landsat_change_detection - GitLite

KO | EN

gitlite — search

#cpp #hacktoberfest #graphql #terminal #privacy #cross-platform #java #javascript #gpt #go

landsat_change_detection

(★ 11)

No description available.

File Explorer

ZIP Download

.gitignore
.python-version
landsat_vegetation_change_h3.ipynb
pyproject.toml
README.md
uv.lock

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

Command Glossary

Commands referenced in this DOCs, explained below.

🔍

`duckdb`

View Details ▼

Client for DuckDB, an in-process analytical SQL engine.

duckdb

Start an interactive shell with a transient in-memory database:

duckdb {{path/to/dbfile}}

Start an interactive shell on a database file. If the file does not exist, a new database is created:

duckdb -c "{{SELECT * FROM 'data_source.[csv|csv.gz|json|json.gz|parquet]'}}"

Query a CSV, JSON, or Parquet file using SQL:

🔍

`git clone`

View Details ▼

Clone an existing repository.

git clone {{remote_repository_location}} {{path/to/directory}}

Clone an existing repository into a new directory (the default directory is the repository name):

git clone --recursive {{remote_repository_location}}

Clone an existing repository and its submodules:

git clone {{[-n|--no-checkout]}} {{remote_repository_location}}

Clone only the `.git` directory of an existing repository:

🔍

`uv sync`

View Details ▼

Update the project's environment to match the lockfile.

uv sync

Sync the project environment with the lockfile:

uv sync --all-extras

Sync and include all optional dependencies:

uv sync --extra {{extra_name}}

Sync with specific optional dependencies:

🔍

`uvx`

View Details ▼

This command is an alias of `uv tool run`.

tldr uv tool

View documentation for the original command:

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

Detect vegetation changes over time using Landsat imagery aggregated to H3 hexagons.

Compares EVI (Enhanced Vegetation Index) and NDVI (Normalized Difference Vegetation Index) between two time periods (e.g., 1990s vs 2020s) and outputs change metrics per H3 cell.

<img width="1244" height="533" alt="image" src="https://github.com/user-attachments/assets/68e72eff-9ebb-4b14-86cc-db841f99a54b" />

<img width="1250" height="601" alt="Screenshot 2026-01-10 at 3 31 59 PM" src="https://github.com/user-attachments/assets/8b232464-16a1-41f5-aecd-ab0f43e74ee1" />

## Background

**Case Study: Chernobyl Exclusion Zone**

The 1986 Chernobyl disaster was catastrophic—the worst nuclear accident in history. Yet in the decades since evacuation, the Exclusion Zone has become an unintentional experiment in what happens when humans withdraw from a landscape. This notebook uses satellite imagery to examine one aspect of that story: vegetation change between the early 1990s and 2020s.

Inspired by [Cal Flyn](https://www.calflyn.com/)'s *Islands of Abandonment*, which includes her trip to Chernobyl and Pripyat before the Russian invasion.

I think this type of tool could be useful for analysis of change over time and can be repurposed for different types of sensors and other spectral indices.

## Setup & Usage

Install [uv](https://docs.astral.sh/uv/getting-started/installation/) and clone the repo:

```bash
curl -LsSf https://astral.sh/uv/install.sh | sh  # or: brew install uv
git clone https://github.com/kentstephen/landsat_change_detection.git
cd landsat_change_detection
```

**Option A: One-liner with juv** - runs notebook with inline dependencies (PEP 723):
```bash
uvx juv run landsat_vegetation_change_h3.ipynb
```

**Option B: Install to venv** - use `uv sync` then open notebook in your IDE:
```bash
uv sync
```

## Data Sources

This project supports two STAC endpoints for Landsat data:

### Option 1: Element84 Earth Search

Uses requester-pays S3 buckets. You will need to use new or existing AWS secrets. For purposes of accessing Landsat imagery in S3 this way, the IAM user defined for this purpose will only need `AmazonS3ReadOnlyAccess`.

```bash
# Create .env file with your AWS credentials
AWS_ACCESS_KEY_ID=your_key
AWS_SECRET_ACCESS_KEY=your_secret

# Optional: Mapbox token for satellite basemap with labels
MAPBOX_TOKEN=your_mapbox_token
```

It cost me less than 20 cents of compute to develop this example using AWS/Element84.

### Option 2: Microsoft Planetary Computer

This notebook uses `sign_inplace` so no extra credentials are needed. However, AWS/Element84 is generally more reliable.

## How It Works

1. Query STAC catalog for Landsat scenes in your AOI
2. Load imagery with odc-stac (uses dask for parallel S3 fetching)
3. Calculate EVI or NDVI per pixel
4. Aggregate to H3 hexagons using DuckDB's H3 extension
5. Compare early vs recent periods to compute change

## Core Dependencies

- [`odc-stac`](https://github.com/opendatacube/odc-stac) - Cloud-native raster loading
- [`duckdb`](https://github.com/duckdb/duckdb) + [H3 extension](https://github.com/isaacbrodsky/h3-duckdb) - Fast spatial aggregation
- [`h3`](https://github.com/uber/h3-py) - Uber's hexagonal grid system
- [`lonboard`](https://github.com/developmentseed/lonboard) - GPU-accelerated map visualization
- [`dask`](https://github.com/dask/dask) - Parallel array computing with task scheduling
- [`pyarrow`](https://github.com/apache/arrow) - Apache Arrow for zero-copy data interchange

*This notebook is not meant to be conclusive science - just a means of exploration*