Severe

Severe environment, hour by hour

Rates the environment only: whether storms that form could be severe, not whether storms will form. Top row uses blended indices; each model row uses that model's own sounding. Click an hour (or use ← →) to load its hodograph below.

Fetch models to load the signal.

Low Marginal Elevated High Hatched: limited data No data

Experimental severe probabilitiesHISTORICAL REPLAY

Explore the research model’s tornado, wind and hail maps from 12 archived 2024 cases.

Archive only. No live forecast. Research v0.2 has not established dependable forecast skill. These maps describe the selected historical window.

Archived case

Hazard

Display

Map theme

Probability opacity 85%

Choose a historical case

Saved location

—

Tornado probability

A qualifying report within 25 miles of the sampled grid center during this four-hour window. This is a neighborhood probability, not a direct-hit probability.

Dashed circle: the area assessed around the selected sample.

Chance of a qualifying tornado report within 25 miles of a sampled grid center · four-hour window.

Filled contours interpolate between archived samples about 15 km apart. Point readouts always use an original sample.

Colors show experimental probabilities, not official SPC outlook categories. No EF2+, significant wind or significant hail hatching is available from this model.

Open the viewer to load archived research results.

How well did it perform?

36 calendar-selected cases: 2022 for training, 2023 for calibration and 12 dates in 2024 for development evaluation. Lower Brier scores are better. Nearby grid samples are correlated.

Every paired uncertainty interval includes zero. This small sample does not establish a dependable advantage over the baselines. Only two evaluation days had tornado reports. 2025 remains untouched for a later evaluation.

Targets: tornado reports, wind reports of at least 50 kt and hail reports of at least 1 inch. Report absence is an imperfect non-event proxy. This is separate from official outlooks and from the live environment guidance above.

Method, provenance and development results · Nadocast reference viewer

Hodograph & sounding summary

Profile

Selected hour

0–1 km 1–3 km 3–6 km 6–9 km RM / LM: Bunkers right- and left-mover motion · ■ 0–6 km mean wind · rings every 10 kt

Day by day

Blend value first, with the highest single model in brackets, so one model's worst case stays visible. STP and SCP are ≈ fixed-layer estimates.

SCP

How signals are set

HIGH STP ≥ 1 or SCP ≥ 4.

ELEVATED SCP ≥ 1, or CAPE ≥ 1000 J/kg with 0–6 km shear ≥ 40 kt.

MARGINAL CAPE ≥ 500 J/kg with 0–6 km shear ≥ 25 kt.

LTD Limited data. Shown on top of any signal level when shear, SRH or LCL can't be computed for that hour, so the signal could be understated. Hover for what's missing.

LOW None of the above, or CAPE under 100 J/kg.

These are Heads-Up status colors used for signal strength in model guidance, not an instruction to act or an official outlook.

Ingredient flags

HAIL CAPE ≥ 1000, 700–500 hPa lapse rate ≥ 7 °C/km, 0–6 km shear ≥ 30 kt.

WIND CAPE ≥ 500, with a 0–3 km lapse rate ≥ 8 °C/km or a 0–6 km mean wind ≥ 35 kt.

TOR CAPE ≥ 250, 0–1 km SRH ≥ 100, LCL ≤ 1250 m, 0–6 km shear ≥ 30 kt.

A flag needs every one of its inputs. Missing data never raises or clears a flag.

How these are computed

Open-Meteo gives each model's CAPE, CIN and lifted index, plus winds, temperatures and heights on pressure levels. Shear, helicity, Bunkers motion, lapse rates and LCL are calculated here from those levels, about 50–150 hPa apart, and the 10 m wind. STP and SCP are the fixed-layer versions: STP uses 0–1 km SRH and 0–6 km shear, and SCP uses 0–3 km SRH and 0–6 km shear in place of the effective-layer values. Expect a rougher picture than SPC mesoanalysis or a full-resolution sounding. Which kind of CAPE each model reports varies.