Downloads · 30 days
0
aerostratospheric/uogw-tornado-trends
uogw-tornado-trends is a tabular classification model from aerostratospheric. Use it for the tabular classification task on the model card, and read the license before you ship it in a product. It is set up for sklearn. The card lists the license as cc-by-4.0.
Publisher: Aerostratospheric Library: scikit-learn (.joblib heads) Sibling data: aerostratospheric/uogw · spc-tornado-history
Downloads · 30 days
0
Access
Public
Updated Oct 23, 2026
Repo size
37 MB
Likes
0
Public
Click a slice to open those files.
.joblib37 MB · 100%
From the Hugging Face model README
Publisher: Aerostratospheric
Library: scikit-learn (.joblib heads)
Sibling data: aerostratospheric/uogw · spc-tornado-history
Research package for tornado climatology and Midwest trend screening (SPC Severe Weather Database 1950–2024), with UOGW Clark County desk products as context — not a warning service.
| Is | Is not |
|---|---|
| Open climatology / trend research heads | A watch, warning, or official outlook |
| SPC path + count feature models | Real-time radar nowcasting |
| Education and open-science screening | Sole life-safety decision support |
Trained 2026-09-18T15:34:31Z (~71.8k SPC rows). Hub re-touch 2026-09-23 — weights not re-trained. Intended retrain ~12:30 UTC.
| Head | File | What it does |
|---|---|---|
state_month_count | models/state_month_count.joblib | Expected tornado counts by U.S. state × month × year |
ef_severity | models/ef_severity.joblib | Path geometry proxies → EF/mag bin (0–5) |
midwest_annual_trend | models/midwest_annual_trend.joblib | Linear long-term Midwest annual count trend |
State id map: state_id_map.json / config.json. Holdout metrics: metrics.json.
Note: A Clark County pre-tornado research fit appears in metrics history with
n_days: 0for this build — noclark_pre_tornado_research.joblibis shipped until history is available. Do not invent that head.
import json
from pathlib import Path
import joblib
from huggingface_hub import snapshot_download
root = Path(snapshot_download("aerostratospheric/uogw-tornado-trends"))
cfg = json.loads((root / "config.json").read_text())
heads = {h: joblib.load(root / "models" / f"{h}.joblib") for h in cfg["heads"]}
state_ids = cfg.get("state_id_map") or json.loads((root / "state_id_map.json").read_text())
# Example features for state_month_count: st_id, mo, sin_m, cos_m, yr (see metrics.json)
Requires scikit-learn, joblib, huggingface_hub.
pre-tornado-clark desk products when present in the archiveSee FRESHNESS.md.
| Artifact | URL |
|---|---|
| SPC companion dataset | https://huggingface.co/datasets/aerostratospheric/spc-tornado-history |
| UOGW scientific suite | https://huggingface.co/aerostratospheric/uogw-scientific-suite |
| Weather forecast suite | https://huggingface.co/aerostratospheric/uogw-weather-forecast |
| Open-models index | https://huggingface.co/aerostratospheric/msds-open-models |
| Site | https://midwestsds.com |
| Intro | https://calendly.com/aerostratospheric/15min |
Aerostratospheric / Midwest Stratospheric Data Systems (2026).
UOGW Tornado Trends Suite. Hugging Face.
https://huggingface.co/aerostratospheric/uogw-tornado-trends
Always cite NOAA/SPC as the system of record for storm reports. License: CC BY 4.0 for this packaging.
This tornado trend packaging is part of the open UOGW commons from Aerostratospheric, an Illinois nonprofit and home of Midwest Stratospheric Data Systems. Severe weather researchers, storm spotter trainers, and classrooms who want to pair their own reports or methods with these layers are welcome to reach out about data collaboration.
Live context sits on xDataHub, our public open atmospheric data dashboard (https://www.midwestsds.com/msds-data-hub.html), including NOAA KILX soundings next to our own flight packages. API docs: https://www.midwestsds.com/data/v1/api.html
Contact [email protected] or book a 15 minute intro at https://calendly.com/aerostratospheric/15min. Collaboration only; we are not asking for donated gear or services.