---
name: olmo-earth-skill-876858987222
description: "Sentinel-2 satellite imagery to OlmoEarth embeddings on geodesic tiles, serialized via zig-syrup. Use for satellite imagery pipelines, earth observation embeddings, or WholeEarthModel operations."
---

# Whole Earth Model: Sentinel-2 + OlmoEarth + Zig-Syrup

Downloads Sentinel-2 satellite imagery from Microsoft Planetary Computer (free, no account needed), runs AI2's OlmoEarth foundation model to produce per-tile embeddings, and writes them into the zig-syrup WholeEarthModel binary format for CapTP exchange.

```
Planetary Computer (free Sentinel-2 COGs)
    |
    v
Download 12-band L2A tiles (10-60m resolution)
    |
    v
OlmoEarth FlexiVit Encoder (768-dim Base model)
    |
    v
Per-tile embeddings as f32 vectors
    |
    v
Binary file -> Zig EmbeddingLoader -> WholeEarthModel.setEmbedding()
    |
    v
Syrup serialization -> CapTP exchange at ~400 Hz
```

## Usage

### Download + Embed a region

```bash
python /Users/alice/.claude/skills/olmo-earth-skill/sentinel2_olmoearth.py \
    --lat 37.77 --lon -122.42 \
    --size 0.1 \
    --start 2024-07-01 --end 2024-08-31 \
    --max-cloud 10 \
    --model base \
    --output /tmp/sf_bay_embeddings.bin
```

### Load into zig-syrup WholeEarthModel

```zig
const bridge = @import("olmoearth_bridge");
const earth = @import("whole_earth");

var model = try earth.WholeEarthModel.init(allocator, .level_3, 768);
defer model.deinit();

var loader = bridge.EmbeddingLoader.init(allocator, .base);
var embeddings = try loader.loadFromFile("/tmp/sf_bay_embeddings.bin");
defer embeddings.deinit();

for (0..embeddings.tile_count) |i| {
    if (embeddings.getEmbedding(@intCast(i))) |emb| {
        const tile_id = model.tileAt(lat, lon) orelse continue;
        try model.setEmbedding(tile_id, emb);
    }
}

const syrup_val = try model.toSyrup(allocator);
```

### Compare tile embeddings over time

```bash
python sentinel2_olmoearth.py --lat 37.77 --lon -122.42 \
    --start 2024-01-01 --end 2024-03-31 --output /tmp/sf_q1.bin

python sentinel2_olmoearth.py --lat 37.77 --lon -122.42 \
    --start 2024-07-01 --end 2024-09-30 --output /tmp/sf_q3.bin

python sentinel2_olmoearth.py --diff /tmp/sf_q1.bin /tmp/sf_q3.bin
```

## OlmoEarth Model Specs

| Model | Embed Dim | Params | HuggingFace ID |
|-------|-----------|--------|----------------|
| Nano  | 192       | 1.4M   | `allenai/OlmoEarth-v1-Nano` |
| Tiny  | 384       | 6.2M   | `allenai/OlmoEarth-v1-Tiny` |
| Base  | 768       | 89M    | `allenai/OlmoEarth-v1-Base` |
| Large | 1024      | 308M   | `allenai/OlmoEarth-v1-Large` |

## Binary Embedding File Format

Readable by `olmoearth_bridge.zig` `EmbeddingLoader.loadFromFile()`:

```
Offset  Size    Field
0       4       tile_count (u32 little-endian)
4       2       embed_dim (u16 little-endian)
6       4*N*D   embeddings (f32 little-endian, row-major)
6+4*N*D 8*N     coordinates (f64 lat, f64 lon per tile, little-endian)
```

Where N = tile_count, D = embed_dim.

## Data Source

STAC endpoint: `https://planetarycomputer.microsoft.com/api/stac/v1`
Collection: `sentinel-2-l2a`
The `planetary-computer` Python package handles URL signing transparently.


---
Source attribution: plurigrid
Original project: https://github.com/plurigrid/asi/blob/293ce9f76e3abcd5ba18db284688bc0d9ac6691d/skills/olmo-earth-skill/SKILL.md
License: MIT
License information: https://opensource.org/license/mit
Original upstream instructions; downloaded from Mega AI Hub.

## Upstream license notice

MIT License

Copyright (c) 2025 SkillCreator.ai

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.



# Mega AI Hub resource metadata

Creator / publisher: plurigrid
Original project: https://github.com/plurigrid/asi/blob/293ce9f76e3abcd5ba18db284688bc0d9ac6691d/skills/olmo-earth-skill/SKILL.md
Attribution basis: Source repository record

Fields of work: Data & Analytics, Development & Engineering, Product Management
Roles: Data Scientist, ML / AI Engineer, Product Manager
Specialties: API
Tools: Microsoft Planetary Computer, Sentinel 2, Zig Syrup
Country applicability: Country not specified
Classification basis: Source catalog labels
Instruction SHA-256: ce66ba65bc1e2c07498cc250e89c77f3fa8e11811beab4c21c2348b6eb81b2f9