengthened as the whole-year data sets were reduced to the heavily polluted periods by the EM clustering and water quality evaluation. This study demonstrates how unsupervised machine learning algorithms play an objective and effective role in data mining real-time monitoring information and highlighting spatio–temporal relationships between pollutants in wastewater discharges and surface water to support scientific water resource management.

scholarly journals Multi-Step Ahead Wind Power Generation Prediction Based on Hybrid Machine Learning Techniques

Energies ◽  
2018 ◽  
Vol 11 (8) ◽  
pp. 1975 ◽  
Author(s):  
Wei Dong ◽  
Qiang Yang ◽  
Xinli Fang
Keyword(s):  
Machine Learning ◽  
Phase Space ◽  
Fuzzy Inference ◽  
Weather Prediction ◽  
Wind Farms ◽  
Machine Learning Techniques ◽  
Multiple Time ◽  
Inference System ◽  
Learning Techniques ◽  
Benchmark Solutions

Accurate generation prediction at multiple time-steps is of paramount importance for reliable and economical operation of wind farms. This study proposed a novel algorithmic solution using various forms of machine learning techniques in a hybrid manner, including phase space reconstruction (PSR), input variable selection (IVS), K-means clustering and adaptive neuro-fuzzy inference system (ANFIS). The PSR technique transforms the historical time series into a set of phase-space variables combining with the numerical weather prediction (NWP) data to prepare candidate inputs. A minimal redundancy maximal relevance (mRMR) criterion based filtering approach is used to automatically select the optimal input variables for the multi-step ahead prediction. Then, the input instances are divided into a set of subsets using the K-means clustering to train the ANFIS. The ANFIS parameters are further optimized to improve the prediction performance by the use of particle swarm optimization (PSO) algorithm. The proposed solution is extensively evaluated through case studies of two realistic wind farms and the numerical results clearly confirm its effectiveness and improved prediction accuracy compared to benchmark solutions.

Sign in / Sign up

Export Citation Format

Share Document