Forecast Models

Forecast Models 🧪

In addition to observed data, SPCBot monitors several experimental and operational forecast models to provide a look-ahead at severe weather threats.

The Colorado State University Machine Learning Probabilities model provides daily severe weather forecasts for Days 1–8.

Automated Posting: The bot polls for new CSU-MLP runs daily and automatically posts the consolidated 6-panel summaries.
On-Demand (/csu): Users can use the /csu command to retrieve specific products (Individual Hazards, Significant Severe, etc.) via an interactive dropdown.
Source: Pulls from the official CSU MLP image archive.

WxNext2 is an AI-driven convective hazard forecast developed by NCAR.

Automated Posting: Posted daily for Days 1–8.
On-Demand (/wxnext): Retrieves the latest AI hazard maps, which provide an alternative to traditional SPC categorical outlooks.

The Supercell Composite Parameter (SCP) maps are generated twice daily based on CFSv2 and GEFS data.

Frequency: Automated posts occur shortly after the 00z and 12z model cycles.
On-Demand (/scp): Retrieves the latest NIU SCP graphics.
Context: These maps are particularly useful for long-range (1–2 week) severe weather pattern recognition.

The Weather Prediction Center (WPC) Excessive Rainfall Outlooks (ERO) for Days 1–3.

On-Demand (/wpc): Retrieves the latest flash flood probability graphics.
Automated Monitoring: The bot tracks ERO updates and can be configured to post High-Risk ERO events automatically.

The Weather Prediction Center (WPC) Surface Fronts analysis map shows the latest synoptic surface analysis.

On-Demand (/fronts): Retrieves the most recent 3-hourly analysis cycle (00/03/06/09/12/15/18/21Z) from WPC.
Auto-Discovery: The bot dynamically detects the latest analysis by checking the Last-Modified header of available cycle images.
Automated Posting: Posts automatically to the Weather Chat channel every 15 minutes when the image changes (SHA-256 dedup).