Geospatial Data Analysis, Prediction, and GeoAI:
New Theories, Methods, and Software
This lecture series provides a comprehensive overview of the latest theories, analytical methods, and software tools for spatial research and applications. Students will learn how to design and conduct spatial analysis, handle diverse geospatial and remote sensing data, and apply advanced modeling and GeoAI techniques to address complex problems in sustainability, urban development, natural, built and social environments, and decision-making.
By completing this eight-topic course, students will develop the skills required to conduct end-to-end spatial analysis using advanced models and software, and will be equipped to apply these methods to practical problems and high-level research publications.
Download the Lecture Outline and Reading List (PDF) for foundational references and topic-by-topic readings.
- Each day, write a 200-word essay on that day's topic.
- Submit a combined Word document with essays for at least four days out of the eight.
- Submission: PDF file (named with student's name), emailed to yongze.song@outlook.com
- IEEE IGARSS Summer School Training Materials: https://ausgis.github.io/igarss25ss/
Topic 1. Introduction to Spatial Modelling and Prediction
Why spatial methods are needed for prediction; GeoAI concepts, history, and emerging trends; sustainable infrastructure case study; OPGD model and the GD R package for determinant analysis.
Topic 2. Spatial Modelling for Heterogeneity and Determinant Analysis
Concepts of local and spatial stratified heterogeneity; the GOZH model for optimal zone-based analysis; LISP for local driver analysis.
Topic 3. Spatial Modelling for Prediction
General process of spatial prediction; Kriging fundamentals and geostatistics; Second-Dimension Spatial Association (SDA); Geographically Optimal Similarity (GOS) model.
Topic 4. Spatial Patterns for Prediction
Spatial pattern identification and quantification (clusters, gradients, hotspots); Geocomplexity theory and measurement; Second-Dimension Outliers (SDO) for prediction.
Topic 5. Research Design, Spatial & Remote Sensing Data Collection, and Pre-processing
Experiment design for spatial prediction; data collection strategies; spatial stratified random sampling; pre-processing workflow (missing values, multicollinearity, outliers, normalization).
Topic 6. Spatial Validation Methods
Why spatial validation is critical; k-fold cross-validation; spatial block cross-validation; comparison of random CV vs. block CV vs. NNDM approaches.
Topic 7. Spatial Decision-Making and Applications
Model-based spatial decision-making; entropy-weighting for multi-criteria aggregation; constructing composite urbanization indices; spatial mapping and visualization.
Topic 8. Geospatial Intelligence (GeoAI) and Spatial Big Data for Urban Sustainability
GeoAI definition, scope, and history; emerging trends (heterogeneity-aware GeoAI, geo-foundation models, spatial representation learning); machine learning with GAM using the caret package.
Answers to students' questions are continuously updated here.
- Topic 4
- analysis code.R, line 27: change sf_data to sf.data
- Topic 5
- Australia extent:
ext_aus <- ext(112, 154, -44, -10) - China extent:
ext_china <- ext(73, 135, 18, 54) - Canada:
ext_canada <- ext(-141, -52, 41, 83) - Europe:
ext_europe <- ext(-25, 45, 34, 72) - Africa:
ext_africa <- ext(-20, 55, -35, 38) - South America:
ext_south_america <- ext(-82, -34, -56, 13) - US:
ext_us <- ext(-125, -66.5, 24, 49.5)
- Australia extent: