solar/docs/AVAILABLE_DATA.md

Data available to store for comparisons (weekdays, weekends, day/night, holidays)

What we already store (daily, per location)

Data	Source	Use for comparison
Date	—	Weekday vs weekend, holiday (join with calendar)
Temperature min/max	Open-Meteo Forecast	Season, weather type
Solar (shortwave sum), sunshine duration	Open-Meteo	Production, clear vs cloudy
Sunrise, sunset, daylight	Open-Meteo	Day length, season
Wind speed/gusts/direction	Open-Meteo	ML/RL, weather type
Precipitation, rain, rain probability	Open-Meteo	Rainy vs dry days
UV index max	Open-Meteo	Clear-sky proxy

---

Also stored (AQ + weather code; hourly in hourly_metrics)

Data	Source	Use for comparison
Air quality (PM2.5, PM10, European AQI)	Open-Meteo Air Quality API	Weekday vs weekend, traffic/industry, health
Weather code (WMO)	Open-Meteo Forecast	Day type: clear (0) vs cloudy (1–3) vs rain (61–82) vs thunder (95–99)

---

Other data available from Open-Meteo (not yet stored)

From Forecast API (daily)

• Cloud cover (total, low/mid/high) – daily max/mean; good for “clear vs overcast”.
• Relative humidity – daily min/max; comfort and fog.
• Surface pressure – daily min/max; weather systems.
• Apparent temperature – daily min/max; “feels like”.
• Precipitation hours – number of hours with rain; intensity of rainy days.
• Snowfall sum – for winter comparisons.

From Forecast API (hourly – for day vs night)

• Same variables at hourly resolution. To compare day vs night you need time-of-day:
  • Option A: We already store hourly_metrics; aggregate by hour-of-day (e.g. 6–18 = day, 18–6 = night) when analysing.
  • Option B: Derive “day” vs “night” from sunrise/sunset and store e.g. solar_day_sum and solar_night_sum (night is 0 for solar) or average temp day vs night from hourly in the ETL.

From Air Quality API (hourly)

• Ozone (O₃), Nitrogen dioxide (NO₂), Sulphur dioxide (SO₂), Carbon monoxide (CO) – for traffic/industry vs background.
• Dust (e.g. Saharan) – episodes.
• Pollen (grass, tree, weed) – seasonal and health.

From other sources (not Open-Meteo)

• Holidays – Belgian (or regional) holiday calendar: add a small table or CSV (date, region, name) and join by date (and optionally location) so you can compare holiday vs non-holiday (e.g. traffic, air quality, demand).
• Day of week – derived from date in SQL or in the app (no extra storage).

---

Suggested next steps for comparisons

1. Weekday vs weekend – use stored date; derive day_of_week (e.g. in SQL: EXTRACT(DOW FROM date)) or store a small calendar table with date, is_weekend, is_holiday.
2. Day vs night – add hourly storage for key series (solar, temp, PM2.5) and aggregate by hour-of-day; or add pre-aggregated “day” and “night” columns in a later iteration.
3. Holidays – add a holidays table (e.g. Belgian official holidays) and join on date (and optionally location_id if you have regional holidays).
4. Weather type – use weather_code (and optionally cloud cover) to group days into clear / cloudy / rainy / stormy for solar and demand comparisons.