Field Guide

Surface-Based CAPE: the buoyant energy available to a parcel lifted from the surface, integrated from its level of free convection to equilibrium.

How it's derived Taken from the model’s own surface-based CAPE output where shipped, otherwise integrated from the model sounding with the virtual-temperature correction.

g gravity; z_LFC level of free convection, z_EL equilibrium level; T_v virtual temperature of the surface parcel vs environment.

Reading it Lift a parcel from the surface and this is the buoyancy it gets. 500 to 1500 J/kg supports storms with adequate forcing, and above about 2500 J/kg updrafts can carry large hail and severe wind given shear. Surface-based values overstate the threat when the boundary layer is capped or storms are elevated, so cross-check MLCAPE and CIN.

Caveat A surface parcel does not account for layers that are not well mixed and grossly underestimates buoyancy when ascent is elevated above a stable layer, and it overestimates realized instability when the moist layer is shallow.

Mixed-Layer CAPE: buoyant energy for a parcel with the mean temperature and moisture of roughly the lowest 100 mb, which is closer to what a real storm draws in than a surface parcel.

How it's derived Computed from the lowest-100-mb mean parcel against the model sounding with the virtual-temperature correction, the same construction SPC mesoanalysis uses.

Same integral as CAPE, but the lifted parcel uses the mean temperature and moisture of the lowest 100 mb above ground.

Reading it This is the instability term most supercell and tornado parameters expect, and it represents inflow buoyancy better than a surface parcel does. STP uses it. Less prone than SBCAPE to overstating instability from a shallow moist layer.

Caveat Averaging over the lowest 100 mb underestimates elevated or surface-based buoyancy when the layer is not well mixed, and there is heavy overlap in MLCAPE between ordinary and severe thunderstorms.

Most-Unstable CAPE: buoyant energy for the most buoyant parcel in the lowest 300 mb.

How it's derived The parcel with the highest theta-e in the lowest 300 mb, lifted against the model sounding.

The lifted parcel is the highest-theta-e parcel found in the lowest 300 mb above ground; the integral is the standard virtual-temperature CAPE.

Reading it Use this when convection is elevated, rooted above a stable boundary layer, as with nocturnal MCSs and cold-side hail. When MUCAPE far exceeds SBCAPE, expect elevated storm modes.

Caveat Tall thin CAPE in a high equilibrium-level sounding is more susceptible to water loading than short fat CAPE, so equal MUCAPE values do not imply equal updraft depth or strength.

Surface-Based Convective Inhibition: the negative buoyancy a surface parcel must overcome before it can convect freely.

Negative-area integral from the surface to the LFC for the surface parcel; the result is negative.

Reading it This is how strong the cap is. Values of -25 to -75 J/kg can be overcome by strong heating or forcing, while beyond -100 J/kg initiation from the boundary layer becomes unlikely without a strong trigger. Zero CIN with large CAPE means storms can fire readily.

Caveat CIN is sensitive to the parcel chosen and to small changes in surface dewpoint or boundary-layer mixing ratio, and surface-based CIN can be very misleading when the instability is elevated, so there is no exact threshold below which convection is guaranteed.

Downdraft CAPE: the energy available to a saturated downdraft descending from mid-levels, which gauges how hard rain-cooled air can hit the surface.

How it's derived Integrated for a parcel descending moist-adiabatically from the lowest-theta-e level in the 700 to 500 mb layer.

The parcel starts at the minimum-theta-e level z_i within 700 to 500 mb and descends moist-adiabatically to the surface; the integrand is the environment-minus-parcel buoyancy deficit.

Reading it Above about 1000 J/kg supports strong downbursts and damaging outflow winds, so weight it heavily when the threat is severe wind. The outlook tool requires an 80-km average DCAPE of at least 700 J/kg as its wind-hazard environment check.

Mixed-Layer CIN: convective inhibition for the lowest-100-mb mean parcel.

Negative-area integral from the surface to the LFC for the lowest-100-mb mean parcel.

Reading it For surface-based storm potential this represents the cap better than SBCIN does. It reads slightly stronger, meaning more negative, than SBCIN on hot afternoons with a superadiabatic surface layer.

0–3 km CAPE: the portion of buoyancy in the lowest 3 km.

The CAPE integral truncated at 3 km above ground (or the equilibrium level, whichever is lower).

Reading it Low-level buoyancy accelerates stretching of near-ground rotation, which is why it matters for QLCS and landspout tornadoes. SPC contours it from 25 J/kg, and 100 J/kg or more with strong low-level shear is a meaningful tornado signal even when total CAPE is modest.

0–3 km lapse rate: how quickly temperature falls with height in the lowest 3 km.

Temperature drop from the surface to 3 km above ground divided by the depth, in degC per km.

Reading it Steep low-level lapse rates, 7 degC per km or more, promote strong low-level stretching and gusty outflow. They also deepen mixing, which can erode low-level moisture. The dry-adiabatic limit is about 9.8 degC per km.

700–500 mb lapse rate: how steep the mid-level temperature profile is, which is what flags elevated-mixed-layer air.

How it's derived Computed from the 700 and 500 mb temperatures and geopotential heights.

Temperature difference between the 700 and 500 mb levels divided by their geopotential-height separation, in degC per km.

Reading it Values of 7 degC per km or more mark a plume of steep mid-level lapse rates, often elevated-mixed-layer air off the Rockies or the Mexican Plateau, and they build large hail and high CAPE. SPC treats roughly 5.5 to 6.0 as the moist-adiabatic floor and 9.8 as the dry-adiabatic ceiling. SHIP uses it directly.

Caveat Lapse rates alone do not determine parcel buoyancy, since the key to convective growth is the presence of CAPE, and steep mid-level lapse rates can mark very dry air aloft that actually inhibits deep moist convection.

Lifted Condensation Level height: the altitude where a lifted surface parcel saturates, which is roughly where cloud base sits.

Espy approximation: LCL height in meters above ground from the surface temperature minus dewpoint spread in degC; the exact value is the height where the dry adiabat through T meets the mixing-ratio line through T_d.

Reading it Lower is better for tornadoes. Cloud bases under about 1000 m keep evaporative cooling of outflow weak, preserving the buoyant, moist inflow tornadoes need. Above about 1500 m, cold-pool-dominant storm modes and gusty straight-line winds become more likely than true tornadoes. The STP LCL term goes to zero at 2000 m.

Caveat LCL height varies sharply over small time and space scales and depends on the parcel origination level the same way CAPE and CIN do, so a single surface or model value is a crude estimate of actual cloud base.

Theta-e (equivalent potential temperature) at 2 m: the temperature a parcel would have if all its moisture condensed and it were brought to a standard pressure, so it folds heat and moisture into one number.

theta is potential temperature with reference pressure p0 = 1000 mb; L_v latent heat of vaporization, r mixing ratio, c_p specific heat, T_LCL temperature at the LCL.

Reading it High low-level theta-e often marks the warm, moist inflow available to storms. Watch for a theta-e axis advecting into a boundary or convergence zone, especially where lift can erode CIN. Sharp theta-e gradients may help reveal fronts, drylines, and outflow boundaries, but assess them alongside convergence, shear, lapse rates, and storm-relative inflow when anticipating initiation or supercell maintenance.

Caveat Potential instability shown by theta-e decreasing with height does not by itself produce convection, since the layer still has to be lifted by a front or other forcing for storms to result.

Boundary-layer height: the depth of the turbulently mixed near-surface layer.

Reading it Deep boundary layers, over 2 km, mean strong mixing, gusty surface winds, and deep dry layers. Shallow stable layers trap pollutants and fog. Taken directly from the model PBL-height diagnostic, so the exact definition varies by model.

K-Index: a legacy thunderstorm-potential index built from 850, 700, and 500 mb temperature and moisture.

All terms in degC: 850-500 temperature lapse, plus 850 mb dewpoint, minus the 700 mb dewpoint depression.

Reading it A legacy thunderstorm-coverage flag built from 850, 700, and 500 mb temperature and moisture. Values above about 35 degC usually mean the column is moist enough for numerous storms, but the index is blind to shear, capping, and storm organization, so use it as a broad coverage cue rather than a severe parameter.

Total Totals Index: a 1970s-era index combining 850 mb warmth and moisture against 500 mb cold.

Total Totals in degC: the sum of the vertical totals (T850 minus T500) and cross totals (Td850 minus T500).

Reading it A 1970s-era index combining 850 mb warmth and moisture against 500 mb cold. Treat 50 degC or more as scattered strong storms by the old criteria, then go look at CAPE and shear. Kept for continuity with older guidance.

Lifted Index: the 500 mb environment temperature minus a lifted surface parcel’s temperature.

Environment temperature at 500 mb minus the temperature of a surface parcel lifted dry- then moist-adiabatically to 500 mb, in degC.

Reading it Negative means unstable: -2 to -5 degC is moderate, beyond -6 strong. It carries the same blind spot as SBCAPE, a surface parcel that knows nothing about the cap. SPC compares the lifted parcel to the environment at 500 mb by default.

Caveat Because it uses temperature at only a few mandatory levels rather than the full integrated buoyancy, the Lifted Index is a coarser measure of instability than CAPE and can miss inhibition that lies between those levels.

Showalter Index: the Lifted Index computed from the 850 mb parcel instead of the surface, which makes it insensitive to shallow surface layers.

Environment temperature at 500 mb minus the temperature of an 850 mb parcel lifted to 500 mb, in degC.

Reading it The Lifted Index computed from the 850 mb parcel instead of the surface, which makes it insensitive to shallow surface layers. A negative Showalter with positive surface indices points to elevated storms.

Freezing level: the lowest height (meters above sea level) at which the temperature crosses 0 °C.

How it's derived Interpolated through the model’s mandatory pressure levels, scanning bottom-up. Clamped at zero, so sub-freezing columns read as frozen to the surface rather than a negative height.

The lowest height z_FZ, scanning upward, where the environmental temperature profile crosses 0 degC.

Reading it Low freezing levels favor hail reaching the ground because there is less melting depth, and they matter for mountain snow levels. High freezing levels melt small hail into heavy rain. Interpolated through the model levels bottom-up and clamped at zero, so sub-freezing columns read as frozen to the surface.

1000-500 mb thickness, the depth of the lower half of the atmosphere. Cold air is dense and shallow, warm air deep, so thickness is a clean proxy for the mean temperature of that layer.

How it's derived Geopotential height at 500 mb minus the height at 1000 mb, in decameters.

Reading it The classic rain/snow rule of thumb is the 540 dam line: precipitation tends to fall as snow below about 540 dam and rain above, shifting toward 534 dam in shallow cold air and over high terrain. Read it with snow level and precip type, not alone.

Caveat A bulk-layer average. It misses shallow cold-air damming and warm noses, which is exactly where precip type gets interesting, so treat 540 as a guide rather than a hard line.

Snow level, the altitude where falling snow melts. It follows the wet-bulb 0 degC height rather than the dry freezing level, because evaporative cooling into unsaturated air lets snow survive a few hundred meters lower.

How it's derived The height where the wet-bulb temperature crosses 0 degC, from the temperature and dewpoint profile by Normand rule.

Reading it The elevation that splits rain from snow in terrain. Below it expect rain or a rain/snow mix; above it, accumulating snow. The gap between snow level and freezing level widens in dry air.

Caveat Interpolated through the mandatory levels, so in deep valleys or sharp inversions it is approximate. It marks where snow melts aloft, not whether it survives a warm surface layer to reach the ground as snow.

Aircraft icing potential, where supercooled liquid water is most likely. Icing forms when an aircraft flies through humid air cold enough to hold supercooled droplets.

How it's derived The column maximum of a 0 to 1 score combining relative humidity and temperature, peaking near -7 °C and falling to zero outside -18 to 0 °C or below 70 percent RH.

Reading it A general-aviation planning aid for the altitudes and regions to watch for airframe icing. The highest values mark saturated, moderately supercooled layers.

Caveat A simple humidity-and-temperature proxy, not a calibrated icing algorithm. It does not distinguish droplet size or supercooled large drops, and it works off the mandatory levels rather than a fine vertical grid.

0–1 km bulk shear: the magnitude of the vector wind difference between the surface and 1 km above ground.

How it's derived Vector difference of the model winds at the surface and 1 km, in knots.

u and v are the east-west and north-south wind components at the surface (subscript 0) and 1 km above ground (subscript 1).

Reading it This tells you how much the wind turns and strengthens in the lowest kilometer, which weighs directly on tornado potential. Values of 15 to 20 kt support low-level mesocyclones, and 30 kt or more is strong. Read it together with low cloud bases and 0 to 1 km storm-relative helicity.

0–6 km bulk shear: the deep-layer shear that organizes storms.

How it's derived Vector difference between the surface and 6 km winds.

u and v are the wind components at the surface (subscript 0) and 6 km above ground (subscript 6); the result is a vector difference magnitude, not a wind speed difference.

Reading it This is the deep-layer shear that organizes storms into supercells. About 25 to 30 kt marks the transition from multicells to supercells, and 40 to 60 kt is where supercells become favored. It feeds SCP, SHIP, and the outlook recipes, converted to meters per second internally to match the published formulas.

Caveat Deep-layer 0 to 6 km bulk shear has limited ability to distinguish supercells that produce significant tornadoes from those that do not, which is better separated by the 0 to 1 km layer.

Effective bulk shear: SPC defines this as deep-layer shear measured across the storm's actual inflow layer (effective inflow base to half the storm depth) rather than fixed heights. On this site the field is computed as plain 0 to 6 km bulk shear, not the true effective bulk wind difference.

How it's derived Computed here as plain 0 to 6 km bulk shear, the vector wind difference between the surface and 6 km. It is not the true effective bulk wind difference, which SPC defines from the effective inflow base to 50 percent of the most-unstable equilibrium-level height, and it does not use the effective-inflow-layer algorithm.

Vector wind difference from the effective inflow base up to one half of the most-unstable-parcel equilibrium level (EL) height; the effective inflow layer is bounded where lifted-parcel CAPE is at least 100 J/kg and CIN is weaker than -250 J/kg.

Reading it This field currently uses fixed 0 to 6 km bulk shear rather than true effective bulk shear, so it is not a replacement for effective-layer diagnostics in elevated or shallow-inflow setups, where the real inflow layer departs from fixed heights. Read the magnitude as deep-layer shear: supercell probability climbs as 0 to 6 km values pass 25 to 40 kt.

0–1 km Storm-Relative Helicity: the streamwise spin available to a storm’s inflow in the lowest kilometer.

How it's derived Integrated from the model hodograph against the Bunkers right-mover storm motion (the model’s own storm-motion output where shipped).

V is the wind vector, C the Bunkers right-mover storm motion, z height above ground, and the integral runs over the lowest 1 km; result in m^2/s^2.

Reading it This measures the streamwise spin a storm's inflow can tilt into rotation in the lowest kilometer, and it carries more weight on tornado potential than almost anything else. About 100 m^2/s^2 is meaningful and 200 to 300 is strong. Localized maxima along boundaries matter more than broad areas, and STP uses it scaled by 150.

Caveat SRH is very sensitive to the assumed storm motion and to small changes in the wind profile, so it needs frequently updated wind input, and a wide spectrum of values has been observed with any single tornadic event.

0–3 km Storm-Relative Helicity: streamwise spin through the storm-bearing layer.

u and v are wind components, c_x and c_y the Bunkers right-mover storm-motion components, z height above ground, integrated over the lowest 3 km; result in m^2/s^2.

Reading it This integrates storm-relative streamwise vorticity through the storm-bearing layer, so it captures rotation over the depth of the mesocyclone. SCP scales it by 50 m^2/s^2. Large 0 to 3 km values with weak 0 to 1 km values favor mid-level rotation without tornadoes.

Caveat The 0 to 3 km signal for tornadic supercells is weaker than the 0 to 1 km layer, and like all SRH it depends heavily on the assumed storm motion, which differs between models.

0–3 km bulk shear.

u and v are the wind components at the surface (subscript 0) and 3 km above ground (subscript 3).

Reading it This is the layer to watch for QLCS and mesovortex potential. Roughly 30 to 40 kt oriented across a convective line supports embedded circulations and bowing segments. It bridges the low-level tornado shear and the deep-layer supercell shear.

0-500 m storm-relative helicity, the shallow low-level streamwise vorticity a storm can tilt upright and stretch. The lowest 500 m is the layer most tied to tornado potential.

How it's derived Storm-relative helicity integrated from the surface to 500 m AGL using the interpolated wind column and the Bunkers right-mover storm motion.

Reading it A sharper tornado-environment signal than 0-1 km SRH because it isolates the near-ground inflow. Large values with a low LCL and strong low-level shear are the classic significant-tornado setup.

Caveat Interpolated from the mandatory wind levels, so coarser than a model that ships native 0-500 m helicity. Sensitive to the storm-motion estimate, which assumes a right-moving supercell.

Mid-level (4-6 km) storm-relative wind, the flow a storm feels relative to its own motion through the middle of the updraft. It shapes where precipitation falls relative to the mesocyclone.

How it's derived Magnitude of the mean 4-6 km AGL wind minus the Bunkers storm motion, in knots. This is deliberately the MID-level layer: true anvil-level (9-10 km) winds are not computable from the mandatory pressure levels this site ingests, so mid-level thresholds are used instead of anvil climatology.

Reading it A storm-mode hint. Weak mid-level SR flow (below about 16 kt) lets precipitation wrap into the mesocyclone toward high-precipitation, messier modes; roughly 16-28 kt is the classic-supercell range; strong flow (above about 40 kt) ventilates precipitation well downshear toward classic or low-precipitation structures.

Caveat A kinematic tendency, not a guarantee of mode. It interpolates the mandatory wind levels and shares the storm-motion assumption of the SRH fields.

Corfidi upshear vector magnitude: the mean 850 to 300 mb cloud-layer wind minus the low-level jet, in knots. This is the back-building component of MCS motion.

How it's derived Cloud-layer wind taken as the pressure-weighted mean through the 850 to 300 mb layer, so the deep middle levels carry more weight than the top and bottom; the low-level jet taken as the stronger of the 925 and 850 mb winds. The vector difference follows Corfidi, Merritt, and Fritsch (1996) and Corfidi (2003), the same construction behind SPC mesoanalysis Corfidi vectors.

V_cloud is the pressure-weighted mean wind through the 850 to 300 mb cloud layer, V_LLJ is the stronger of the 925 and 850 mb winds; the upshear (back-building) vector is their difference, in knots.

Reading it This is the upshear propagation component of MCS motion, and it is why storms train. A system moves with the cloud-layer flow while new cells build upstream toward the feeding jet, so when the two nearly cancel the system barely moves and cells track over the same ground. Small values under 10 kt with concurrent high precipitable water along an instability axis are a precursor for flash flooding potential. This is flow geometry only and says nothing about whether storms exist, so read it with reflectivity and the QPF probabilities.

Significant Tornado Parameter (fixed-layer): a normalized product of surface-based CAPE, cloud-base height, 0–1 km SRH, 0–6 km shear, and CIN, tuned so values of 1 or more pick out significant-tornado environments.

How it's derived The SPC fixed-layer formulation with all five terms multiplied together, using the SURFACE-BASED parcel throughout (SBCAPE, SB LCL, SBCIN) with the published clamps: the LCL term is 1 below a 1000 m cloud base and 0 above 2000 m; the CIN term is 1 above -50 J/kg and 0 below -200; the shear term is zeroed below 12.5 m/s (about 24 kt) and capped at 1.5 above 30 m/s (about 58 kt). The map, hover, pin, and meteogram all read the same server-computed grid.

SBCAPE in J/kg, ESRH (0-1 km SRH proxy) in m^2/s^2, EBWD (0-6 km shear proxy) in m/s, SBLCL in m, SBCIN in J/kg.

Reading it STP of 1 or more with storms present points to a significant-tornado environment, and 4 or more is a volatile setup. The CIN term multiplies the whole product, so a strong cap drives the value toward zero even when the other ingredients are favorable. It describes the environment only and says nothing about whether storms will form.

Caveat As a single composite it carries a relatively high false-alarm rate and inherits every limitation of its CAPE, shear, and helicity inputs, so high values can appear where no significant tornado follows. ESRH/EBWD are approximated with the fixed 0-1 km SRH and 0-6 km shear layers rather than the effective-inflow versions.

Supercell Composite Parameter: MUCAPE, 0–3 km SRH, deep shear, and a most-unstable CIN penalty multiplied into a supercell-likelihood index, following the SPC operational formulation.

How it's derived The full four-term SPC formulation: MUCAPE over 1000, SRH over the published 50 m²/s² scaling, the shear (EBWD) term zeroed below 10 m/s, scaled by 20 between 10 and 20 m/s, and capped at 1 above 20 m/s, and the most-unstable CIN factor (-40 J/kg divided by MUCIN, held at 1 when CIN is weaker than -40 J/kg) so capped environments are penalized. The map, hover, pin, and meteogram all read the same server-computed grid.

MUCAPE in J/kg, ESRH (0-3 km SRH proxy) in m^2/s^2, EBWD (0-6 km shear proxy) in m/s, MUCIN in J/kg.

Reading it SCP of 1 or more supports supercells once storms are present, and 10 or more marks an exceptional combination of instability and shear. Use the hazard-specific parameters to separate the hail, wind, and tornado threats.

Caveat It describes only the supercell potential of the environment and says nothing about whether storms form, and it inherits the same uncertainties as its CAPE, shear, and helicity inputs. ESRH/EBWD are approximated with the fixed 0-3 km SRH and 0-6 km shear layers rather than the effective-inflow versions.

Significant Hail Parameter: MUCAPE, parcel moisture, mid-level lapse rate, 500 mb cold, and deep shear combined toward the likelihood of hail two inches or larger, following the SPC five-term formulation.

How it's derived The full five-term SPC formulation over the published 42,000,000 divisor, with the published clamps — parcel mixing ratio held to 11-13.6 g/kg, the 500 mb temperature floored at -5.5 °C, shear clamped to 7-27 m/s — and all three low-end reduction factors: scaled down when MUCAPE is below 1300 J/kg, when the 700-500 mb lapse rate is below 5.8 °C/km, and when the freezing level sits below 2400 m AGL. The one remaining approximation: the most-unstable parcel mixing ratio is taken from the 2 m mixing ratio (the clamp bounds most of the difference). The map, hover, pin, and meteogram all read the same server-computed grid.

MUCAPE in J/kg, r_MU is the parcel mixing ratio in g/kg (clamped 11-13.6), gamma_75 is the 700-500 mb lapse rate in degC/km, T500 is the 500 mb temperature in degC (floored at -5.5), BWD06 is 0-6 km shear in m/s (clamped 7-27).

Reading it SHIP of 1 or more with supercells present supports significant hail of two inches or larger, and 2 or more strongly so. Pair it with mid-level lapse rates and the explicit hail diagnostic.

Caveat As an environment composite it does not account for updraft strength, hailstone residence time, or the melting depth below the wet-bulb-zero height, all of which govern whether large hail actually reaches the ground.

0–1 km Energy-Helicity Index: CAPE multiplied by 0–1 km SRH, scaled down by 160,000.

CAPE in J/kg (this site uses MLCAPE, falling back to SBCAPE), SRH over 0-1 km in m^2/s^2.

Reading it A quick combined look at buoyancy and low-level rotation. 1 is notable and 2 or more marks a strong tornado environment. STP has largely superseded it for operational use.

Caveat High CAPE can over-inflate EHI while strong CIN or weak surface-based CAPE can wreck it, and in low-CAPE high-shear setups it can underestimate tornado potential.

0–3 km Energy-Helicity Index. As the 0–1 km version but for mesocyclone-depth rotation.

CAPE in J/kg (this site uses MLCAPE, falling back to SBCAPE), SRH over 0-3 km in m^2/s^2.

Reading it The mesocyclone-depth version of the energy-helicity index, using rotation through the storm-bearing layer rather than just the lowest kilometer. Read it alongside the 0-1 km version, since large 0-3 km values with weak 0-1 km values favor mid-level rotation without tornadoes.

Caveat Observed values overlap between tornadic and non-tornadic storm classes, so it is not always a reliable discriminator and inherits the limitations of its CAPE and helicity inputs.

Craven-Brooks significant severe parameter: MLCAPE multiplied by 0–6 km shear.

MLCAPE in J/kg (falls back to SBCAPE), BWD06 is the 0-6 km bulk shear in m/s.

Reading it This is a blunt product of instability and deep-layer shear that holds up well in practice. About 20,000 m^3/s^3 or more historically separates significant-severe environments (two-inch hail, 65 kt wind, or an EF2+ tornado). Use it as a broad coverage flag, then go to the hazard-specific parameters.

Bulk Richardson Number: CAPE divided by a measure of deep-layer shear energy. The version here uses a nonstandard shear denominator, so it is a proxy and its thresholds are not directly interchangeable with textbook BRN values.

How it's derived The formal BRN shear denominator is the 0-6 km mean wind minus the 0-500 m mean wind, but the site uses the 0-6 km bulk shear vector difference as a nonstandard proxy, so values differ from the textbook definition. Treat the traditional storm-mode thresholds as a rough guide rather than directly equivalent.

MLCAPE in J/kg (the mixed-layer parcel better represents inflow buoyancy than a surface parcel; falls back to SBCAPE when MLCAPE is unavailable), U is the 0-6 km bulk shear in m/s used as the BRN shear proxy (floored at 1 m/s, result capped at 200).

Reading it Use this to sort out likely storm type. Roughly 10 to 45 favors supercells, while higher values mean weak shear relative to buoyancy and favor multicell and pulse modes. It breaks down at very low shear, where the small denominator inflates the number, so read it together with the hodograph.

Caveat Above about 4000 J/kg of CAPE the value stays large regardless of the shear denominator, and BRN ignores low-level hodograph curvature, so a low number can occur near tornadic storms.

GOES-19 ABI Channel 13 (10.3 µm clean infrared) brightness temperature — live satellite imagery, day and night.

Reading it Cold brightness temperatures mark high cloud tops: the colorized enhancement starts at -30 °C, storm anvils typically read -50s to -60s, and overshooting tops push past -70 °C. Watch cooling trends frame-to-frame for strengthening updrafts, and warm dark ground for clearing. This is an observation loop (5-minute CONUS scans), not model output.

Caveat IR sees cloud TOP temperature only — a cold anvil can hide what is underneath, and thin cirrus reads colder than its impact deserves. Pair with radar for precipitation structure.

Composite reflectivity: the maximum simulated radar reflectivity in the column, which is the model showing you what it expects the radar mosaic to look like.

Reading it Read storm mode straight off this field: discrete cells, clusters, and lines appear as they would on the radar mosaic, so it is the storm-organization check. Cores of 50 dBZ or more flag possible hail, and 60 dBZ or more marks a likely hail core. Trust it most as a comparison field, run against run and against the observed MRMS mosaic, because the model places convection imperfectly.

Hourly-max 2–5 km updraft helicity: the strongest mid-level rotating-updraft signal in each hour.

How it's derived The model’s hourly-max diagnostic over 2–5 km above ground. The verification literature treats this layer as the standard stand-in for any severe hazard.

w is vertical velocity (m/s), zeta is vertical vorticity (1/s), z is height above ground (m); contributions are summed only in the rising-air part of the layer.

Reading it Tracks where the model produces rotating mid-level updrafts, which is how a supercell shows up in the grid. Isolated streaks of 25 to 50 m^2/s^2 are ordinary rotating storms, while 75 m^2/s^2 or more on a 3 km grid is the calibrated all-severe threshold. This is a storm signal, not a forecast of reports, so check the run against radar and use the 0-2 km layer for a tornado-specific read.

Hourly-max 0–3 km updraft helicity: low-to-mid-level rotation.

w is vertical velocity (m/s), zeta is vertical vorticity (1/s), z is height above ground (m); same integrand as the 2-5 km field over a lower layer.

Reading it Samples rotation closer to the ground than the 2-5 km layer, which makes it better for QLCS mesovortices and low-level tornado potential. Values of 50 m^2/s^2 or more indicate organized low-level rotation. Magnitudes run smaller than the 2-5 km layer, so do not compare the two on one color scale.

Hourly-max 0–1 km vertical vorticity: raw near-ground spin.

zeta is vertical vorticity (1/s); u and v are the horizontal wind components and x and y are eastward and northward distance. The field is the hourly maximum of this quantity in the lowest 1 km.

Reading it Raw near-ground spin, reported as the hourly maximum in the lowest kilometer. Use it alongside low-level updraft helicity to corroborate mesovortex and landspout signals. Values near 250e-5 per second are extreme low-level rotation in the model.

Maximum hail diameter diagnostic.

How it's derived The model’s explicit hail diagnostic (HRRR/RRFS). A microphysics product, noisier than environment-based hail reasoning.

Reading it The model's explicit hail-size diagnostic from HRRR and RRFS, computed inside the microphysics rather than from the environment, so it is noisier than parameter-based hail reasoning. Treat 1 inch or more as a strong in-model hail signal and 2 inches or more as a large-hail signal. Verify the environment with SHIP and mid-level lapse rates before trusting the magnitude.

Hourly-max 10 m wind speed: the model’s peak surface wind each hour, including convective gusts.

Reading it The model's peak 10 m wind each hour, including convective gusts. 50 kt meets severe criteria, which equals 58 mph, not 58 kt. Because this is direct hazard output, it feeds the outlook's wind hazard without a separate environment check.

Vertically integrated liquid: a reflectivity-derived estimate of vertically integrated liquid water.

Z is radar reflectivity factor in mm^6/m^3, the bracket is the layer-mean reflectivity between two levels, delta h is layer thickness in meters, and the result is column liquid in kg/m^2 (Greene and Clark 1972).

Reading it A reflectivity-derived estimate of vertically integrated liquid water in the column. High cores correlate with large hail and intense water loading, since the reflectivity-to-liquid relation saturates where big hydrometeors dominate. A sudden collapse of a VIL core can precede a downburst, so watch the time trend, not just the peak.

Hourly-max 0–2 km updraft helicity: near-ground rotation, the most tornado-specific layer the convection-allowing models publish.

How it's derived The hourly-max diagnostic over 0–2 km above ground, ingested from HRRR and RRFS.

w is vertical velocity (m/s), zeta is vertical vorticity (1/s), z is height above ground (m); the same updraft-helicity integrand restricted to the lowest 2 km.

Reading it The most tornado-specific updraft-helicity layer the convection-allowing models publish, ingested from HRRR and RRFS. Values of 14 m^2/s^2 or more, paired with a supportive environment, flag near-ground rotation. Magnitudes run far smaller than the 2-5 km layer, so keep them on separate scales.

REFS probability that hourly-max 2–5 km updraft helicity exceeds 25 m²/s² nearby, so it covers where the ensemble expects rotating storms.

How it's derived A NOAA-computed neighborhood probability from the REFS ensemble, ingested directly rather than recomputed here.

Reading it Read it as the ensemble flagging any organized storm. Drives the experimental outlook’s REFS mode with environment corroboration.

REFS probability of hourly-max 2–5 km updraft helicity above 75 m²/s², meaning strong mesocyclones.

Reading it Weighted heavily in the outlook’s tornado hazard. Double-digit values localize the day’s most intense rotation threat.

REFS probability of composite reflectivity above 40 dBZ, meaning strong-storm coverage.

Reading it Where the ensemble puts strong cores. The outlook’s thunder line falls back to it when the lightning probability is unavailable.

REFS probability of MLCAPE above 1000 J/kg.

Reading it Ensemble instability support, one of the two environment checks in the outlook’s ingredients rule.

REFS probability of 0–3 km storm-relative helicity above 200 m²/s².

Reading it Ensemble rotation support, the other outlook check. Blended additively with the instability probability so high-shear, low-instability regimes are not zeroed out.

REFS probability of 10 m wind above 40 kt.

Reading it A direct strong-wind probability. It enters the outlook unweighted because it measures the hazard directly.

REFS probability of 2 m temperature at or below freezing.

Reading it The winter-impact baseline: where the ensemble says surface freezing is likely during the hour.

REFS probability that the precipitation type is snow.

Reading it Precipitation-type guidance from the full ensemble, which carries more information than any single deterministic call.

REFS probability that the precipitation type is freezing rain.

Reading it The ice-storm flag. Even 20 to 30 percent deserves attention given the impact asymmetry.

REFS probability of more than 1 inch of snow in one hour.

Reading it The banded heavy-snow signal, where crippling rates may set up.

REFS probability of at least a hundredth of an inch of freezing rain in one hour.

Reading it The accruing-glaze signal for travel and power infrastructure.

REFS probability of more than half an inch of rain in one hour.

Reading it Heavy-rain coverage. Pair with precipitable water and storm motion for flash-flood reasoning.

REFS probability of more than 1 inch of rain in one hour.

Reading it The serious flash-flood-rate signal.

REFS lightning probability.

How it's derived NOAA-computed from the ensemble’s lightning diagnostics.

Reading it The general-thunder line. SPC’s TSTM category is defined as at least a 10 percent chance of thunder, and this is that probability from a real ensemble.

250 mb geopotential height: the jet-stream level.

Reading it Jet streaks mark favorable regions for deep ascent, often near the left-exit and right-entrance regions of the streak, which can support severe weather and heavy rain when other ingredients align.

500 mb geopotential height: the synoptic steering pattern at mid-levels.

Reading it Troughs (height minima) bring cold air aloft, and ascent is often favored downstream of trough axes rather than at the trough itself; the gradient gives steering flow. Heights are in meters here. SPC charts often label decameters.

700 mb geopotential height (~3 km): the layer between the low-level jet and the mid-level steering flow.

Reading it Classic chart for tracking shortwaves that are too shallow to show at 500 mb, and the level where warm-nose intrusions and dry punches organize. Pairs with 700 mb RH for frontal moisture structure.

850 mb geopotential height.

Reading it Low-level jet and warm-advection patterns live here. 850 mb lows track moisture transport.

700 mb relative humidity: mid-level moisture, the classic "dry slot" and cloud-deck chart.

How it's derived Direct model RH at 700 mb where shipped, otherwise computed from 700 mb temperature and dewpoint (Magnus).

Reading it High values mark deep moisture and elevated cloud decks; sharp dry tongues behind a system mark the dry slot and downslope drying. Winter forecasters read it with 700 mb heights for overrunning precipitation; fire forecasters watch mid-level drying ahead of wind events.

500 mb absolute vorticity: spin at mid-levels (relative vorticity plus the planetary contribution). The classic shortwave-tracking chart.

How it's derived Finite differences of the model’s 500 mb winds on the native grid (plus the Coriolis parameter and spherical curvature term), smoothed to an effective ~25 km radius so convection-allowing grids read synoptically.

Reading it Vorticity maxima mark shortwave troughs; by quasi-geostrophic reasoning, ascent is favored downstream (east) of a max where positive vorticity advection increases with height. Track maxima frame-to-frame to see energy ejecting into a warm sector.

Caveat Values below the planetary background (~8-10 at midlatitudes) are dynamically quiet; the color ramp only lights up above it. On 3 km models small convective circulations still peek through the smoothing at short ranges.

850 mb temperature.

Reading it Warm-air advection at 850 mb drives overnight elevated convection. The 0 °C line at 850 mb is a first-guess rain-versus-snow discriminator in lowland winter setups.

250 mb wind speed. Jet-streak placement; streaks above 100 kt matter for severe-weather ventilation and turbulence.

500 mb wind speed. Mid-level flow strength, a supercell organization and storm-motion input.

850 mb wind speed.

Reading it The low-level jet. Nocturnal 850 mb jets of 40 kt or more feed MCSs and overnight tornado risk.

Surface wet-bulb temperature: the temperature air reaches when cooled by evaporating water into it.

Reading it Surface wet-bulb helps assess evaporative cooling and near-surface freezing potential. It cannot determine precipitation type by itself; use the full temperature and wet-bulb profile to identify melting and refreezing layers.

2 m air temperature.

How it's derived Model 2 m output converted to °F.

Reading it With dewpoint, the basic state of the boundary layer.

Apparent temperature, the "feels like" value that combines air temperature, humidity, and wind. Warm humid air feels hotter and cold windy air feels colder.

How it's derived Steadman apparent temperature from the 2 m temperature, the 2 m dewpoint (as water-vapor pressure), and the 10 m wind, converted to °F.

T in °C, e the water-vapor pressure (hPa) from the dewpoint, ws the 10 m wind (m/s); the result is converted to °F. Steadman (1994), the basis for the Australian apparent temperature.

Reading it Reads the heat and cold stress on people directly. Around 105°F it signals dangerous heat, and well below freezing with wind signals dangerous cold. One continuous field, so it spans summer heat and winter wind chill in a single layer.

Caveat A whole-body estimate for someone in the shade. Direct sun, exertion, and clothing shift the real feel. It is not the NWS heat index or wind chill, which are separate single-regime formulas.

Heat index, the NWS "feels like" temperature for hot, humid conditions. High humidity slows sweat evaporation, so the body cannot shed heat and the air feels hotter than the thermometer reads.

How it's derived NWS Rothfusz regression in °F and relative humidity (derived from the dewpoint) with the standard low- and high-humidity edge corrections. Below about 80°F it falls back to a simple form near the air temperature.

T in °F, R the relative humidity in percent. The full Rothfusz regression has nine terms plus humidity-edge corrections; it is only meaningful in hot, humid air and reduces to the temperature when it is cool.

Reading it A summer heat-stress layer. NWS categories: caution above 80°F, extreme caution above 90°F, danger above 103°F, extreme danger above 125°F. Read it alongside the local heat-advisory and excessive-heat-warning thresholds.

Caveat Defined for hot conditions; in cool air this layer just tracks the temperature. It assumes shade and a light wind, so full sun can add roughly 15°F to the real feel.

Wind chill, the NWS "feels like" temperature for cold, windy conditions. Wind strips the thin warm layer next to the skin, so the body loses heat faster and the air feels colder than the thermometer reads.

How it's derived NWS 2001 wind chill from the 2 m temperature and the 10 m wind. Valid for temperatures at or below 50°F with wind at or above 3 mph; in calmer or warmer conditions it falls back to the air temperature.

T in °F, V the 10 m wind speed in mph. NWS / Environment Canada 2001 formula, which replaced the older Siple-Passel index.

Reading it A winter cold-stress layer. The hazard is frostbite: exposed skin can freeze in about 30 minutes near a −15°F wind chill and in about 10 minutes near −40°F. Drives wind-chill advisories and warnings.

Caveat Defined for cold, windy conditions; in calm or warm air this layer just tracks the temperature. It estimates heat loss from exposed skin, not the deeper danger of wet clothing or immersion.

Dewpoint depression, the gap between temperature and dewpoint. A direct read on how close the surface air sits to saturation.

How it's derived The 2 m temperature minus the 2 m dewpoint, in Fahrenheit degrees.

Reading it Small values of a few degrees flag fog, low stratus, and saturated inflow; large values mark dry, well-mixed air and a high cloud base. A fast way to spot moisture boundaries and dryline gradients.

Caveat A surface-only measure. It says nothing about moisture aloft, so pair it with precipitable water or the RH profile for the full column.

2 m water-vapor mixing ratio, the absolute moisture in the near-surface air in grams of vapor per kilogram of dry air. Unlike relative humidity, it does not change as the air warms or cools.

How it's derived From the model 2 m specific humidity when shipped, otherwise from dewpoint and surface pressure by the Magnus relation.

Reading it The true low-level moisture for severe weather and the boundary layer. Values of 12 g/kg and up support strong storms, and the mid-to-high teens are very moist. It cuts through the diurnal swing that misleads relative humidity.

Caveat A single-level value at 2 m. Deeper moisture matters too, so read it with dewpoint and precipitable water.

2 m dewpoint: the absolute moisture content of the near-surface air.

Reading it Read this as the fuel for severe weather. Dewpoints in the 60s °F support strong storms; the 70s are volatile. Watch dewpoint pooling along boundaries.

2 m relative humidity.

Reading it Fire weather on the low end (below 20 percent with wind), fog or stratus on the high end (above 95 percent overnight).

10 m sustained wind speed.

Reading it The surface wind field: boundaries, drylines, and warning-relevant sustained winds.

80 m wind speed, turbine-hub height.

Reading it The wind-energy planning level. Also a view of the nocturnal low-level jet decoupled from the surface.

10 m wind gust: the model’s estimate of peak short-duration wind.

How it's derived The model gust diagnostic. ECMWF defines this product as a maximum since the previous output step, so its analysis hour (f000) is encoded as zero.

Reading it Gusts of 35 kt disrupt outdoor operations; 50 kt is severe criteria.

Mean sea-level pressure.

Reading it Cyclone tracks, mesolows, and wake lows. The map can overlay its isobars on any field.

Station (surface) pressure, the actual pressure at model ground level. Used internally for terrain masking in the sounding and time-height tools; on the map it mostly traces elevation.

Total cloud cover. Heating potential ahead of convection; overnight decks moderate cooling.

Surface visibility.

How it's derived Converted to miles and capped at 10. Each model encodes unlimited visibility with a different stand-in value (HRRR 90 km, GFS 24 km); the cap makes models comparable.

Reading it Fog and blowing-precipitation impacts. Under 1 mile is dense-fog or whiteout territory.

Vapor pressure deficit, how far the air sits from saturation. It is the drying power of the air and the single best driver of how fast fine fuels and vegetation lose moisture.

How it's derived Saturation vapor pressure at the temperature minus saturation vapor pressure at the dewpoint, in kilopascals.

Reading it Fire and agriculture. High VPD means thirsty air pulling moisture from fuels, soil, and plants; values above about 4 kPa are a strong fire-spread signal. It rises with heat and falls with humidity, folding both into one number.

Caveat A surface dryness measure with no wind or fuel-loading information, so read it with the Hot-Dry-Windy and Fosberg indices for the full picture.

Hot-Dry-Windy Index, a fire-weather potential that multiplies how dry the air is by how hard the wind blows in the layer fire actually feels.

How it's derived The maximum vapor pressure deficit (hPa) times the maximum wind speed (m/s) in the lowest 500 m. Srock et al. (2018).

Reading it A clean read on the days fire spreads fast and smoke plumes loft. It deliberately ignores fuel state to isolate the meteorology. Watch for sharp local maxima downslope and behind dry fronts.

Caveat Atmosphere only: it assumes fuels are there to burn, so a high index over green, wet fuels is not a fire. The lowest-500 m sampling is coarse where the column is thin.

Fosberg Fire Weather Index, a classic 0 to 100 measure of how readily fire spreads given temperature, humidity, and wind.

How it's derived An equilibrium fine-fuel moisture term from T and RH, damped into a moisture factor, then scaled by wind speed and normalized to roughly 0 to 100.

Reading it A fast-responding, wind-dominated fire-spread signal. Values climbing through 50 to 60 mark critical conditions and 80 plus is extreme. Long used operationally, so the thresholds are familiar.

Caveat Based on dead fine-fuel moisture, not live fuels or drought, so it can read high in spring before fuels cure. Sensitive to the surface wind, which models smooth over terrain.

Ventilation rate, how vigorously the boundary layer flushes smoke and pollutants away. It pairs the depth of the mixed layer with the wind moving through it.

How it's derived PBL mixing height times the mean transport wind through the mixed layer itself — the wind column is sampled at fixed heights and averaged only up to the local mixing height, so a shallow stable layer under a strong low-level jet is not credited with dispersion it does not feel. Square meters per second.

Reading it Smoke management and air quality. Low ventilation traps smoke and haze near the ground, often overnight and in valleys; high ventilation disperses it. Prescribed-burn go/no-go decisions key off this.

Caveat A bulk dispersion estimate that does not resolve terrain channeling, nocturnal drainage, or plume rise. The mixing height itself is a modeled diagnostic that varies between models.

Precipitation type: rain, snow, sleet (ice pellets), or freezing rain. Which one falls depends on the temperature profile the hydrometeor passes through on the way down.

How it's derived A warm-nose classification from the warmest ABOVE-GROUND temperature among 925/850/700 mb (levels under the model surface are excluded, so high terrain is not misread through below-ground extrapolations) and the 2 m temperature. Snow requires both a cold column and a near-freezing surface; a warm nose over a cold surface deck refreezes to sleet when the deck is deep and cold, or freezing rain when it is shallow.

Reading it Where the hazard is. Freezing rain glazes roads and downs power lines; sleet accumulates like tiny ball bearings; the rain/snow line is the forecast headline. Pair it with snowfall and thickness.

Caveat A simplified profile method, not a microphysics scheme. It can miss thin warm or cold layers the mandatory levels do not resolve and does not weight intensity. It is masked to where the model is precipitating, so dry areas read blank.

Snowfall accumulation, the depth of snow from the model liquid precipitation using a variable snow-to-liquid ratio instead of a flat 10:1.

How it's derived Accumulated hour by hour: each interval adds the Kuchera (2007) snow-to-liquid ratio for THAT hour (set by the column maximum temperature) times that interval's liquid precipitation, and only when that hour's near-surface is at or below freezing — so rain before a changeover is never converted to snow, and snow already on the map does not vanish when the surface later warms.

Reading it A first-guess accumulation map that beats a fixed 10:1 in cold, fluffy events where ratios run 15-25:1. Read trends and gradients more than exact inches.

Caveat Inherits all of the model QPF error, and the ratio is a single-variable estimate that ignores riming, compaction, and mixed precip. Marginal events near freezing are the least reliable.

Precipitable water: total column water vapor, in inches.

Reading it High precipitable water supports heavy rain, but it does not set the ceiling alone: lift, warm-cloud depth, precipitation efficiency, storm motion, and antecedent conditions all matter too. 1.75 to 2.0 inches marks rich tropical moisture; compare against climatology for the season. High precipitable water with slow storm motion is a common flash-flood signal.

Total accumulated precipitation since the run’s start.

How it's derived The model’s run-total accumulation message where both bucket and total are shipped; the decoder explicitly selects the longest window. GEFS publishes 6-hour buckets, so its map shows bucket values. The point tools handle the resets.

Reading it Read differences between hours for interval rainfall. The precipitation panel does this per step at the pinned point.

Instantaneous precipitation rate.

Reading it Where it is raining in the model at this hour, and how hard. An inch per hour or more supports flash-flood rates.

EXPERIMENTAL run-period severe outlook: SPC-style categories (TSTM through HIGH) derived from this model’s own storm and environment fields over the run’s first 24 forecast hours.

How it's derived A union of severe signals over the period, smoothed to the neighborhood scales from the verification literature, then mapped through SPC’s published probability-to-category table. Full methodology with citations below.

Reading it First-guess, model-specific guidance. It is not an SPC product and is not calibrated against storm reports. The underlying percentages are model surrogate-signal coverage, not calibrated probabilities of a severe-weather report. Do not brief this alone. Its best use is comparison: against SPC’s actual outlook to see where this model is more or less aggressive, and against the same product from another model.

EXPERIMENTAL hourly severe signal: 0 to 100 neighborhood coverage of this hour’s severe signals.

How it's derived A sharp 40-km neighborhood coverage of the same signals the period outlook unions, computed per hour. SPC categories are 24-hour quantities, so this product does not use them.

Reading it Timing. Scrub it to see when the model focuses its severe signal, and read magnitudes as relative within the run. The percentages are model surrogate-signal coverage, not calibrated probabilities of a severe-weather report. This is a storm signal, not a forecast of reports.