Overview

Altira selects DEM tiles from georeferenced rasters, downloads Copernicus, ALOS, or SRTM data, and provides elevation interpolation, height statistics, and local DEM support. All elevation arrays and queries use WGS84 ellipsoidal meters.

Installation

Use the altira Conda environment for GDAL and PROJ. From this checkout:

conda env create -f resources/envs/altira.yml
conda activate altira
python -m pip install -e .

The base package supports georeferenced raster paths, DEM downloads, coordinate queries, and local DEMs. GDAL reads raster georeferencing directly.

DEM usage

from altira import Dem

dem = Dem(cache_dir="D:/DEM", local_paths=(), resolution=30)
tiles = dem.preload_image("D:/Images/scene.tif")

heights = dem.interpolate_heights([13.3, 13.31], [52.5, 52.51])
minimum, maximum, average = dem.get_height_bounds(
    13.3, 13.31, 52.5, 52.51, compute_avg_height=True
)

preload_image returns a tuple of DemTile objects with info, heights, and transform attributes. info.path identifies the unchanged native source raster; heights is a float32 array converted to ellipsoidal meters. Raster paths must provide an affine transform and CRS. GDAL reads these directly, including projected and rotated grids.

For footprints supplied by a caller, load DEM tiles using explicit WGS84 bounds:

tiles = dem.preload_dem_tiles(west=13.3, east=13.31, south=52.5, north=52.51)

Use west > east for bounds crossing the antimeridian. Product metadata loading and geocoding are the caller's responsibility.

altira download "D:/Images/scene.tif" --cache-dir "D:/DEM"
altira download "D:/Images/scene.tif" --cache-dir "D:/DEM" --source alos-ellipsoidal

Select the online dataset explicitly with source:

dem = Dem(cache_dir="D:/DEM", source="alos-ellipsoidal", resolution=30)
tiles = dem.preload_image("D:/Images/scene.tif")

ALOS and SRTM use OpenTopography's anonymous bulk storage, without an account or API key. Published VRT indexes determine coverage; only selected native tiles are downloaded. Ellipsoidal variants skip geoid conversion. These are OpenTopography's published mirrors, which may differ in version from the original producers' current releases. ALOS is a surface model, so elevations can include vegetation and buildings. See the source dataset pages for attribution: ALOS, ALOS Ellipsoidal, SRTM, and SRTM Ellipsoidal.

Local directories take priority in their supplied order, with the finest resolution preferred within each directory. Supported formats include SRTM HGT, CGIAR SRTM, DTED, TanDEM-X, edited TanDEM-X, Copernicus, and ZIP/TAR deliveries. Coverage gaps are logged, without switching source or resolution. Interpolation fills unavailable or nodata posts with avg_height=0; use avg_height=float("nan") to retain gaps. Raster and network failures raise.

See the usage guide for cache, datum, and statistics details.

Documentation

The documentation site includes the usage guide and API reference. The documentation source is in docs/.

Development

Install the development dependencies:

> pip install -e .[dev]

Run the core test suite directly with pytest:

> pytest

Run linting and formatting with Ruff:

> python -m ruff check --fix .
> python -m ruff format .

To include coverage:

> pytest --cov --cov-report=term-missing