← Back to Forecast

About This Forecast

Chapel Hill Composite Weather Forecasting System

How It Works

This forecast system blends multiple weather models to provide accurate precipitation type diagnosis for Chapel Hill, NC. It's designed specifically for marginal winter events where precipitation type is uncertain (snow vs. sleet vs. freezing rain).

Model Blending

We use time-dependent weighting to combine forecasts from ECMWF and NOAA HRRR:

  • 0-6 hours: HRRR (70%) + ECMWF (30%) — HRRR excels at short-range
  • 6-24 hours: HRRR (60%) + ECMWF (40%)
  • 24-48 hours: ECMWF (60%) + HRRR (40%)
  • 48-72 hours: ECMWF (80%) + HRRR (20%) — ECMWF excels at long-range

Precipitation Type Diagnosis

The system analyzes temperature profiles at three levels:

  • Surface (2m): Determines if precipitation reaches ground frozen
  • 850 hPa (~1.5 km): Detects warm/cold layers aloft
  • 700 hPa (~3 km): Identifies deep cold columns

The algorithm applies rules based on these temperature profiles:

  • Rain: Surface temp above 2°C
  • Snow: Deep cold column (all levels below freezing)
  • Sleet: Warm nose aloft + surface below freezing
  • Freezing Rain: Warm nose aloft + surface near freezing

Cold-Air Damming (CAD) Detection

The system detects CAD events — a Carolina/Virginia specialty where cold air gets trapped against the Appalachian Mountains. CAD increases the likelihood of snow and sleet over freezing rain in marginal setups.

CAD signature: Cold surface (≤2°C), cold 850 hPa (≤-2°C), and NE surface winds during Nov-Mar.

Data Sources

ECMWF AIFS/IFS

The European Centre for Medium-Range Weather Forecasts provides global forecast data via their Open Data program. We use both surface variables and upper-air temperature profiles.

Accessed via: Open-Meteo.com API

Contains modified Copernicus ECMWF data. ECMWF data is licensed under the Copernicus License.

NOAA HRRR

The High-Resolution Rapid Refresh model provides hourly 3-km resolution forecasts for the continental US. We use it for short-range accuracy and categorical precipitation type flags.

Accessed via: AWS S3 (noaa-hrrr-bdp-pds)

NOAA data is in the public domain.

Open-Meteo.com

Open-Meteo provides a simplified API for accessing ECMWF forecasts, including pressure-level temperature data on their free tier.

Open-Meteo is a free weather API licensed under CC BY 4.0. Attribution is required.

⚠️ Important Disclaimer

This is an experimental forecast system for educational and research purposes.

For official weather forecasts, warnings, and watches, always refer to:

Technical Details

Location: Chapel Hill, NC (35.9132°N, 79.0558°W)
Forecast Range: 72 hours (3 days)
Temporal Resolution: Hourly
Update Frequency: 6-hourly at 00, 06, 12, 18 UTC (+15 minutes)
Backend: Python 3.11+ with FastAPI, SQLite caching
Frontend: Next.js 15 with Tailwind CSS
Deployment: Railway/Render (backend), Vercel (frontend)