Our project leverages cutting-edge remote sensing to identify and monitor land cover changes, visualizing our changing world with effective and efficient data sources. By integrating advanced algorithms and datasets, we provide a comprehensive tool for understanding the dynamics of our changing planet — supporting researchers and professionals across geography and geology to detect, measure, and analyze the changes that interest them.
Click here for an interactive dashboard.
For a narrative walkthrough, open notebooks/AlphaEarth_Story.ipynb (visualizations are interactive and cannot be rendered on GitHub; see screenshots below).
Mentor: Dr. Brendon Hall — Sr. Manager in AI for Energy & Utilities, Deloitte
conda env create -f environment.yml
conda activate alphaearthearthengine authenticate
python -m ipykernel install --user --name alphaearth --display-name "alphaearth"Need a GEE project? Register at earthengine.google.com. Free tier is sufficient.
jupyter labRun in this order:
| # | Notebook | Purpose |
|---|---|---|
| 1 | notebooks/alpha_tutorial.ipynb |
AlphaEarth K-means clustering intro |
| 2 | notebooks/AlphaEarth_Story.ipynb |
Main narrative: cosine similarity, dam detection, Austin urban growth |
| 3 | notebooks/AlphaEarth_Interactive_Mapping.ipynb |
Draw AOI → real-time change detection + inspector |
| 4 | notebooks/AlphaEarth_LandTrendr_ChangeComparison.ipynb |
Side-by-side comparison across 15 sites + IoU analysis |
| 5 | notebooks/LandTrendr_AlphaEarth.ipynb |
Deep-dive LandTrendr statistics; exports CSVs to outputs/ |
python -m scripts.smoke_testDetect and compare land-use change across 15 western U.S. sites using two complementary algorithms:
| Algorithm | Method | Time range |
|---|---|---|
| AlphaEarth | Cosine similarity on 64-dim satellite embeddings (GOOGLE/SATELLITE_EMBEDDING/V1/ANNUAL) |
2017–2024 |
| LandTrendr | Spectral-temporal NDVI trajectory fitting (Landsat) | 2016–2024 |
Both produce three standardized layers — YOD (year of change), MAG (magnitude), DUR (duration) — and IoU analysis compares their overlap.
Study sites (15):
| Category | Sites |
|---|---|
| Urbanization | Austin TX, Dallas TX, Houston TX, Bend OR, Portland OR, Sacramento CA |
| Wildfires | Bootleg OR (2021), Camp Fire CA (2018), Dixie CA (2021), Mosquito CA (2022), Santiam OR (2020) |
| Forest / Logging | Angelina TX, Coos Bay OR, Mt Hood OR, Shasta-Trinity CA |
AlphaEarthHack/
├── notebooks/ # Main Jupyter notebooks (run in order above)
├── scripts/ # smoke_test.py
├── src/ # Shared constants (EE asset paths, thresholds)
├── images/ # Figures referenced in notebooks and README
├── outputs/ # Generated CSVs / PNGs (git-ignored)
├── data/ # Local data placeholder (git-ignored)
├── backup/ # Archived exploratory notebooks
├── environment.yml # Conda environment spec (Python 3.12)
└── pyproject.toml # Ruff lint config
| Global View | Years Side-By-Side |
|---|---|
 |
 |
| Cosine Similarity (areas of change) | Similar Feature Detection |
|---|---|
 |
 |
| AlphaEarth Magnitude | AlphaEarth Duration |
|---|---|
 |
 |
| LandTrendr Magnitude | LandTrendr Duration |
|---|---|
 |
 |
| Interactive Dissimilarity Plot | Export Options |
|---|---|
 |
 |
Google DeepMind · Google AlphaEarth · LandTrendr · GeoAI · NumPy · tqdm · ipyleaflet · Google Earth Engine · Leaflet.js · Plotly.js
Brown, C. F., et al. (2025). AlphaEarth Foundations. arXiv:2507.22291.
Kennedy, R. E., et al. (2018). Implementation of the LandTrendr Algorithm on Google Earth Engine. Remote Sensing, 10(5), 691.
Gorelick, N., et al. (2017). Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18–27.
Harris, C. R., et al. (2020). Array programming with NumPy. Nature, 585, 357–362.
Janowicz, K., et al. (2020). GeoAI. Int. J. Geographical Information Science, 34(4), 625–636.
da Costa-Luis, C. O. (2019). tqdm. Journal of Open Source Software, 4(37), 1277.
BSD 3-Clause — see LICENSE.