Purpose
This guide mirrors how professional forecasters think and work, but scaled for
a single analyst using HRRR, Herbie, and Python.
It is designed as a reference — not something you must learn all at once.
Core Philosophy
- Do not rely on single fields
- Focus on processes, not colors
- Compare time evolution
- Always ask “why did this happen?”
Essential Data Stack
| Component | Tool | Purpose |
|---|
| Model Data | HRRR | High-resolution convection & boundary layer |
| Fetching | Herbie | Automated GRIB access |
| Analysis | xarray / MetPy | Diagnostics & calculations |
| Mapping | Cartopy | Geospatial plotting |
| Visualization | Matplotlib | Static maps & animations |
Temperature Analysis (Beyond 2m)
Recommended Fields
| Field | Why It Matters |
|---|
| 850 mb Temperature | Air mass & potential highs |
| 925 mb Temperature | Cold pools, nighttime lows |
| Thickness (1000–500) | Large-scale thermal structure |
Forecaster Tip:
High temperatures are estimated by combining 850 mb temperature with expected
mixing depth — not by trusting surface output.
Convective Forecast Toolkit
| Category | Fields |
|---|
| Instability | SBCAPE, MLCAPE, CIN |
| Moisture | PWAT, Td @ 850 mb |
| Shear | 0–6 km Bulk Shear |
| Storm Evolution | Composite Reflectivity |
Radar Reflectivity Animation Workflow
- Loop HRRR forecast hours (0–24)
- Fetch composite reflectivity
- Mask weak returns
- Save consistent frames
- Compile into animation
REFC > 15 dBZ is typically sufficient to remove noise
Multi-Panel Thinking
Forecasters rarely analyze a single map.
Effective combinations include:
- 850 mb Temperature + Winds
- CAPE + Shear
- Reflectivity + Updraft Helicity
- 925 mb Temperature + MSLP
Automation = Your Forecast Team
A single script should:
- Fetch model data
- Generate core maps
- Create animations
- Save time-stamped outputs
Verification & Learning
Verification is where real forecasting skill develops:
- Modeled reflectivity vs real radar
- Forecast highs vs observations
- Instability vs SPC mesoanalysis
How to Use This Guide
- Start with one section at a time
- Add tools gradually
- Revisit after real weather events
- Update it as your workflow evolves