TecVolSA: InSAR and Machine Learning for Surface Displacement Monitoring in South America

2021 ◽  
Author(s):  
Sina Montazeri ◽  
Homa Ansari ◽  
Francesco De Zan ◽  
René Mania ◽  
Robert Shau ◽  
...  
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Signal Processing ◽  
South America ◽  
Surface Deformation ◽  
Supervised Machine Learning ◽  
Research Centre ◽  
Prototype System ◽  
Data Archives ◽  
Deformation Velocity

<p>TecVolSA (Tectonics and Volcanoes in South America) is a project dedicated to the development of an intelligent Earth Observation (EO) data exploitation system for monitoring various geophysical activities in South America. Three partners from the German Aerospace Center (DLR) and the German Research Centre for Geosciences (GFZ) are involved to combine their expertise in signal processing, geophysics and Artificial Intelligence (AI).</p><p>The first milestone of the project is to perform interferometric processing on tens of terabytes of SAR data to generate deformation products. Efficient algorithms have been designed to accommodate big data processing. Employing these algorithms, five-year data archives of Sentinel-1 have been processed thus far. The data archives span an area of over 770,000 km² surrounding the central volcanic zone of the Andes. Products in the form of surface deformation velocity and displacement time series are generated as point-wise measurements. To ensure highly accurate deformation estimates, two novel techniques have been utilized: large-scale atmospheric correction and covariance-based phase estimation for distributed scatterers.</p><p>The second milestone is automatic mining of the wealth of the deformation products to gain insights about anthropogenic and geophysical signals in the region. Here two challenges are faced: the variety of crustal deformation processes as well as the sheer volume of the data. A closer analysis of the estimated deformation velocity verifies the presence of various signals including tectonic movements, volcanic unrest and slope-induced deformations. Such variety requires the classification of the observed signals. Furthermore, the dataset includes displacement time series and velocity estimates of over 750 million data points. This data volume necessitates the incorporation of AI for efficient mining of the products. The aforementioned challenges are met by combining geophysical and signal processing expertise of the project partners, and translating them to the AI algorithms.</p><p>The use of AI in EO is a growing topic with numerous successful applications. However, compared to the well-established AI applications of cartography and ground cover classification, there is not enough training data available for the analysis of tectonic and volcanic signals. Therefore, there is a need for synthetic data generation. GFZ produces geophysical models for the simulation of a diverse database that is used for the training of neural networks to autonomously discover significant events in deformation products.</p><p>DLR employs supervised machine learning techniques based on simulated data to automatically detect volcanic deformation from InSAR products. Apart from this application, signals which are not attributed to volcanic deformation are automatically clustered for further studies by expert geologists. For this approach, we depend on InSAR and geometrical feature engineering as well as advanced unsupervised learning algorithms. In the presentation, examples of clustering similar points in terms of temporal progression and a prototype system for the automatic detection of volcanic deformations will be illustrated.</p><p>Our system is being developed with scalability and transferability in mind. South America serves as a generic and challenging case for this development, as it reveals manifold geophysical and anthropogenic signals. Our ultimate goal is to apply the developed AI-assisted system for global processing.</p><p> </p>

scholarly journals Identifying Spatiotemporal Patterns in Land Use and Cover Samples from Satellite Image Time Series

2021 ◽  
Vol 13 (5) ◽  
pp. 974
Author(s):  
Lorena Alves Santos ◽  
Karine Ferreira ◽  
Michelle Picoli ◽  
Gilberto Camara ◽  
Raul Zurita-Milla ◽  
...  
Keyword(s):  
Machine Learning ◽  
Land Use ◽  
Time Series ◽  
Satellite Image ◽  
Spatiotemporal Patterns ◽  
Supervised Machine Learning ◽  
Learning Methods ◽  
Self Organizing Maps ◽  
Machine Learning Methods ◽  
Land Use And Cover

The use of satellite image time series analysis and machine learning methods brings new opportunities and challenges for land use and cover changes (LUCC) mapping over large areas. One of these challenges is the need for samples that properly represent the high variability of land used and cover classes over large areas to train supervised machine learning methods and to produce accurate LUCC maps. This paper addresses this challenge and presents a method to identify spatiotemporal patterns in land use and cover samples to infer subclasses through the phenological and spectral information provided by satellite image time series. The proposed method uses self-organizing maps (SOMs) to reduce the data dimensionality creating primary clusters. From these primary clusters, it uses hierarchical clustering to create subclusters that recognize intra-class variability intrinsic to different regions and periods, mainly in large areas and multiple years. To show how the method works, we use MODIS image time series associated to samples of cropland and pasture classes over the Cerrado biome in Brazil. The results prove that the proposed method is suitable for identifying spatiotemporal patterns in land use and cover samples that can be used to infer subclasses, mainly for crop-types.

scholarly journals Study of Systematic Bias in Measuring Surface Deformation with SAR Interferometry

2020 ◽  
Author(s):  
Homa Ansari ◽  
Francesco De Zan ◽  
Alessandro Parizzi
Keyword(s):  
Time Series ◽  
Surface Deformation ◽  
Sar Interferometry ◽  
Systematic Bias ◽  
Earth Surface ◽  
Phase Component ◽  
Measuring Surface ◽  
Deformation Velocity ◽  
Temporal Decorrelation ◽  
Efficient Processing

<div>This paper investigates the presence of a new interferometric signal in multilooked Synthetic Aperture Radar (SAR) interferograms which cannot be attributed to atmospheric or earth surface topography changes. The observed signal is short-lived and decays with temporal baseline; however, it is distinct from the stochastic noise usually attributed to temporal decorrelation. The presence of such fading signal introduces a systematic phase component, particularly in short temporal baseline interferograms. If unattended, it biases the estimation of Earth surface deformation from SAR time series. <br></div><div>The contribution of the mentioned phase component is quantitatively assessed. For short temporal baseline interferograms, we quantify the phase contribution to be in the regime of 5 rad at C-band. The biasing impact on deformation signal retrieval is further evaluated. As an example, exploiting a subset of short temporal baseline interferograms which connects each acquisition with the successive 5 in the time series, a significant bias of -6.5 mm/yr is observed in the estimation of deformation velocity from a four-year Sentinel-1 data stack. A practical solution for mitigation of this physical fading signal is further discussed; special attention is paid to the efficient processing of Big Data from modern SAR missions such as Sentinel-1 and NISAR. Adopting the proposed solution, the deformation bias is shown to decrease to -0.24 mm/yr for the Sentinel-1 time series.</div>Based on these analyses, we put forward our recommendations for efficient and accurate deformation signal retrieval from large stacks of multilooked interferograms.

TIME-SERIES SAR INTERFEROMETRY ANALYSIS OF SURFACE DEFORMATION AT MT. BROMO INDONESIA

2019 ◽  
Vol 3 ◽  
pp. 771
Author(s):  
Arliandy Pratama Arbad ◽  
Wataru Takeuchi ◽  
Yosuke Yosuke ◽  
Mutiara Jamilah ◽  
Achmad Ardy
Keyword(s):  
Time Series ◽  
Surface Deformation ◽  
Sar Interferometry ◽  
Time Interval ◽  
Infrastructure Management ◽  
Alos Palsar ◽  
Volcanic System ◽  
Deformation Velocity ◽  
Small Baseline Subset ◽  
Small Baseline

One of the most active volcanoes in Indonesia is Mt. Bromo, volcanic activities at Mt. Bromo has been recorded in 1775. We observe the surface deformation of the Mt. Bromo which located at eastern Java Indonesia area that includes neighborhood volcanic system on TNBTS (Taman Nasional Bukit Tengger Semeru). Recently, remote sensing has played as an important role to observe volcano behavior. We apply the SAR Interferometry (InSAR) algorithm referred to as Small Baseline Subset (SBAS) approach that allows us to generate mean deformation velocity maps and displacement time series for the studied area. The common SBAS technique, the set of interferometric phase observations writes as a linear combination of individual SAR scene phase values for each pixel independently. Particularly, the proposed analysis is based on 22 SAR data acquired by the ALOS/PALSAR sensors during the 2007–2017 time interval. A fewer studies have been able to show capability of InSAR analysis for investigating cycle of volcano especially of Mt. Bromo which characterized eruption stratovolcano in ranging one to five years. The results expected in this work represent an advancement of previous InSAR studies of the area that are mostly focused on the deformation affecting the caldera. According to the result, we expected this study could implement on risk management or infrastructure management.

scholarly journals Study of Systematic Bias in Measuring Surface Deformation with SAR Interferometry

2020 ◽  
Author(s):  
Homa Ansari ◽  
Francesco De Zan ◽  
Alessandro Parizzi
Keyword(s):  
Time Series ◽  
Surface Deformation ◽  
Sar Interferometry ◽  
Systematic Bias ◽  
Earth Surface ◽  
Phase Component ◽  
Measuring Surface ◽  
Deformation Velocity ◽  
Temporal Decorrelation ◽  
Efficient Processing

<div>This paper investigates the presence of a new interferometric signal in multilooked Synthetic Aperture Radar (SAR) interferograms which cannot be attributed to atmospheric or earth surface topography changes. The observed signal is short-lived and decays with temporal baseline; however, it is distinct from the stochastic noise usually attributed to temporal decorrelation. The presence of such fading signal introduces a systematic phase component, particularly in short temporal baseline interferograms. If unattended, it biases the estimation of Earth surface deformation from SAR time series. <br></div><div>The contribution of the mentioned phase component is quantitatively assessed. For short temporal baseline interferograms, we quantify the phase contribution to be in the regime of 5 rad at C-band. The biasing impact on deformation signal retrieval is further evaluated. As an example, exploiting a subset of short temporal baseline interferograms which connects each acquisition with the successive 5 in the time series, a significant bias of -6.5 mm/yr is observed in the estimation of deformation velocity from a four-year Sentinel-1 data stack. A practical solution for mitigation of this physical fading signal is further discussed; special attention is paid to the efficient processing of Big Data from modern SAR missions such as Sentinel-1 and NISAR. Adopting the proposed solution, the deformation bias is shown to decrease to -0.24 mm/yr for the Sentinel-1 time series.</div>Based on these analyses, we put forward our recommendations for efficient and accurate deformation signal retrieval from large stacks of multilooked interferograms.

Damage Classification of Composites Based On Analysis of Lamb Wave Signals Using Machine Learning

2020 ◽  
Author(s):  
Shweta Dabetwar ◽  
Stephen Ekwaro-Osire ◽  
Joao Paulo Dias
Keyword(s):  
Machine Learning ◽  
Signal Processing ◽  
Composite Materials ◽  
Fatigue Damage ◽  
Research Question ◽  
Machine Learning Algorithms ◽  
Supervised Machine Learning ◽  
Data Repository ◽  
Damage Classification

Abstract Composite materials have a myriad of applications in complex engineering systems, and multiple structural health monitoring strategies have been developed. However, these methods are challenging due to signal attenuation and excessive noise interference in composite materials. Signal processing can capture a small difference between the input-output signals associated with the severity of the damage in composites. Thus, the research question is "can signal processing techniques reduce the required number of features and assess the randomness of fatigue damage classification in composite materials using machine learning algorithms?" To answer this question, piezoelectric signals for carbon fiber reinforced polymer test specimens were taken from NASA Ames prognostics data repository. A framework based on a comparative analysis of signals was developed. For the first specific aim, the effectiveness of features based on statistical condition indicators of the sensor signals were evaluated. For the second specific aim, actuator-sensor signal pair were analyzed using cross-correlation to extract two features. These features were used to train and test four supervised machine learning (ML) algorithms for damage classification and their performance was discussed. For the third specific aim, randomness in the dataset of fatigue damage of the specimens was assessed. Results showed that by signal processing, the requirement of features for training ML was reduced with the improvement in the performance of ML. The randomness was captured by the utilization of two specimens from the same material. This work contributes to the improvement of intelligent damage classification of composite materials, operating under complex working conditions.

Comparative Analysis between Supervised Machine Learning and Time Series Models for stock price prediction

2021 ◽  
Author(s):  
D.N. Hendahewa ◽  
A.A.D.S.N.A. Perera ◽  
T.D.N. De Meraal ◽  
I.K.P.H. Indikadulle ◽  
S.D. Perera ◽  
...  
Keyword(s):  
Machine Learning ◽  
Time Series ◽  
Comparative Analysis ◽  
Stock Price ◽  
Supervised Machine Learning ◽  
Time Series Models ◽  
Stock Price Prediction ◽  
Price Prediction

Echo-State Networks for Predicting ENSO Beyond One Year

2021 ◽  
Author(s):  
forough hassanibesheli ◽  
Niklas Boers ◽  
Jurgen Kurths
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Time Series ◽  
Recurrent Neural Networks ◽  
Southern Oscillation ◽  
Numerical Models ◽  
Weather Prediction ◽  
Supervised Machine Learning ◽  
Climate Indices ◽  
Training Procedure

&lt;p&gt;Most forecasting schemes in the geosciences, and in particular for predicting weather and&lt;br&gt;climate indices such as the El Ni&amp;#241;o Southern Oscillation (ENSO), rely on process-based&lt;br&gt;numerical models [1]. Although statistical modelling[2] and prediction approaches also have&lt;br&gt;a long history, more recently, different machine learning techniques have been used to predict&lt;br&gt;climatic time series. One of the supervised machine learning algorithm which is suited for&lt;br&gt;temporal and sequential data processing and prediction is given by recurrent neural networks&lt;br&gt;(RNNs)[3]. In this study we develop a RNN-based method that (1) can learn the dynamics&lt;br&gt;of a stochastic time series without requiring access to a huge amount of data for training, and&lt;br&gt;(2) has comparatively simple structure and efficient training procedure. Since this algorithm&lt;br&gt;is suitable for investigating complex nonlinear time series such as climate time series, we&lt;br&gt;apply it to different ENSO indices. We demonstrate that our model can capture key features&lt;br&gt;of the complex system dynamics underlying ENSO variability, and that it can accurately&lt;br&gt;forecast ENSO for longer lead times in comparison to other recent studies[4].&lt;/p&gt;&lt;p&gt;&amp;#160;&lt;/p&gt;&lt;p&gt;Reference:&lt;/p&gt;&lt;p&gt;[1] P. Bauer, A. Thorpe, and G. Brunet, &amp;#8220;The quiet revolution of numerical weather prediction,&amp;#8221;&lt;br&gt;Nature, vol. 525, no. 7567, pp. 47&amp;#8211;55, 2015.&lt;/p&gt;&lt;p&gt;[2] D. Kondrashov, S. Kravtsov, A. W. Robertson, and M. Ghil, &amp;#8220;A hierarchy of data-based enso&lt;br&gt;models,&amp;#8221; Journal of climate, vol. 18, no. 21, pp. 4425&amp;#8211;4444, 2005.&lt;/p&gt;&lt;p&gt;[3] L. R. Medsker and L. Jain, &amp;#8220;Recurrent neural networks,&amp;#8221; Design and Applications, vol. 5, 2001.&lt;/p&gt;&lt;p&gt;[4] Y.-G. Ham, J.-H. Kim, and J.-J. Luo, &amp;#8220;Deep learning for multi-year enso forecasts,&amp;#8221; Nature,&lt;br&gt;vol. 573, no. 7775, pp. 568&amp;#8211;572, 2019.&lt;/p&gt;

Sign in / Sign up

Export Citation Format

Share Document