❄️ How Our Daily Snow Analysis Works
Every morning we combine official datasets, field observations and local notes into one clear snowpack snapshot for
Arnøya,
Vannøya and
Kågen.
The system refreshes automatically and produces both a readable overview and machine‑readable files.
1) Data Sources
🛰️ seNorge (Operational)
Daily 1×1 km gridded data from MET Norway’s operational archive.
For each island, we extract:
- RR — 24‑hour precipitation
- TG — daily mean temperature
- SD — snow depth (when available)
- HN24 / HN72 — estimated new snow (24 h / 72 h)
- SD shock — change in snow depth compared to yesterday
🌦️ MET Locationforecast 2.0
Official MET Norway forecast service.
We sample the rolling 24‑hour window ending
06:00 local time, collecting:
- Temperature
- Precipitation
- Wind (min / max / gust)
Used both as a fallback and for wind‑based slab detection.
🏔️ Regobs (NVE)
Recent public and professional snow observations within ~60 km of our islands:
- Whumpfs
- Shooting cracks
- Wet/dry snow comments
- Persistent weak‑layer hints
Regobs helps validate what models suggest.
📓 Local Notes
We ingest local field notes from our own website (via JSON:API).
These notes are scanned for:
- Weak‑layer keywords (facets, depth hoar, surface hoar, etc.)
- Qualitative snowfall descriptions (converted to rough HN estimates)
2) Elevation‑Band Reconstruction (0–1000 m)
For each island, we analyze elevation bands:
0–200 m · 200–400 m · 400–600 m · 600–800 m · 800–1000 m
Each band receives:
- Band temperature (with inversion guard on calm/cold/dry winter days)
- HN24 estimate based on temperature and precipitation
- Precipitation type (snow / mixed / rain)
- ROS flag (rain falling onto snow)
- Slab flag:
- From wind loading (wind ≥ threshold + snow/mixed precip), or
- From Regobs danger signs (cracking / whumpf)
- SD shock (same per band for visibility)
3) Deep‑Instability Score
We compute a simple daily score to highlight potential for deep persistent slab issues.
The score considers:
- Large HN24 or HN72 loading
- Strong temperature gradients between bands
- SD shock (rapid loading)
- Rain‑on‑snow while bands stay below 0 °C
- Regobs danger signs
- Weak‑layer keywords in local notes
If enough signals combine →
Deep Instability: YES.
4) Daily Outputs
Our system generates several files each morning:
Human‑readable
- daily_summary.md
Overview of freezing level, wind, instability, plus detailed band tables.
Machine‑readable
- daily_summary.json
- avi_inputs_latest.csv (per‑band values)
- drupal_ingest.json (stable feed for the website)
- provenance.json (which sources contributed that day)
5) Transparency & Limitations
- All outputs are refreshed automatically every morning.
- Every number is traceable back to seNorge, MET, Regobs or field notes.
- SD shock is included only when seNorge provides snow‑depth data for that day.
- This pipeline provides decision support, not an avalanche forecast.
Always compare with official avalanche warnings and your own observations in the field.
