clms-aoi

A lightweight Python library and CLI for extracting and summarising Copernicus Land Monitoring Service (CLMS) data for any area of interest, via Sentinel Hub's Statistical API.

Point it at a boundary file, pick a product and year, and get back a pandas DataFrame, a CSV, and/or a bar chart — no hand-rolled Sentinel Hub request boilerplate required.

Status: early (0.1.0). The Python API is the primary interface today; the CLI currently covers config/auth/AOI validation only (see CLI usage).

Who is this for?

Land use analysts, GIS practitioners, students, NGOs, and small consultancies who want quick AOI-level summaries without setting up a full geospatial processing pipeline.

Supported products

Grasslands is not currently implemented.

Installation

For users:

pip install clms-aoi

For devs:

1. Clone the repository
2. Change directory into the cloned folder

cd clms-aoi

3. Create a virtual environment and activate it

python3 -m venv clms_aoi_env && source clms_aoi_env/bin/activate

4. Do an editable install which means edits to the source code keep working without reinstalling

pip install -e .

5. To run tests for this project

pip install pytest

and run:

pytest

Configuration

Create a YAML config file with your Sentinel Hub credentials (get these from the Copernicus Data Space Ecosystem):

sentinelhub:
  client_id: "your-client-id"
  client_secret: "your-client-secret"

Credentials can also be supplied via the CLMS_SH_CLIENT_ID / CLMS_SH_CLIENT_SECRET environment variables instead of (or as a fallback for) the YAML file.

Never commit a config file containing real credentials. Keep your local config out of version control (it's covered by .gitignore by default) — a config file with a placeholder/example only should be committed, if any.

CLI usage

The CLI currently covers configuration and AOI validation. Product analysis (land cover, forest type, etc.) is available through the Python API below.

# Validate a config file's structure
clms-aoi validate-config config/config.yaml

# Check Sentinel Hub credentials are valid
clms-aoi check-auth --config config/config.yaml

# Validate an AOI boundary file (.geojson or .gpkg)
clms-aoi check-aoi --aoi tests/data/test_aoi.geojson
clms-aoi check-aoi --aoi tests/data/test_aoi.gpkg

# Run with verbose (debug) logging
clms-aoi -v check-aoi --aoi tests/data/test_aoi.geojson

# Errors are reported cleanly instead of raising a traceback
clms-aoi check-aoi --aoi missing.geojson

Options

Python API

You can call the library directly from notebooks or pipelines:

from clms_aoi import LandCover, ForestType

lc = LandCover(config_path="config.yaml")

# Single year
result = lc.analyse(aoi="boundary.geojson", year=2020)

# Multiple years
result = lc.analyse(aoi="boundary.geojson", years=[2018, 2021, 2024])

result.to_csv("landcover.csv")
result.to_chart("landcover.jpg")

# Access the underlying DataFrame directly
result.data

# Fetch and display the color-mapped map for a single year
lc.visualize(aoi="boundary.geojson", year=2020)

ForestType has the same shape:

from clms_aoi import ForestType

ft = ForestType(config_path="config.yaml")
result = ft.analyse(aoi="boundary.geojson", years=[2018, 2021, 2024])
result.to_csv("forest_type.csv").to_chart("forest_type.jpg")

Pass exactly one of year= (a single int) or years= (a list/range of ints) to analyse().

Outputs

CSV / DataFrame

result.data is a tidy pandas.DataFrame with columns for class name, pixel count, percentage of the AOI, and area in hectares. When multiple years are requested, a year column is added and all years are included in a single table. result.to_csv(path) writes that table to the path you give it.

Chart (JPG)

result.to_chart(path) renders a bar chart of area per class. When multiple years are requested, the chart groups bars by class with one bar per year so trends are visible at a glance.

Dependencies

• sentinelhub — authenticated Statistical API requests
• geopandas, shapely, pyproj — AOI handling and reprojection
• numpy — raster array summarisation
• matplotlib — chart and map rendering
• pandas — tabular output
• click — CLI
• requests — Sentinel Hub OAuth token requests
• PyYAML — config parsing

Limitations and known constraints

• CLI analysis commands not yet available. land-cover/forest-type-style CLI subcommands are planned but not implemented; use the Python API in the meantime.
• Only two products are fully wired up. Dynamic Land Cover and Forest Type are exposed as LandCover/ForestType. Tree Cover Density and CLC+ Backbone exist in the codebase but aren't yet exposed as public classes.
• No Grasslands product.
• No change detection. The library reports values per year but does not compute transition matrices or gain/loss statistics between years.
• No map outputs to disk. visualize() returns/plots an image array; it does not currently save clipped raster maps to a file for you.
• AOI size. Sentinel Hub has request size and pixel count limits; very large AOIs may hit them.

Quick start

Install the package and add your Sentinel Hub credentials to a config file, then run:

from clms_aoi import LandCover

lc = LandCover(config_path="config.yaml")
result = lc.analyse(aoi="./my-boundary.geojson", year=2022)

result.to_csv("./results/land-cover-2022.csv")
result.to_chart("./results/land-cover-2022.jpg")

You'll find a CSV table and a bar chart in ./results/, ready to share.