jakenotjay

jakenotjay / aef-loader

Public

Quickly access multiple data sources of Google's Alpha Earth Foundations embeddings as virtual zarr

14
1
100% credibility
Found Mar 03, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

aef-loader is a tool for quickly accessing and organizing satellite image embeddings from the Alpha Earth Foundations dataset stored in cloud storage.

How It Works

1
🌍 Discover satellite insights

You hear about special satellite image summaries called AEF embeddings that help analyze Earth changes over years.

2
📦 Get the easy loader

You add this simple tool to your workspace to grab and organize the satellite data without hassle.

3
Pick your data source
🆓
Free public option

Tap into shared community storage at no cost, great for most users.

💰
Official paid option

Use Google's storage for the freshest data, covering your own download fees.

4
🔍 Search your area

Tell the tool the map area and years you care about, like your city from 2020 to 2023.

5
🗺️ Load organized maps

Watch as your satellite summaries load neatly grouped by map regions, ready to explore without downloading everything upfront.

6
🔄 Blend and adjust

Mix maps from different regions into one view and tweak for perfect analysis.

🚀 Unlock Earth secrets

You now have fast access to powerful satellite insights for your projects, saving time and effort.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 14 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is aef-loader?

aef-loader is a Python library for quickly accessing Google's Alpha Earth Foundations (AEF) embeddings—a yearly 64-channel dataset from DeepMind derived from satellite imagery—from GCS (requester pays) or free Source Cooperative AWS buckets. It queries indexes rapidly with bbox and year filters, then loads Cloud-Optimized GeoTIFFs (COGs) lazily as virtual Zarr via xarray DataTrees organized by UTM zone. Users get analysis-ready data cubes for reprojection, dequantization, and band splitting without full downloads.

Why is it gaining traction?

It beats rioxarray by 2x on single-tile loads via virtual-tiff and obstore async I/O, while handling multi-zone reprojections with dask for seamless composites. No auth needed for Source Coop data, and utilities like dequantize/requantize make embeddings ML-ready out of the box. Devs dig the pip-install quickstart yielding geospatial xarray workflows in minutes.

Who should use this?

Remote sensing analysts querying AEF for land cover or change detection over custom bboxes and years. ML engineers fine-tuning models on global satellite embeddings who need fast, lazy access to avoid GCS egress costs. GIS devs building earth observation pipelines with xarray and odc-geo.

Verdict

Solid pick for AEF workflows if you're okay with 1.0% credibility (14 stars, early 0.1.0)—docs shine on readthedocs with benchmarks, but test in non-prod first. Grab it for prototypes; watch for updates.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.