Earth Observation Data Analytics Using Machine and Deep Learning: Modern tools, applications and challenges

Description

Earth Observation Data Analytics Using Machine and Deep Learning: Modern tools, applications and challenges covers the basic properties, features and models for Earth observation (EO) recorded by very high-resolution (VHR) multispectral, hyperspectral, synthetic aperture radar (SAR), and multi-temporal observations. Approaches for applying pre-processing methods and deep learning techniques to satellite images for various applications - such as identifying land cover features, object detection, crop classification, target recognition, and the monitoring of earth resources - are described. Cost-efficient resource allocation solutions are provided, which are robust against common uncertainties that occur in annotating and extracting features on satellite images. This book is a valuable resource for engineers and researchers in academia and industry working on AI, machine and deep learning, data science, remote sensing, GIS, SAR, satellite communications, space science, image processing and computer vision. It will also be of interest to staff at research agencies, lecturers and advanced students in related fields. Readers will need a basic understanding of computing, remote sensing, GIS and image interpretation. Cover Contents About the editors Foreword 1 Introduction 1.1 Earth observation data 1.1.1 Organization 1.2 Categories of EO data 1.2.1 Passive imaging system 1.2.2 Active imaging system 1.3 Need of data analytics in EO data 1.4 Data analytics methodology 1.4.1 Machine learning 1.4.2 Deep learning 1.5 Data visualization techniques 1.5.1 Cartogram map 1.5.2 Heat map 1.5.3 Choropleth map 1.6 Types of inferences from data analytics (application areas) 1.6.1 Agriculture 1.6.2 Forestry 1.6.3 Land cover classification 1.6.4 Flooding 1.6.5 Maritime 1.6.6 Defence and security 1.6.7 Wetland 1.7 Conclusion References Part I. Clustering and classification of Earth Observation data 2 Deep learning method for crop classification using remote sensing data 2.1 Sources of remote sensing data collection 2.2 Tools for processing remote sensing data 2.3 Crop classification using remote sensing data 2.3.1 Methods for crop classification 2.3.2 Case study 2.4 Performance evaluation 2.5 Conclusion References 3 Using optical images to demarcate fields in L band SAR images for effective deep learning based crop classification and crop cov 3.1 Introduction 3.1.1 Motivation 3.1.2 Research contribution 3.1.3 Organization 3.2 Related work 3.3 Proposed methodology 3.3.1 SAR image pre-processing and decomposition 3.3.2 Edge detection & field extraction 3.3.3 Classification using deep learning 3.4 Study area 3.5 Experimental setting 3.5.1 Dataset 1 3.5.2 Dataset 2 3.6 Experimental result and analysis 3.7 Conclusion References 4 Leveraging twin networks for land use land cover classification 4.1 Introduction 4.2 Related literature 4.3 Methodology 4.3.1 Dataset 4.3.2 Siamese network 4.3.3 Encoders 4.4 Results and discussion 4.5 Conclusion and future work References 5 Exploiting artificial immune networks for enhancing RS image classification 5.1 Introduction 5.1.1 The immune system 5.1.2 Classification based on the AIS 5.2 Data used and study area 5.3 Experimental approach 5.3.1 Initialization 5.3.2 Randomly choose an antigen 5.3.3 Select the highest affinity 5.3.4 Clone the selected Ab’s 5.3.5 Allow each Ab’s in clone set 5.3.6 Calculate the affinity aff * j 5.3.7 Select the highest affinity 5.3.8 Decide 5.3.9 Replace 5.3.10 A stopping criterion 5.4 Result 5.5 Conclusion References 6 Detection and segmentation of aircrafts in UAV images with a deep learning-based approach 6.1 Introduction 6.2 Background 6.2.1 Digital images and spatial resolution 6.2.2 Neural networks 6.2.3 CNNs 6.3 Methodology 6.3.1 Dataset 6.3.2 Object detection 6.3.3 Semantic segmentation 6.4 Model training and results 6.4.1 Object detection 6.4.2 Semantic segmentation 6.5 Conclusions and discussion References Part II. Rare event detection using Earth Observation data 7 A transfer learning approach for hurricane damage assessment using satellite imagery 7.1 Introduction 7.2 Literature review 7.3 Image processing techniques 7.3.1 Statistical-based algorithms 7.3.2 Learning-based algorithms 7.4 Transfer learning 7.4.1 AlexNet 7.5 Implementation 7.6 Conclusion References 8 Wildfires, volcanoes and climate change monitoring from satellite images using deep neural networks 8.1 Introduction 8.2 Background and related work 8.3 Modern DL methods 8.3.1 U-Net 8.3.2 AlexNet 8.3.3 Inception-v3 8.3.4 Other neural networks 8.4 Benefits of using this approach 8.5 Long-term climate change monitoring using DL methods 8.6 Other applications of this approach 8.7 Possible problems 8.8 Conclusion References 9 A comparative study on torrential slide shortcoming zones and causative factors using machine learning techniques: a case study 9.1 Introduction 9.2 Discussions on landslide influencing factors 9.3 Materials and methods 9.4 Dataset collections 9.5 Rainfall characteristics in Kerala 9.6 Landslide impacted earthquake 9.7 Anthropogenic activities 9.8 Machine learning techniques for landslide study using satellite images 9.8.1 Highlights of machine learning techniques in satellite images 9.9 Emergency rescue and mitigation 9.10 Conclusion References 10 Machine learning paradigm for predicting reservoir property: an exploratory analysis 10.1 Introduction 10.2 Geo-scientific data sources for reservoir characterization 10.2.1 Seismic survey 10.2.2 Well logging 10.3 Research issues and objectives 10.4 Description of the case study 10.4.1 Geological background of the survey area 10.5 ML for reservoir characterization: the proposed approach 10.5.1 Well tie 10.5.2 Seismic signal reconstruction 10.5.3 Smoothing of well log 10.5.4 Seismic attributes selection 10.5.5 Outlier removal 10.6 Experimental results and analysis 10.6.1 Statistical data analysis 10.6.2 Results and analysis of ML modeling 10.6.3 Performance comparison of shallow vs. DNN model 10.7 Discussion and future prospects 10.8 Conclusion Acknowledgment References Part III. Tools and technologies for Earth Observation data 11 The application of R software in water science 11.1 Introduction 11.1.1 What is hydrology? 11.1.2 What is computational hydrology? 11.1.3 What is hydroinformatics? 11.1.4 Free, open-source software (FOSS) 11.1.5 What is GitHub? 11.2 Material and methods 11.2.1 What is R? What is an integrated development environment (IDE)? 11.2.2 What are R packages? 11.2.3 What are cheatsheets? 11.2.4 What are R communities? 11.2.5 What is RPubs? 11.2.6 What are popular conferences in R? 11.2.7 What is joss (open source software)? 11.2.8 What is R studio cloud? 11.2.9 What is R application in hydrology? 11.2.10 What are hydrological packages? 11.2.11 Workflow of R in hydrology 11.2.12 Data for hydrology? How to retrieve datasets? 11.2.13 Preprocessing retrieved hydrological data (data tidying) 11.2.14 Different hydrology model types? 11.2.15 Hydrologic time series analysis tools in R? 11.2.16 Hydrological ML application tools in R? 11.2.17 Remote sensing tools in R 11.3 Conclusion and future prospects References 12 Geospatial big data analysis using neural networks 12.1 Introduction 12.1.1 Geospatial data 12.1.2 Big data analysis 12.1.3 Fog computing 12.1.4 Neural network 12.1.5 Contribution 12.2 Related works 12.2.1 Big data analysis on geospatial data 12.2.2 Data processing techniques in fog environment 12.3 Proposed work 12.4 Methodology and concepts 12.4.1 Data pre-processing on fog environment 12.4.2 Prediction on cloud environment using ANN 12.5 Results and discussion 12.6 Conclusion References 13 Software framework for spatiotemporal data analysis and mining of earth observation data 13.1 Introduction 13.1.1 Visualization 13.1.2 Multidimensional analysis 13.1.3 Data mining 13.2 Related work 13.3 Challenges 13.4 The ST-DAME 13.4.1 Conceptual architecture of the framework 13.4.2 Proposed framework 13.4.3 ST-DAME in action 13.5 Result 13.5.1 Automated system 13.5.2 Customized system 13.6 Conclusion References 14 Conclusion 14.1 Excerpts from various chapters 14.2 Issues and challenges 14.2.1 Collecting meaningful and real-time data 14.2.2 Data storage 14.2.3 Resolution; quality promotion 14.2.4 Budget limitations 14.2.5 Standardization 14.2.6 Lack of ground truth data 14.2.7 Processing and analysis References Index Back Cover