Spatio-temporal analysis sits right at the intersection of statistics, data science, GIS, and environmental applications.
📚 Step 1: Build Foundations
Spatial Statistics
- Learn point pattern analysis, spatial autocorrelation (Moran’s I, Geary’s C), variograms, and kriging.
- Core book: Applied Spatial Data Analysis with R (Bivand, Pebesma, Gómez-Rubio).
Time Series / Temporal Analysis
- Refresh time series models (ARIMA, state-space, spectral methods).
- Add temporal clustering and forecasting methods.
Spatio-Temporal Models
- Combine both: spatio-temporal kriging, Gaussian random fields, Bayesian hierarchical models.
- Book: Hierarchical Modelling of Spatial Data Using R-INLA (Blangiardo & Cameletti).
🛠 Step 2: Learn the Tools
R packages
- sf (modern spatial objects)
- stars (spatio-temporal arrays)
- terra (raster/vector operations)
- spacetime (older but still used)
- gstat (variograms, kriging)
- spdep (spatial dependence)
- INLA (Bayesian spatio-temporal modeling)
- Python equivalents (optional, if you want dual skillset): geopandas, xarray, rasterio, pySTL, pysal, scikit-mobility
🔬 Step 3: Practice with Data
Sources of free spatio-temporal datasets:
- Earth observation: NASA EarthData, Copernicus (Sentinel satellites)
- Climate: ERA5 reanalysis (ECMWF)
- Health: COVID-19 spatio-temporal data (Johns Hopkins)
- Urban/mobility: OpenStreetMap, Google Mobility Reports
- You could start with simple projects:
- Mapping air quality trends across time.
- Spatio-temporal spread of disease (e.g., dengue in Bangladesh).
- Detecting land cover change from satellite images.
🤝 Step 4: Contribute
Open-Source Contributions
- Help improve documentation or examples in sf, stars, or spacetime.
- Build teaching vignettes for gstat or spdep.
Research Contributions
- Apply spatio-temporal models to a new real-world problem (climate, urban growth, seismicity).
- Publish datasets + reproducible code (on GitHub + Zenodo).
Community Involvement
- Join R-Spatial community (https://r-spatial.org).
- Follow Edzer Pebesma, Roger Bivand, and Virgilio Gómez-Rubio.
- Present small case studies in workshops/webinars.
⚡ A good starting project for you could be:
“Spatio-temporal visualization of earthquake aftershocks in Bangladesh/Asia using USGS data”