MET Cloud Layers for Home Assistant

Layered cloud-cover forecasts — low / mid / high clouds and fog — from MET Norway's free Locationforecast API, plus derived sky transmittance and PV-weighted sky quality sensors aimed at solar/battery automation.

Home Assistant's built-in Met.no integration only exposes total cloud cover. For solar work that's not enough: 80% cover can mean thin cirrus (production barely dented) or solid stratus (production gone). This integration exposes the layers and turns them into actionable numbers.

Not affiliated with or endorsed by MET Norway, Yr or NRK.

Sensors

Sensor	Meaning
Cloud cover	Total cloud fraction now (%) — carries the hourly forecast series as a hourly attribute
Low / Mid / High clouds	Per-layer cloud fraction now (%)
Fog	Fog fraction now (%)
Sky transmittance	Estimated fraction of clear-sky irradiance reaching the ground now (%)
Low clouds next 3 h	Time-weighted average — a "will this dip persist?" signal
Sky transmittance next 3 h	Same window, transmittance
Sky quality today remaining	Solar-elevation-weighted transmittance from now to midnight (%)
Sky quality tomorrow	Same for tomorrow's daylight hours (%)
Forecast updated	Timestamp of the MET model run (diagnostic)

The hourly attribute on Cloud cover holds ~48–54 h of {datetime, total, low, medium, high, fog, transmittance} rows for use in Node-RED, templates or ApexCharts.

The transmittance model

A deliberately simple, tunable heuristic:

attenuation   = w_low·low + w_medium·medium + w_high·high + w_fog·fog   (fractions)
transmittance = 1 − min(1, attenuation)

Default weights: low 0.75, medium 0.40, high 0.15, fog 0.85 — all adjustable in the integration options. The sky quality aggregates additionally weight each forecast hour by sin(solar elevation), so a cloudy noon costs the score far more than a cloudy 05:00.

These are estimates, not irradiance physics. Calibrate the weights against your own PV data if you need better absolute numbers; the relative signal (good day / bad day / clearing up / closing in) is the point.

Installation

HACS (custom repository): HACS → Integrations → ⋮ → Custom repositories → add this repo URL, category Integration → install → restart Home Assistant.

Manual: copy custom_components/metclouds/ into your HA config/custom_components/ and restart.

Then Settings → Devices & services → Add integration → MET Cloud Layers. Only a location is needed — no API key.

MET terms-of-service compliance

This integration follows the MET API TOS:

• Sends an identifying User-Agent (ha-metclouds/<version> <repo URL>).
• Truncates coordinates to 4 decimals.
• Uses If-Modified-Since conditional requests; unchanged data is answered with a cheap 304 and reused locally.
• Default poll interval 30 min (MET's data file refreshes about that often).
• Data license: CC BY 4.0 — the sensors carry the required attribution.

Standalone smoke test

python3 scripts/smoke_test.py 59.5450 10.8571

Fetches a live forecast and prints the current layers plus the derived metrics — no Home Assistant required (needs aiohttp).

License

MIT for the code. The weather data itself is © MET Norway, CC BY 4.0.