Advances in Scalable and Intelligent Geospatial Analytics: Challenges and Applications

Cover
Half Title
Title Page
Copyright Page
Table of Contents
Preface
Editors
Contributors
Section I Introduction to Geospatial Analytics
Chapter 1 Geospatial Technology – Developments, Present Scenario and Research Challenges
1.1 Introduction
1.1.1 Concept of Spatial Data
1.1.2 Spatial Data Sources
1.1.3 Geographic Coordinate System
1.1.4 Map Projections
1.1.5 Spatial Data Modelling
1.1.6 Spatial Database Creation
1.1.7 Spatial Relations
1.1.8 Spatial Data Analysis
1.1.9 Spatial Data Interpolation
1.1.10 Digital Terrain Modelling
1.1.11 Network Analysis
1.1.12 Statistical Analysis
1.1.13 Visualisation of Spatial Data Analysis
1.1.14 Spatial Decision Support Systems
1.1.15 Spatial Data Accuracy
1.2 Applications
1.3 Research Challenges
1.4 Open Areas of Research
1.5 Conclusion
References
Section II Geo-Ai
Chapter 2 Perspectives on Geospatial Artificial Intelligence Platforms for Multimodal Spatiotemporal Datasets
2.1 Introduction
2.2 Challenges and Opportunities of Different Geospatial Data Modalities
2.3 Motivation for a Data-Centric, Multimodal Geospatial Artificial Intelligence Platform
2.3.1 Current Challenges in ML-Based Geospatial Analysis
2.3.2 An Example of a Geospatial AI Platform: Trinity
2.3.3 Key Advantages and Observed Benefits of Trinity
2.4 Representation, Alignment, and Fusion of Multimodal Geospatial Datasets
2.4.1 Preliminary: Spherical Mercator Projection and Zoom-q Tiles
2.4.2 Spatial Transformations of Mobility Data
2.4.3 Spatial Transformations of Road Network Geometry
2.4.4 Vector Geometry Data
2.4.5 Temporal Transformations of Mobility Data
2.4.6 Synthetic Generation of Geospatial Data Representations
2.4.7 Self-Supervised Representation Learning from Geospatial Data
2.4.8 Geospatial Imagery
2.4.9 Auxiliary Datasets and Data Provenance
2.5 Design Overview of a Geospatial AI Platform
2.5.1 Machine Learning Operations: MLOps
2.5.2 Components of a Geospatial AI Platform
2.6 ML Feature Management and Feature Platform
2.6.1 Why Do We Need a Feature Platform?
2.6.2 Components of a ML Feature Platform
2.6.3 Design Considerations for a ML Feature Platform
2.7 Label Management and Label Platform
2.7.1 Components of a Label Platform
2.7.1.1 Label Generation and Editing
2.7.1.2 Label Visualization, Analysis, and Validation
2.7.1.3 Label Metadata and Catalog
2.7.1.4 Stratification
2.7.1.5 Active Learning
2.7.2 Design Considerations for a Label Platform
2.8 Machine Learning Infrastructure Components
2.8.1 Data Processing Framework
2.8.2 Storage, Compute, and Metadata Handling
2.9 Machine Learning Modeling Kernel
2.9.1 Serving and Deployment of Trained Models
2.10 Trinity Experiment Lifecycle
2.10.1 Project and Experiment Setup via the User Interface
2.10.2 Data Preparation and Training
2.10.3 Scalable Distributed Inference
2.10.4 Visualization and Evaluation of Predictions
2.10.5 Product Types and Sample Applications
2.11 Conclusions
Note
References
Chapter 3 Temporal Dynamics of Place and Mobility
3.1 Introduction
3.1.1 Social Norms and Historical Contexts
3.1.2 Environmental Influences and Mobility Disruptions
3.1.3 Built Environment and Points of Interest
3.2 Data Types for Temporal Research
3.2.1 Probe Data
3.2.2 Stationary Sensor Data
3.2.3 Place-Based Data
3.2.4 Human-Centric Data
3.3 What’s Going On
3.4 Discussion, Conclusion, and Opportunities
3.4.1 Common Grounds
3.4.2 Data Privacy
3.4.3 Data Ownership
3.4.4 Data Quality and Transparency
References
Chapter 4 Geospatial Knowledge Graph Construction Workflow for Semantics-Enabled Remote Sensing Scene Understanding
4.1 Introduction and Motivation
4.1.1 Image Information Mining for Earth Observation (EO)
4.1.2 Semantic Web
4.1.3 Ontologies and Reasoning
4.1.4 Geospatial Knowledge Representation
4.1.4.1 Ontology-Based Remote Sensing Image Analysis
4.1.4.2 Ontology-Based Approaches for Disaster Applications
4.1.4.3 Knowledge Graphs
4.2 Geospatial Knowledge Graphs Construction
4.2.1 Knowledge Graph Construction Workflow
4.2.1.1 Deep Learning-Based Multi-Class Segmentation
4.2.1.2 Geometry Shape Extraction
4.2.1.3 Resource Description Framework (RDF) Based Serialization
4.2.1.4 Semantic Enrichment of Geospatial KG
4.3 Applications and Use Cases
4.4 Summary
Notes
References
Chapter 5 Geosemantic Standards-Driven Intelligent Information Retrieval Framework for 3D LiDAR Point Clouds
5.1 Introduction and Motivation
5.2 LiDAR—Light Detection and Ranging
5.2.1 Types of LiDAR Data Sources
5.2.2 List of Remote Sensing-Based Open LiDAR Datasets
5.3 Interoperability and Geosemantics Standardization for LiDAR
5.3.1 Need for Interoperability in LiDAR
5.3.2 Geospatial Standardization
5.3.2.1 International Bodies for Geospatial Standardization
5.3.2.2 Geospatial Standards for 3D LiDAR Data
5.3.3 Designing a Geosemantic Standards-Driven Framework for 3D LiDAR Data
5.3.3.1 LiDAR Markup Language (LiDARML)—Toward Interoperability for LiDAR
5.4 Development of a Scalable LiDAR Information Mining Framework: A Systems Perspective
5.4.1 Geo-Artificial Intelligence (GeoAI) Module for 3D LiDAR Point Cloud Processing
5.4.2 GeoSemantics Module: Toward Semantically Enriched LiDAR Knowledge Graph
5.5 Case Study: Knowledge Base Question-Answering (KBQA) Framework for LiDAR
5.5.1 Dataset Details
5.5.2 Problem Formulation: Knowledge Base Question-Answering (KBQA) for LiDAR
5.5.3 Generating LiDAR Scene Knowledge Graph—LiSKG
5.5.4 Natural Language to GeoSPARQL in Knowledge Base Question-Answering (KBQA)
5.6 Summary and Future Trends
5.7 Where to Look for Further Information
Acknowledgments
Notes
References
Chapter 6 Geospatial Analytics Using Natural Language Processing
6.1 Introduction
6.1.1 Geospatial Analytics
6.1.1.3 Sources of Geotext
6.1.2 Introduction to Natural Language Processing
6.1.3 Geospatial Data Meets NLP
6.1.3.1 Researchers Interest in Geospatial Data from Text Using NLP
6.2 Overview of NLP Techniques in Geospatial Analytics
6.2.1 Event Extraction
6.2.2 Parts-of-Speech (POS) Tagging
6.2.3 Temporal Information Extraction
6.2.4 Spatial-Temporal Relationship Extractions
6.2.5 Named Entity Recognition (NER)
6.3 Applications of NLP in Geospatial Analytics
6.3.1 Geoparsing and Toponym Disambiguation
6.3.1.1 Geoparsing
6.3.2 Geospatial Geosemantic in Natural Language
6.3.2.1 Role of NLP in Geosemantic
6.3.2.2 Geosemantic Similarity Using NLP
6.3.3 Geospatial Information Analysis
6.3.3.1 Geospatial Information Extraction (GIE)
6.3.3.2 Geospatial Information Retrieval
6.3.3.3 Geospatial Question Answering
6.3.4 Spatiotemporal/Geospatial Text Analysis from Social Media
6.4 Future Scope of NLP in Geospatial Analytics
6.5 Summary and Conclusions
Notes
References
Section III Scalable Geospatial Analytics
Chapter 7 A Scalable Automated Satellite Data Downloading and Processing Pipeline Developed on AWS Cloud for Agricultural Applications
7.1 Introduction
7.2 Satellite Imagery Resolutions
7.3 Application of Technology in Monitoring Crop Health
7.4 High-Level Solution—Crop Health Monitoring Using Satellite Data
7.5 AWS Components Used in the Solution
7.6 Some of the Key Advantages of Having AWS Based Data Pipeline Were
7.7 Detailed Solution
7.7.1 Key Steps of the ADDPro Pipeline
7.8 Sample Analysis for Field-Level Crop Health
7.9 Time Series of Satellite Data and Crop Condition Information
7.10 Conclusion
Acknowledgment
References
Chapter 8 Providing Geospatial Intelligence through a Scalable Imagery Pipeline
8.1 Geospatial Intelligence R&D Challenges
8.1.1 Challenges to Advancing Geospatial Intelligence
8.1.1.1 Compute Power and Startup Costs
8.1.1.2 Scalability
8.1.1.3 Speed and Resolution
8.1.1.4 Data Privacy and Security
8.1.2 ORNL High-Performance Computing Resources
8.2 Pushing the Boundaries of Geospatial Intelligence
8.2.1 Enabling Research
8.2.2 Mapping
8.2.3 Large-Scale Modeling
8.3 Building the Imagery Pipeline
8.3.1 Imagery Ingest
8.3.2 Orthorectification
8.3.3 Pan-Sharpening
8.3.4 Cloud Detection
8.3.5 Postprocessing and Output
8.4 Future Considerations
8.4.1 Adding Atmospheric Compensation to the Pipeline
8.4.2 Leveraging Cloud Computing to Advance Our Imagery Processing Capabilities
8.4.3 Adapting Pipe to Other Applications
Acknowledgments
Notes
References
Chapter 9 Distributed Deep Learning and Its Application in Geo-spatial Analytics
9.1 Introduction
9.2 High-performance Computing (HPC)
9.2.1 Need for High-performance Computing
9.2.2 Parallel Computing
9.2.3 Distributed Computing
9.2.3.1 Distributed Computing for Geo-Spatial Data
9.2.4 Challenges in High-performance Computing
9.3 Distributed Deep Learning for Geo-Spatial Analytics
9.3.1 Distributed Deep Learning
9.4 Apache Spark for Distributed Deep Learning
9.4.1 Distributed Hyper-Parameter Optimization
9.4.2 Deep Learning Pipelines
9.4.3 Apache Spark on Deep Learning Pipeline
9.5 Applications of Distributed Deep Learning in Real World
9.6 Conclusion Summary and Perspectives
References
Chapter 10 High-Performance Computing for Processing Big Geospatial Disaster Data
10.1 Introduction
10.2 Recent Advances in High-Performance Computing for Geospatial Analysis
10.3 Damage Assessment and Sources of Disaster Data
10.3.1 Images
10.3.1.1 Airborne
10.3.1.2 Satellite
10.3.2 LiDAR
10.4 Key Components of High-Performance Computing
10.4.1 Domain Decomposition
10.4.2 Spatial Indexing
10.4.3 Task Scheduling
10.4.4 Evaluation Metrics
10.5 Hardware and Its Programming Model
10.5.1 Graphics Processing Unit
10.5.2 General Architecture of GPU
10.5.3 Jetson Nano—Embedded HPC
10.6 HPC for Building Damage Detection for Earthquake-Affected Area
10.6.1 Point Cloud Outlier Removal
10.6.2 Buildings Extraction
10.6.3 Iterative Closest Point
10.6.4 Classification Results
10.7 Summary and Future Work
References
Section IV Geovisualization: Innovative Approaches for Geovisualization and Geovisual Analytics for Big Geospatial Data
Chapter 11 Dashboard for Earth Observation
11.1 Introduction
11.2 Canonical Use Cases and High-Level Requirements (Science)
11.2.1 COVID-19 Dashboard
11.2.2 MAAP Dashboard
11.2.3 Community Workshops, Tutorials, and Hackathons
11.3 Technology Landscape Analysis
11.3.1 Data Stores
11.3.2 Data Processing
11.3.3 Data Services
11.3.3.1 Discovery Services
11.3.3.2 Data Access Services
11.3.3.3 Mapping Services
11.3.4 Visualization Front End
11.3.4.1 Visualization Libraries
11.3.4.2 Technical Considerations
11.3.4.3 Dynamic Tilers
11.3.4.4 User Interactivity and Engaging End User Experience
11.3.4.5 Map Projections
11.3.4.6 Analysis Clients
11.4 VEDA
11.4.1 Overview
11.4.2 Implementation
11.4.2.1 Federated Data Stores
11.4.2.2 Data Processing (Extract, Transform, and Load)
11.4.2.3 Data Services
11.4.2.4 APIs
11.4.2.5 Data Visualization, Exploration, and Analysis Clients
11.5 Summary
References
Chapter 12 Visual Exploration of LiDAR Point Clouds
12.1 Introduction
12.2 Visualization Systems for Airborne LiDAR Point Clouds
12.3 Distributed System Architecture for Visual Analytics Tool
12.3.1 Browser-based Visualization Tool
12.3.1.1 Canvas Rendering Using BufferGeometry
12.3.1.2 Asynchronous and Parallel Processing
12.3.1.3 Service Interface
12.3.2 Distributed System for Semantic Classification
12.3.3 Backend Services
12.4 System Implementation
12.4.1 Visualization Tasks
12.4.2 Graphical User Interface Design
12.4.2.1 Navigation Controls on the Canvas
12.4.2.2 Classification Tool
12.4.2.3 Analytics Widgets
12.4.2.4 Selection and Exploration of Regions of Interest
12.4.3 Subsampling for Efficient Rendering
12.4.4 Custom Partitioning in Spark
12.4.5 Distributed Data Model in Cassandra
12.4.6 System Specifications
12.5 Case Study: Visual Analytics of Airborne LiDAR Point Clouds
12.6 Conclusions
Acknowledgment
Note
References
Section V other Advances in Geospatial Domain
Chapter 13 Toward a Smart Metaverse City: Immersive Realism and 3D Visualization of Digital Twin Cities
13.1 Introduction
13.2 Metaverse for Digital Twin Cities
13.2.1 Digital Twin Cities
13.2.2 Metaverse
13.2.3 Smart Metaverse City
13.2.3.1 Immersive Realism
13.2.3.2 Scientific Virtual Object Creation
13.2.3.3 Public Engagement through Avatars
13.2.4 Potential Applications
13.3 Geospatial Framework for Metaverse Cities
13.3.1 Overall Framework Design
13.3.2 Geospatial Data Acquisition
13.3.3 Digital Twin City Construction
13.3.4 Immersive 3D Geovisualization
13.4 Use Cases
13.4.1 Public Engagement, Education, and Training
13.4.2 Training Computer Vision Applications
13.4.3 Spatial Co-simulation
13.5 Future Opportunities
Disclaimer
Acknowledgement
References
Chapter 14 Current UAS Capabilities for Geospatial Spectral Solutions
14.1 History
14.2 Current State of the Art
14.2.1 Types of Platforms
14.2.1.1 Multirotor
14.2.1.2 Fixed-Wing
14.2.1.3 Hybrid Airframes
14.2.2 Propulsion Systems
14.2.3 Sensor Payloads
14.2.3.1 RGB
14.2.3.2 Multispectral
14.2.3.3 Hyperspectral
14.2.3.4 Thermal/LWIR
14.2.3.5 LiDAR
14.2.3.6 SAR
14.2.3.7 Sensor Precautions
14.2.4 Use Cases/Literature Review
14.2.4.1 Statistics
14.2.4.2 Agricultural
14.2.4.3 Forestry
14.2.4.4 Riparian Zones
14.2.4.5 Wetlands and Coastal Systems
14.2.4.6 Land Surface Temperature
14.2.4.7 Animals
14.2.4.8 Archeology
14.2.4.9 Atmospheric Dynamics
14.2.4.10 Optimization
14.2.4.11 Automation
14.3 Communications
14.3.1 Ground Control
14.3.2 Networked UAS
14.4 Processing Techniques
14.4.1 Onboard Processing
14.4.2 Postprocessing
14.5 Current Issues
14.6 Future Directions
14.6.1 Sensors
14.6.2 Processing
14.6.3 Communications
14.7 Conclusion
References
Chapter 15 Flood Mapping and Damage Assessment Using Sentinel – 1 & 2 in Google Earth Engine of Port Berge & Mampikony Districts, Sophia Region, Madagascar
15.1 Introduction
15.1.1 Background
15.2 Study Area
15.3 Data Used and Methodology
15.3.1 Data Used
15.3.2 Methodology
15.4 Results and Discussions
15.4.1 Flood Inundation Map
15.4.2 Land Use/Land Cover Map
15.4.3 Flood Damage Assessment
15.5 Conclusions
Acknowledgments
References
Section VI case Studies from the Geospatial Domain
Chapter 16 Fuzzy-Based Meta-Heuristic and Bi-Variate Geo-Statistical Modelling for Spatial Prediction of Landslides
16.1 Introduction
16.2 LSZ Mapping and Associated Modeling Approaches
16.2.1 Fuzzy Set Theory and FAHP
16.2.1.1 Extent Analysis on FAHP
16.2.1.2 Triangular Fuzzy MF
16.2.1.3 Fuzzy Operational Laws
16.2.2 Yule Coefficient
16.3 Application of RS and GIS in LSZ Studies
16.4 Application of FAHP and YC Models for LSZ Mapping in Parts of Kalimpong Region of Darjeeling Himalaya
16.4.1 Description of the Area
16.4.2 ThemSatic Layer Development
16.4.2.1 Landslide Inventory
16.4.2.2 Landslide Causative Factors
16.4.3 Model Implementation
16.4.3.1 Factors Weights Determination Using FHAP Model
16.4.3.2 Factors Subclasses Weights Determination Using YC Model
16.4.4 Landslide Susceptibility Zonation (LSZ) and Validation
16.5 Discussion and Conclusion
Acknowledgments
Funding
Availability of Data and Material
Code Availability
Declarations
Conflicts of Interest
References
Chapter 17 Understanding the Dynamics of the City through Crowdsourced Datasets: A Case Study of Indore City
17.1 Introduction, Background and Need of Study
17.2 Literature Review
17.2.1 Location Intelligence
17.2.2 Rise of Social Media and Urban Datasets
17.2.3 Using the Social Media/Urban Datasets for Various Urban Studies
17.2.4 Different Approaches and Clustering-Based Algorithms
17.2.5 Incorporation of Text-Based Classification
17.3 Framework and Methodology
17.3.1 Identification of Platform for Extraction of Data
17.3.2 Case Study – Indore City
17.3.3 Data Collection/Extraction
17.3.4 Data Analysis
17.3.5 Limitations of the Study
17.4 Observation and Inferences
17.4.1 Activity Mapping
17.4.2 Landuse Change Detection
17.4.3 Point of Interest (POI)
17.4.4 Sentiment Mapping
17.5 Conclusion and Way Forward
References
Chapter 18 A Hybrid Model for the Prediction of Land Use/Land Cover Pattern in Kurunegala City, Sri Lanka
18.1 Introduction
18.2 Method and Materials
18.2.1 The Study Area
18.2.2 Data Source
18.2.3 Data Pre-Processing
18.2.4 Data Analysis
18.2.4.1 Supervised Classification
18.2.4.2 Selection of Drivers Variables
18.2.4.3 Multi-Layer Perceptron Neural Network and CA-Markov Model
18.3 Results and Discussion
18.3.1 Land Use/Land Cover Pattern
18.3.2 Land Use/Land Cover Changes
18.3.3 Land Use/Land Cover Simulation and Prediction
18.4 Conclusion
References
Chapter 19 Spatio-Temporal Dynamics of Tropical Deciduous Forests under Climate Change Scenarios in India
19.1 Introduction
19.2 Materials and Methods
19.2.1 Study Area
19.2.2 Data Used
19.2.2.1 Forest Cover Data
19.2.2.2 Predictors Used
19.2.3 Data Processing
19.2.4 Random Forest Model Building
19.2.5 Model Evaluation and Spatial Prediction
19.3 Results
19.3.1 Pearson Correlation
19.3.2 Accuracy and Predictors Importance
19.3.3 Spatial Prediction of Tropical Deciduous Forest Cover
19.4 Discussion
19.5 Conclusion
Acknowledgment
References
Chapter 20 A Survey of Machine Learning Techniques in Forestry Applications Using SAR Data
20.1 Introduction
20.2 SAR and Machine Learning
20.3 Forest Classification
20.4 Forest Degradation/Deforestation Mapping
20.5 Forest Tree Height Estimation
20.6 Forest Biomass Estimation
20.7 Future Perspective and Conclusion
References
Index