RF Propagation API

HTTP API for radio visibility, terrain profiles, Fresnel/LOS checks, link budget, viewshed, and coverage calculations.

This repository follows SPEC.md. The first implementation pass creates the full service skeleton, pure RF/math kernels, and Copernicus DEM sampling. Integrations that require PostGIS data, pycraf, ITM/P.1812, landcover/canopy rasters, or external viewshed binaries are exposed through stable interfaces and return explicit "not implemented" responses until the corresponding data pipeline is connected.

Layout

api/ - FastAPI service, core calculations, service orchestration, tests.
scripts/ - data bootstrap/import entry points.
data/ - mounted data volume for DEM, landcover, and canopy rasters.
docker-compose.yml - local stack with API, worker, Redis, PostGIS, optional tiler.
API.md - current HTTP endpoints, examples, and implementation status.

Quick Start

cp .env.example .env
docker compose up --build api redis postgis

The API is served at http://localhost:${API_PORT:-5603}, with OpenAPI docs at /docs.

DEM Bootstrap

Download Copernicus DEM GLO-30 COG tiles for Saint Petersburg and Leningrad Oblast:

python scripts/bootstrap_dem.py --bbox 27.3,58.4,35.8,61.4 --output-dir data/dem
docker compose restart api worker

The API samples all .tif/.tiff files under DEM_PATH recursively. In Docker, the default DEM_PATH=/data/dem points to the mounted ./data/dem directory.

Buildings Bootstrap

Download an OSM PBF extract and import building polygons into PostGIS:

mkdir -p data/osm
wget -O data/osm/northwestern-fed-district-latest.osm.pbf \
  https://download.geofabrik.de/russia/northwestern-fed-district-latest.osm.pbf
sh scripts/load_buildings.sh data/osm/northwestern-fed-district-latest.osm.pbf
docker compose restart api worker

The script builds a local radio-osm2pgsql:latest image from scripts/Dockerfile.osm2pgsql on first run. It imports the PBF through standard osm2pgsql tables, then normalizes planet_osm_polygon into the API buildings table with db/sql/normalize_buildings.sql.

The import writes a buildings table with geom, height_m, levels, building_type, and source. Heights come from height, then building:levels * 3.0, then an estimated default by building type.

Check the API:

curl -s -X POST http://localhost:5603/api/v1/buildings/query \
  -H 'Content-Type: application/json' \
  -d '{"bbox":[30.30,59.93,30.33,59.95]}' | jq '.features | length'

Check data availability and run a minimal integration smoke test:

curl -s http://localhost:5603/api/v1/status/data | jq
API_BASE=http://localhost:5603 sh scripts/smoke_api.sh

Landcover And Canopy

Place ESA WorldCover GeoTIFF/COG files under data/landcover and optional canopy height GeoTIFF/COG files under data/canopy, then restart the API:

python scripts/bootstrap_landcover.py --bbox 27.3,58.4,35.8,61.4 --landcover-dir data/landcover
python scripts/bootstrap_canopy.py --bbox 27.3,58.4,35.8,61.4 --output-dir data/canopy
docker compose restart api worker

Or download manually into data/landcover and data/canopy. If the canopy tile index is unavailable, pass a newline-separated list of COG URLs with scripts/bootstrap_canopy.py --urls-file urls.txt --output-dir data/canopy.

The /api/v1/landcover/path endpoint samples all .tif/.tiff files under LANDCOVER_PATH recursively. Canopy data is optional; when it is missing, canopy_height_m is returned as null.

When WorldCover is available, /api/v1/link/budget and /api/v1/terrain/los can include P.833 vegetation attenuation. Tune the coefficients with P833_GAMMA_DB_PER_M and P833_MAX_ATTENUATION_DB in .env.

For local Python development:

cd api
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
python -m pytest
python -m ruff check .

3.9 KiB Raw Blame History