Getting started
The viewer shows UKV model soundings for the UK and Ireland, with radiosonde ascents to compare against. Pick a place on the map, a run and a time; the skew-T is in the middle and the diagnostics for the chosen profile mode are on the right.
Choosing an ascent
- Dots are UKV columns every 0.5° over the whole domain, so you can follow storms and convection coming in from the sea, and every 0.25° over land and within about 0.1° of the coast. Squares are radiosonde stations; their times come from the University of Wyoming.
- Find a place: the box above the map takes a name (towns of 1,000 people or more in the UK and Ireland) or a latitude and longitude such as 57.2, -3.8; / jumps to it. The button beside it uses your own location, which stays in your browser. Each selects the nearest UKV point.
- Sharing a view: the address in the browser carries the place, the time and the mode, so copying it shares exactly what you see. A plain visit opens at the present time, at the place and in the mode you used last.
- Background fields on the map help you place a click against fronts, showers, jets and fog. Each tab has its own, from the UKV run at the slider's time (also while a radiosonde is shown), averaged to about 5.5 km:
- Convection, top to bottom: MSLP every 4 hPa; the precipitation rate at that moment, so it matches the ascent's time, in blues to purple from 0.1 mm/h; the UKV's own most-unstable CAPE (parcels from below 500 hPa) shaded from 50 J/kg. Unstable air with no precipitation over it has not set off showers in the model, often because it is capped: the open data has no CIN, so check the ascent. The model's MUCAPE is its own calculation and will not exactly match the table's, which uses the viewer's parcel. Runs built before mid-November 2026 have no MUCAPE.
- Aviation, top to bottom: MSLP every 4 hPa; the precipitation rate; total cloud fraction in greys from 30%. Runs built before the cloud was kept (November 2026) show the precipitation and MSLP only.
- Lee Waves: 850 hPa height every 3 dam, and 850 hPa wind speed from 20 kt. 850 hPa (about 1.5 km) is above nearly all UK ground, close to where the lee wave finder starts.
- Winter: lines of the wet-bulb freezing level every 100 m below 1000 m above sea level, where snow may reach low ground, over the precipitation type (rain, mixed, snow, and hail or graupel where it is at least a fifth of the total), stronger colour for heavier rates.
- Temperatures: the 1000–850 hPa thickness shaded from cold blue to warm red every 20 gpm, with lines every 10 gpm: a guide to how warm the lowest 1.5 km is.
- Fog: MSLP, and visibility below 2500 m with a bold line at 1000 m. UKV visibility is cut by heavy rain and snow too, not only by fog.
- Run: four a day. The 03 and 15 UTC runs go 3-hourly out to T+120; the 09 and 21 UTC runs stop at T+54. Changing run keeps the place and the nearest valid time. Runs are kept for 14 days.
- Time: the slider under the plot, the buttons beside the time, or ← →; Shift with an arrow, or the double arrows, goes back or on 24 hours. While you stay at one place, the plot keeps its position and the scales of the left strip and hodograph, so the changes from one time to the next are what you see.
- Flags give a heads-up where a run's UKV points cross a few thresholds. The ⚑ Flags button on the map counts the kinds of flag on the tab you are in and lists its areas, then one row per kind, with its peak and how many points and times it covers (the numbers on the tab buttons count each tab's areas, below). Each row and group is labelled by how widespread it is at its busiest time, as a share of the points it can occur at (for lee waves, those near hills): Very ISOL (very isolated) under 1%, ISOL (isolated) under 5%, OCNL (occasional) under 20%, WDSPR (widespread) from 20%. A row opens into its groups of points close together in space and time, the most important first: how far over the threshold its peak is, times its points, doubled when the peak is within the next 24 hours. A click on a group opens its strongest point and time, in its tab. With more than 30 groups in a run, rows start closed and groups of a single point are held back behind a Show button.
Once you have checked the profiles behind a kind of flag, one group of it, or the whole tab (Mute all), Mute greys it, moves it to the bottom of the list and turns its rings grey; the button stops counting a muted kind, and when every flag of a tab at the slider's time is muted, its number on the tab button turns grey. Mutes are kept in this browser and carry over to later runs over the same valid times (a group's mute where it shares a point), unless the flag grows stronger: a peak 25% higher, or a group half as big again and 3 points more. Then it comes back marked ↑ Stronger. Unmute undoes a mute.
Areas, at the top of the list: the points flagged by any of the tab's kinds that lie close together (within about 1°) at the same or the next 3-hourly time, so an area follows its weather through the run. Each gives where it is, the kinds in it, its points and times, and its strongest point (the furthest over its threshold), most important first; a click opens that point and time. The first six are shown, the rest behind a button; By kind below lists each kind with its own groups, for muting. Context flags (unusually moist or warm, moist air: PW and θw850) are listed and in the briefing, but are not hazards in themselves, so they join no area, ring no points and are not counted.
A number on a mode button counts the areas with a flag at the slider's time (its tip gives the points too; a small ring: flags at other times only), and the map rings the flagged points in the tab's colour on a white halo, so they show on any background field, thin and grey when muted. Only the points flagged at the slider's time are ringed: step through the run to see where each time's flags are. Where the ascent shown has one, a note above the summary says what crossed and what it means, and the badge behind it is ringed. They are model flags, not warnings, worked out for each UKV column at its own ground with undiluted parcels, so the tables can differ once the ground height, ECAPE or Normand's point is changed.- Convection: SHIP 0.3 or more; SCP 1 or more (by size, so left-movers count); STP 0.3 or more, effective or fixed layer; CAPE 1000 J/kg or more (the largest of the surface, mixed layer and most unstable parcels) with effective bulk shear of 20 kt or more.
- Aviation: strong Cb, CAPE 1000 J/kg or more; severe icing at levels adding up to 3000 ft or more of the column, with a reason for it to be severe: large supercooled drops (freezing drizzle or freezing rain risk moderate or high), ascent of 0.5 m/s or more at a severe level, or embedded convection (most unstable CAPE of 250 J/kg or more); trapped lee waves steep enough to break, near hills.
- Lee Waves: lee wave vertical velocity 1000 ft/min or more (Shutts 1997), near hills: within 30 km of UKV ground heights that span 300 m or more. The lee wave finder knows the air but not the hills, so away from them it finds waves nothing would launch; within 30 km keeps the waves that run on downwind, over lower ground or the sea.
- Winter, only where the UKV has 0.1 mm/h of precipitation or more in the point's map cell at that time: freezing rain risk moderate or high; freezing drizzle risk high; Lowndes' mean chance of snow 70% or more.
- Fog: the UKV's fog fraction 70% or more.
- Temperatures: the UKV's maximum for the UTC day 32 °C or more (once a day, at the warmest time); its minimum in the next 18 hours −10 °C or lower (once a spell, at the coldest time); screen wet bulb 24 °C or more.
- Unusual for the season, against the ERA5 climate (1993–2022) of the calendar month at the point, each quoting the percentile it crossed. Convection: precipitable water, or the 850 hPa wet-bulb potential temperature, at or above the 99th percentile; most unstable CAPE at or above the 99th percentile of ERA5's afternoon CAPE and 500 J/kg or more (in winter, when that percentile is a few hundred J/kg at most, simply 500 J/kg or more). Temperatures: 850 hPa temperature at or above the 99th percentile. Winter: 850 hPa temperature at or below the 1st percentile; the freezing level (above sea level) at or below the 1st percentile, where that percentile is 300 m or more above the ground. ERA5 is about 30 km across (precipitable water and 850 hPa fields 1.5°, so the nearest of fewer, larger cells), 6-hourly, and its CAPE is for 15 UTC only, so a night's CAPE is compared with the afternoon's.
- The Winter, icing and lee wave flags are found by running this page's own code over every point and time, so they agree with the summaries. A run shows them a few minutes after it first appears.
Briefing, the switch at the top of the flags list, gathers every tab's flags day by day (UTC), most important first: what it is, how widespread it is that day, the areas with most points (“mainly …”), and the windows of time it is flagged at. The UKV is kept every 3 hours, so 06–12Z means flagged at 06, 09 and 12Z, not checked continuously in between. Muted flags are left out. Click a line for its strongest point and time; Copy puts the briefing on the clipboard as plain text, for a shift handover or a chat.
Above the time slider, a strip of bars shows how many points are flagged at each time on the tab in use (taller for more, grey when muted, the current time picked out): the times worth stepping to. Click a bar to go there. - Pins overlay up to six ascents, model or observed, in colour. Click a pin's name to make it the primary ascent; the others are drawn fainter. The pencil renames a pin (for example "Peak instability London on Tue"); the name is used in the comparison table, the barb keys and exported images, and lasts until the page is closed. Clear all removes every pin, and turns into Undo for 8 seconds. Add nearest ascent pins the closest radiosonde within 6 hours of a UKV time, or the UKV column for a radiosonde.
- Is the data current? The dot beside the title: green when the newest run and ascents are what the schedule says should be here by now (a run about 5 hours after its 03, 09, 15 or 21 UTC start, ascents about 4 hours after 00 and 12 UTC), amber when one is a cycle late or the server's own hourly check has found a problem, red when two or more cycles late. Click it for the details. The page checks again every 10 minutes and when you come back to the tab; a run that arrives meanwhile is added to the list and offered as New: … run, never switched to by itself.
Uploading an ascent
Upload CSV, under Compare ascents, shows any ascent from a CSV file: a radiosonde from anywhere in the world, or a column of ERA5, the ECMWF reanalysis back to 1940. The file is read and worked out in your browser and never sent anywhere. Every mode, table, tool and export works on it as on a radiosonde, with the indices worked out by the same method as the pipeline's.
- Where to get one: rawinsonde.com gives raw data for observed ascents and ERA5 columns already in this format. Download its CSV and upload it unchanged.
- Columns, named in the first row, in any order:
Pressure(hPa),Altitude(m above sea level),TemperatureandDew_point(°C),Wind_direction(degrees, the direction the wind blows from) andWind_speed. Pressure, altitude and temperature are needed on at least 10 levels; a blank dew point or wind is a gap. Then eitherWMO_ID, for an observed ascent, orSigma_level, for ERA5. Other columns are ignored. - Wind speed: rawinsonde.com gives observed ascents in m/s and ERA5 in knots, so by default a file with
WMO_IDis read as m/s and one withSigma_levelas knots. If the barbs look half or twice what they should be, choose the unit beside the button and the file is read again. - Time and place come from the file name as rawinsonde.com writes it:
RAW_03808_2026-11-03_0000.csv(WMO station 03808, 3 November 2026 at 00 UTC), orRAW_-1300_3400_1940-01-01_00UTC.csv(longitude −13.00, latitude 34.00, 1 January 1940 at 00 UTC). A WMO number is looked up among some 2,400 radiosonde stations worldwide (NOAA's IGRA2 list). The ground is the lowest level in the file. - The map marks the ascent if it is within the map, and otherwise says where it is; the background fields are hidden, as they are the UKV's. The time slider has the one time.
- One at a time: a new upload replaces the last. Pin it first to compare it with the next upload, a UKV column or a radiosonde. It is not in the address, so a shared link does not carry it, and it is gone when the page is reloaded.
- ERA5 is a reanalysis about 31 km across on 137 model levels: smoother than a radiosonde, with no sharp inversions or thin moist layers, and its winds are averages over the grid box.
Through the run
Under the skew-T, Through the run opens every lead of the selected run at the selected UKV point (it stays open or shut as you left it):
- Time–height: height above sea level against time, in the height unit chosen. Shading: relative humidity, cloud fraction, derived w or θe (For this mode picks w in Lee Waves and humidity otherwise). θe's colours span its range in the lowest 6 km. Lines: the 0 °C isotherm (bold), −10 and −20 °C. Winds every 6 hours (12 on narrow screens). The top is 12 km in Convection and Aviation, 10 km in Lee Waves, 6 km in Winter, 4 km in Temperatures and 3 km in Fog.
- Indices: the mode's main numbers through the run, with the values at the selected time beside the blue line. Convection: MU and SB CAPE, ML and SB CIN, 0–6 and 0–1 km shear, 0–3 and 0–1 km SRH for the right mover. Lee Waves: the 0.5–3 km wind, the lower-layer 2π/l (shaded where Scorer's test says trapped) and the strongest w below 500 hPa. Winter: screen and 850 hPa temperature, the wet-bulb freezing level and 0 °C height above ground, and the 1000–850 hPa thickness against 1293 gpm. Temperatures: screen temperature and dew point, the 925 and 850 hPa temperatures mixed down, the 850 hPa temperature and the 1000–850 hPa thickness. Fog: screen visibility (log scale, line at 1000 m), screen T and Td, UKV fog fraction and 10 m wind.
Top sets the highest height drawn, in the plot's height unit: the mode's own top (above), or a round height lower or higher, for example 10,000 ft (FL100) for helicopter work, which also brings the barbs closer together. It is shared with the cross-section and kept in this browser. Smooth, on by default and shared with the cross-section, runs the shading linearly from one 3-hourly time to the next on a continuous scale instead of a block per time; the caption above the plot says so while it is on, since the model was only kept every 3 hours. Lines and barbs are not affected.
Click anywhere on either to show that time on the skew-T. Radiosondes have only one time, so the section is drawn for UKV points.
Cross-section
Under Through the run, Cross-section draws the UKV columns along a route at the time on the slider: distance along the route across (nautical miles with flight levels, km otherwise) and height above sea level up, with the ground under each column.
- A route: type the waypoints joined by “to”, for example
EGLL to EGPH,Bristol to Norwichor55.9, -3.4 to 57.2, -2.2. Airfields go by ICAO code or name (Heathrow), places by name. Or press Pick on map and click the points in turn; a click near an airfield takes it. Undo drops the last point; Done or Esc finishes. The route is drawn on the map while the section is open and is kept in this browser. - Columns: each is the 0.25° UKV point nearest the line (within about 30 km of it), in the order they come along it, so a column can sit up to about 15 km to one side. Each column stretches half way to its neighbours.
- Top: as in Through the run, and shared with it: the mode's own top or a round height in the plot's unit.
- Shading as in Through the run, plus wind speed (from 40 kt), or None for lines alone; For this mode picks icing in Aviation, w in Lee Waves and humidity otherwise.
- Layers: Smooth shading, on by default and shared with Through the run, runs the shading linearly from each column to the next, with the shades on a continuous scale, instead of a block per column; it reads more easily but suggests more detail between the columns than the model points hold, so the caption says when it is on. Any number of lines and marks can go over the shading at once: isotherms 0, −10 and −20 °C; isentropes (θ every 2 K, where lee waves show as ripples); θe every 4 K; RH 70 and 90%; isotachs every 20 kt from 40 kt; vertical velocity (up solid, down dashed, at the second and fourth of the shading's steps); an outline of moderate or worse icing; turbulence layers from the Richardson number (paler where possible); ▲ Cb tops where the most unstable parcel has 100 J/kg or more; and wind barbs. Each mode starts with its own set (Convection: isotherms, θe and barbs; Aviation: isotherms, turbulence, Cb tops and barbs; Lee Waves: isotherms, isentropes and barbs; Winter and Fog: isotherms, RH and barbs); a change is kept for that mode until Back to this mode's layers. Heights follow the plot's unit (metres, feet or flight level) unless another is chosen here.
- Click a column to show it on the skew-T; the blue line marks the point shown when it is on the route.
Reading the skew-T
- Red: temperature. Green: dew point. Isotherms lean to the right; 0 °C and −20 °C are picked out.
- Brown solid lines: dry adiabats. Green dashed: saturated (pseudo-)adiabats. Blue dotted: humidity mixing ratio, labelled in g/kg.
- Right of the plot: heights of the isobars, then wind barbs in knots. Each pinned ascent gets its own column of barbs in its colour, beside the primary one. In Convection the primary ascent's barbs take the hodograph's colours by height above ground (0–1, 1–3, 3–6, 6–9 and 9–12 km; plain above), so the two can be read together.
- Left strip: derived vertical velocity in ft/s (m/min with metres), or thermal advection. The switch just above the strip changes it (key a).
- Zoom with the scroll wheel or a pinch; drag to pan. Reset view (key z) returns to the whole ascent. Zooming scales both axes together, so angles on the diagram never change. The pointer is an arrow over the plot and a hand only while you drag it.
Profile modes
Convection, Aviation, Lee Waves, Winter, Fog and Temperatures (keys 1–6, in that order) each have a tab here. The coloured bar under the title shows which mode you are in. Hover over any ? in the page for a short note. Tap or click a coloured badge in a summary to see every badge that could fill that space, None included, each with a short definition; the one showing is ringed. A click anywhere else, or Esc, closes it.
Citing the viewer
If you use a skew-T or other figure from this viewer in a publication, please cite it as:
Interactive Sounding Viewer, Created by Nicholas Silkstone (2026). ORCiD: https://orcid.org/0009-0004-4223-7054
and credit the data as well: the Met Office for the UKV, the University of Wyoming or NOAA IGRA2 for radiosondes, and for ERA5 the Copernicus Climate Change Service (Hersbach et al. 2020, Q. J. R. Meteorol. Soc. 146, 1999–2049).
Thanks to M. Lehnert and N. Armstrong for ideas and suggestions.
Keys
z reset view · ← → time · p pin · n Normand's point · f fog point · m measure · a left strip · u height unit · 1–5 modes.
Convection
The surface parcel is drawn on the plot with its LCL, LFC, EL and Parcel Top. The hail growth zone (−10 to −30 °C) and the effective inflow layer are bracketed at the left.
Convection summary
Under the hodograph, the key measures as coloured badges; every table is under Diagnostics details. On the hodograph the steering wind (the mean of a layer measured on the plot) is a large labelled diamond; with Measured storm motion it is the storm motion too, so it is drawn once. A custom storm motion is a large labelled triangle.
- Parcels
- CAPE, or ECAPE with the switch, for each parcel: Low from 100 J/kg, Moderate from 500, High from 1000. CIN beside it in blue, as a lid rather than a hazard: Weak above −25 J/kg, Moderate to −100, Strong below. Then the EL and Parcel Top above ground with the air temperature at each, a guide to how cold the tops get.
- Bulk shear
- EBWD, from the base of the effective inflow layer to half the most unstable parcel's EL, with each parcel's own effective value, and 0–6 km with 0–1, 0–3 and 0–8 km. SPC's guidance: under 20 kt pulse storms, 20–30 kt multicells, 30–40 kt organised with supercells possible, 40 kt and more supercells.
- Precipitable water
- Moist from 20 mm, Very moist from 30, Extreme from 40 for the UK, with RH 0–2 and 2–5 km and the most unstable parcel's mixing ratio.
- Precipitation efficiency
- A rule of thumb from flash-flood forecasting (Davis 2001; Doswell, Brooks and Maddox 1996), one point for each: a warm cloud (ML LCL to 0 °C) 3 km or deeper, so collision and coalescence have time to work; precipitable water 30 mm or more; a humid column, RH 0–2 km ≥ 80% and 2–5 km ≥ 70%, so little evaporates or is entrained dry; skinny CAPE, normalised CAPE (MU CAPE over the LFC to EL depth) under 0.1 m/s², with slower updraughts that hold drops in the warm cloud; and a near moist-adiabatic 700–500 hPa lapse rate, under 6.5 K/km. 2 points Moderate, 4 High, 5 Very high. Efficient rain floods when cells are slow or train over the same ground: check the mean winds.
- SHIP, SCP, STP
- On UK bands, lower than SPC's, as all three are rare here. SHIP yellow from 0.3 (worth a look, and a flag), orange from 0.5 (large hail possible), red from 1 (hail over 50 mm diameter favoured). SCP, by size as negative values mean left-movers: yellow for anything shown above 0.0, orange from 1 (supercells possible, and a flag), red from 4 (favoured). STP, effective layer and fixed layer: yellow from 0.3 (a flag), orange from 0.5 (marginal), red from 1 (significant tornadoes favoured). All three were tuned on US storms: read them as signals, not probabilities.
Parcel levels
- LCL, lifting condensation level
- The level at which a parcel lifted dry adiabatically from its starting point becomes saturated: the base of cloud formed by lifting (cumulus base for a surface parcel). On the skew-T it is Normand's point, where the dry adiabat through the parcel's temperature meets the mixing ratio line through its dew point. Above it the parcel cools at the saturated rate.
- LFC, level of free convection
- The level above which the lifted parcel is warmer (in virtual temperature) than its surroundings, so it rises on its own. Below the LFC the parcel has to be forced up, through any CIN, by a front, hills, convergence or heating; above it, CAPE drives the updraught. No LFC means the parcel never becomes buoyant.
- EL, equilibrium level
- The level, above the LFC, where the rising parcel becomes cooler than its surroundings again and stops gaining buoyancy. It approximates the height of a storm's anvil; the tops of strong updraughts overshoot it towards the Parcel Top.
- PT, Parcel Top
- Where the negative area above the EL equals the CAPE: an upper bound on overshooting tops, ignoring entrainment and drag.
Parcels and CAPE
- Surface, most unstable, mixed layer
- The surface parcel starts from screen level. The most unstable parcel has the highest θe in the lowest 3 km. The mixed-layer parcel takes θ and mixing ratio averaged over the lowest 500 m. CAPE and CIN use virtual temperature.
- LI
- Environment minus parcel temperature at 500 hPa. Negative is unstable.
- Rough UK guide
- Over 100 J/kg can give showers, over 500 J/kg thunderstorms, more organised with shear. CIN below about −50 J/kg needs a trigger to break.
Entraining CAPE (ECAPE)
The CAPE / ECAPE switch above the parcels table (and above Cb tops in Aviation) swaps undiluted CAPE for entraining CAPE in the parcel tables, the Cb tops and the plot. It follows Peters et al. (2023). A rising updraught mixes in the air around it; how much depends on how wide it is, and Peters et al. (2023) tie its width to the storm-relative inflow: strong inflow feeds a wide updraught that keeps most of its CAPE, weak inflow a narrow one that loses much of it, and more when the free troposphere is dry.
- ECAPE
- The paper's ECAPEA (its eq. 51 and section 4), which also counts the kinetic energy the updraught takes from its inflow, so in very strong inflow it can exceed CAPE. w max = √(2 ECAPE). Constants as the paper: σ 1.6, α 0.8, k² 0.18, Pr 1/3, Lmix 120 m.
- Inputs
- CAPE, LFC and EL of each parcel, lifted with virtual temperature as everywhere else. VSR, the mean storm-relative wind speed 0–1 km above ground, with the storm motion chosen in the Storm motion box (RM, LM, Measured or Custom; RM while Measured has no layer). NCAPE, from the LFC to the EL: how far the environment's moist static energy, averaged from the ground up, falls short of its saturated value, a measure of how much entrainment can dilute the updraught.
- User modified
- Once Normand's point has been moved, its parcel uses the profile with the point written in and the air above it mixed (as for the effective inflow layer) for NCAPE and its entraining lift, and VSR over that profile's effective inflow layer; 0–1 km if it has none. The other parcels keep the profile as it is.
- Least updraught radius, 1.5 km
- The formula ties updraught width to the inflow, so as VSR falls towards nothing so do the radius and ECAPE, though storms still grow in weak shear, where heating and the boundary layer set their width. As the paper suggests, the radius is therefore never taken below 1.5 km: below that ECAPE is the paper's eq. 28 with the entrainment rate of a 1.5 km updraught. Such rows are marked ‡.
- Of CAPE, radius, NCAPE
- The fraction of CAPE kept: the paper finds it separates supercell from ordinary-cell environments well. The radius follows from the inflow (eq. 48), and the entrainment rate from the radius, 2 k² Lmix / (Pr R²).
- Entraining parcel
- On the plot (pink), the surface parcel, or Normand's once moved, lifted again while it mixes in its surroundings at that entrainment rate, pseudo-adiabatically: its positive area is shaded and labelled in place of the undiluted parcel's, which stays as a thinner line. This lift follows the later ECAPE-parcel work rather than Peters et al. (2023), so its area is close to ECAPE but not the same number. Its LFC, EL and Parcel Top fill the parcel levels table.
Peters, J.M., Chavas, D.R., Su, C.-Y., Morrison, H. and Coffer, B.E. (2023) ‘An analytic formula for entraining CAPE in mid-latitude storm environments’, Journal of the Atmospheric Sciences, 80(9), pp. 2165–2186. doi:10.1175/JAS-D-23-0003.1.
Shear and helicity
- Bulk wind difference
- The vector difference between the wind at the top and bottom of a layer. 0–6 km above about 40 kt supports organised storms and supercells. "Effective" runs from the surface parcel's base to half its EL.
- Storm-relative helicity (SRH)
- How much the inflow, relative to the storm, turns with height. Shown for Bunkers right and left movers. US guides (0–1 km above 100 m²/s², 0–3 km above 150) are a starting point; UK tornadoes often come with less.
- Effective inflow layer
- The levels whose parcels would feed a storm. Parcels are lifted every 5 hPa from the ground to 500 hPa; the layer runs from the first that passes to the last before one fails. UK limits: CAPE ≥ 50 J/kg and CIN ≥ −25 J/kg, in place of the usual 100 and −250. ESRH is helicity over this layer. Like Normand's point, every parcel's CAPE and CIN use virtual temperature. Once you move Normand's point, the layer is worked out again with its temperature and dew point in place of the profile's at that level. A point warmer than the air above it mixes that air, as daytime heating does: each level up to where the profile is warmer than the point's dry adiabat takes the adiabat, and the point's moisture where that is moister. A cooler or drier point changes its own level only.
- Streamwise vorticity
- Shear gives the air horizontal vorticity, ωh = (−∂v/∂z, ∂u/∂z). Its streamwise part, ωs, is the share along the storm-relative inflow: an updraught tilts that share into rotation about the vertical, the root of a supercell's mesocyclone (Davies-Jones, 1984). SRH is the integral of the storm-relative speed times ωs, so ωs separates how much vorticity there is from how well the inflow lines up with it. The table under the effective inflow layer gives, for the storm motion chosen above the hodograph: the layer's mean ωs (10⁻³ s⁻¹), the streamwise share of all its horizontal vorticity, SRH over the same layer, and the critical angle, between the 0–500 m shear and the storm-relative wind at the ground (Esterheld and Giuliano, 2008). Large, mostly streamwise vorticity in the lowest few hundred metres, and a critical angle near 90°, favour tornadic supercells (Coffer and Parker, 2017; Coffer et al., 2019). The layer is the effective inflow layer, worked out again once Normand's point is moved, or 0–1 km when there is none. Each 100 m slab gives one value from its shear and mean storm-relative wind; coarse model levels or a noisy ascent can make a thin layer jumpy. The ωs plot beside the hodograph shows ωs against height (line) and its streamwise share (dots), with the effective inflow layer shaded.
- Effective inflow slice on the hodograph
- The faint fan from the storm motion to the hodograph, between the effective inflow layer's base and top (dashed), is the area the storm-relative wind sweeps through the layer: half its SRH for that motion (Davies-Jones, Burgess and Foster, 1990). Orange where the wind veers relative to the storm (positive SRH), blue where it backs (negative), as for warm and cold advection. It follows the storm motion chosen and Normand's point once moved; in storm-relative view the fan's point is the centre.
- Hail growth zone
- Where the environment is between −10 and −30 °C. A deep zone with strong updraft and CAPE within it favours larger hail.
Coffer, B.E. and Parker, M.D. (2017) ‘Simulated supercells in nontornadic and tornadic VORTEX2 environments’, Monthly Weather Review, 145(1). doi:10.1175/MWR-D-16-0226.1.
Coffer, B.E., Parker, M.D., Thompson, R.L., Smith, B.T. and Jewell, R.E. (2019) ‘Using near-ground storm relative helicity in supercell tornado forecasting’, Weather and Forecasting, 34(5), pp. 1417–1435.
Davies-Jones, R. (1984) ‘Streamwise vorticity: the origin of updraft rotation in supercell storms’, Journal of the Atmospheric Sciences, 41(20), pp. 2991–3006.
Davies-Jones, R., Burgess, D. and Foster, M. (1990) ‘Test of helicity as a tornado forecast parameter’, Preprints, 16th Conference on Severe Local Storms. Kananaskis Park, AB: American Meteorological Society, pp. 588–592.
Esterheld, J.M. and Giuliano, D.J. (2008) ‘Discriminating between tornadic and non-tornadic supercells: a new hodograph technique’, Electronic Journal of Severe Storms Meteorology, 3(2), pp. 1–50.
Storm motion and storm-relative winds
Choose the storm motion: RM or LM (Bunkers right and left movers, 7.5 m/s either side of the 0–6 km mean wind), Measured (the mean wind of a layer measured on the plot, a steering level) or Custom (direction and speed of your own, for example from radar). SRH for that motion is shown beneath it.
The Ground-relative / Storm-relative switch above the hodograph subtracts the storm motion from every wind on the hodograph and the barbs. On the hodograph the storm is then at the centre and the ground at the small cross; strong storm-relative inflow near the ground and outflow aloft favour long-lived storms.
On the hodograph, markers are kept small and explained in the legend beneath it: a diamond for the steering wind of a measured layer, a triangle for a custom or measured storm motion, and a cross for the ground in storm-relative view.
Plots beside the skew-T
- Hodograph: coloured by height (0–1, 1–3, 3–6, 6–9, 9–12 km). A long, curved hodograph favours rotating storms.
- RH and θe: θe falling with height marks potential instability; dry mid levels favour downbursts.
- w: the derived vertical velocity in ft/min, the unit of GAMETS and other aviation products (m/min with metres). See Data and limitations.
SCP, STP and SHIP follow the SPC's definitions in runs and ascents from November 2026. SCP = (MUCAPE/1000) × (effective SRH/50) × (effective bulk shear/20, 0 below 10 m/s, 1 above 20) × (−40/MUCIN when MUCIN is below −40). STP effective = (MLCAPE/1500) × ((2000 − MLLCL)/1000) × (effective SRH/150) × (effective bulk shear/20, 0 below 12.5 m/s, 1.5 above 30) × ((200 + MLCIN)/150), and 0 when the inflow base is above the ground; STP fixed layer uses the surface parcel, 0–1 km SRH and 0–6 km shear. SHIP = MUCAPE × mixing ratio (11–13.6 g/kg) × 700–500 hPa lapse rate × −T500 (at least 5.5) × 0–6 km shear (7–27 m/s) / 42,000,000, cut for CAPE under 1300 J/kg, a lapse rate under 5.8 K/km and a freezing level under 2400 m. The effective inflow layer (Thompson et al. 2007) is every level whose parcel has CAPE of 100 J/kg or more and CIN of no more than 250. One difference remains: the mixed-layer parcel here is the lowest 500 m, the SPC's the lowest 100 hPa. All three were tuned on US storms; use them as a guide, not a threshold.
Lee Waves
Lee waves form when stable air flows over hills and the wave energy is trapped below a layer where the Scorer parameter falls sharply.
- Scorer parameter l²
- N²/U² minus a curvature term (U''/U): stability over wind speed squared. It is worked out on a 250 m grid above the ground.
- Scorer's trapping test
- Trapping is favoured where the mean l² falls by at least π²/4H² across a layer top at height H, tested from 1.5 to 5 km, and the mean 0.5–3 km wind is at least 15 kt. 2π/l in the lower layer gives a scale for the wavelength.
- Wind with height
- Waves need a fairly strong flow across the ridge, keeping its direction and increasing with height. A reversal or a big drop in speed aloft tends to absorb waves instead.
- Vertical velocity
- The strongest derived w in the box around the point and its level, in ft/min (m/min with metres), badged like the lee waves' maximum w. UKV at 2 km resolves the longer lee waves but not short ones.
- Strip at the right of the plot
- The turbulence column from Aviation's hazard strip, in the same colours: Turbulence where the Richardson number is low (orange, deeper for Likely), with the green outline beside it marking the trapped lee wave layer, where the leading wave's up- and downdraughts are at least 70% of their strongest. Hover or tap a block for its heights; the key is under the plot.
Turbulence and the hazard strip
The Turbulence: Richardson number table is the same as in Aviation (see there for the method), with heights above ground as elsewhere on this tab, and Possible (amber) or Likely (orange) badges. The T column at the right of the skew-T shows the same layers, with the trapped lee wave's layer of strongest up- and downdraughts outlined beside it in green. Scorer's "Trapping favoured" is badged amber when yes.
Trapped lee waves: Shutts' Lee Wave Finder
This follows G. Shutts (1997), Operational lee wave forecasting, Meteorological Applications 4, 23–35, the method behind the Met Office's Lee Wave Finder, which replaced Casswell's two-layer method.
- The equation
- The vertical structure equation for steady, small-amplitude gravity waves, ŵ'' + (l² − k²) ŵ = 0, is solved as an eigenvalue problem for the horizontal wavenumber k. Temperature and wind are put on a 100 m grid from the ground (by monotone cubic interpolation, so that the wind's curvature has no spikes at the data levels), smoothed with the paper's three-point filter ten times, and l² = g/T (∂T/∂z + g/cp)/U² − U''/U is smoothed again.
- Boundaries
- ŵ = 0 at the ground. At the top, a radiation condition: a wave either dies away above or carries energy upward, so a mode that leaks slowly is found as well as one that is fully trapped. If the wind across the waves falls to zero (a critical level), the top is placed just below it with the Booker–Bretherton condition there; below 2.5 km that ends the search in that direction.
- Directions
- Waves need not face straight into the wind, so the problem is solved in five directions, 15° apart, within 30° of the wind at the 1 km launching height. The direction is given as a wind direction: the waves face air coming from it, and their crests lie across it.
- Which modes count
- A mode is kept if it is well trapped (the real part of k² at least six times the imaginary part) and low (the centre of mass of ŵ² below 6 km). Up to three distinct waves are listed. The first is the paper's choice: the lowest centre of mass, unless the next lowest is closer to the launching wind, in which case that is likely the more strongly trapped.
- Amplitude
- The equation fixes the shape, not the size. Following Shutts, the vertical velocity at the 1 km launching height is set to W₀ = γ k0.2 U₀, with γ = 0.55 and U₀ the wind across the waves at 1 km, from an assumed power-law spectrum of hill heights. γ was tuned on a region of complex terrain about 200 km across (the Lake District), so it is not right for an isolated hill. ŵ grows with height as (ρ₀/ρ)½.
- Results
- For each wave: the wavelength (2π/k), the direction, the largest vertical velocity and the height at which it occurs (the elevation of the maximum amplitude). Vertical velocity is in ft/min, with m/s alongside, for feet and flight levels; in m/s, with m/min alongside, for metres. The maximum w is badged against the thresholds forecasters use for wave charts: amber from 500 ft/min (the least that goes on the F415), orange from 750 ft/min and red from 1000 ft/min (2.5, 3.8 and 5.1 m/s); the dominant wave's row stays highlighted. The Lee waves plot shows how the amplitude varies with height for each wave; a node between two maxima marks a reversal of phase.
- Limits
- The theory is linear: it cannot describe wave breaking, rotors or downslope windstorms, and it ignores the boundary layer and latent heat in wave cloud. Results are sensitive to the smoothing of the wind, and over complex terrain several wavelengths are excited at once. In the paper's cases the predicted amplitudes were close to those observed (2.6 m/s over the Lake District; 5.6 to 6.4 m/s for the Sheffield gale of 1962, when a pilot reported 2000 ft/min).
Winter
Rain or snow, how wet the snow is, and freezing rain and drizzle, for the ground you choose. On the plot: the wet-bulb curve, the dendritic growth zone (−12 to −17 °C) bracketed at the left, Bourgouin's melting (orange) and refreezing (blue) areas with their size in J/kg, a dashed green line at the ground you set, and a dashed dark blue line at the 0 °C wet-bulb freezing level, labelled WBFL with its height above that ground (or as a flight level), at the right like the parcel's levels.
Ground height
Right under the graph. It starts at the profile's own ground (the UKV's smoothed terrain, or the station). Type a hill or valley height in metres above sea level and everything below it is worked out for that ground: the chance-of-snow table (Boyden's thickness, the 0 °C height in hPa and the ground temperature), the heights above ground, the snow ratio, the precipitation type and freezing drizzle. Above the profile's ground the column is cut at the new height. Below it, the bottom level is carried down at 6.5 K/km with the same dew-point depression, only a stand-in for air the profile does not hold. Use profile's goes back.
Winter summary. The table under the ground height gathers the answers: the precipitation type at the ground (Bourgouin), the chance of snow (the mean of the four Lowndes predictors, with their range), the 0 °C wet-bulb height above ground with its Table 5.9 category, Kuchera's snow-to-liquid ratio, wet or dry snow, and the freezing rain and freezing drizzle risks. Open Diagnostics details for the tables behind them, described below.
- Chance of snow (Lowndes et al. 1974, Table 5.8)
- For each predictor, the chance that precipitation reaching the ground is snow. The column nearest the current value is highlighted. Use the predictors together.
- Boyden's adjusted thickness
- The 1000–850 hPa thickness adjusted for the ground's height, (H1000 − HGR)/30.
- Height of the 0 °C wet-bulb above ground (Table 5.9)
- Above the ground set, not above sea level, so it falls as the ground rises. Below about 1000 ft, mostly snow; 1000–2000 ft, persistent rain readily turns to snow; above 3000 ft, almost always rain. UKV's own wet-bulb freezing level is shown for comparison, moved to the same ground.
- Dendritic growth zone
- Where snowflakes grow fastest. Ascent and high humidity in the zone favour heavy snow. Mean RH is over water: saturation over ice at −15 °C reads about 86%.
Snow-to-liquid ratio
How deep the fresh snow is for each depth of water: 10:1 turns 1 mm of water into 10 mm (1 cm) of snow. Kuchera's method, from the NOAA Forecast Operations Guide, worked out from the ground set:
- Kuchera
- From Tmax, the warmest temperature between the ground and 500 hPa: 12 + 2 (271.16 − Tmax) when Tmax is above 271.16 K (−2 °C), otherwise 12 + (271.16 − Tmax). So 12:1 at −2 °C, 8:1 at 0 °C, 16:1 at −6 °C. It assumes dendrites, so it can run high in very cold, windy air, where flakes break up, and where columns and plates grow instead.
- Wet or dry
- From Kuchera's ratio. Wet, below 8:1 (always so with the ground at +0.5 °C or more, since Tmax is never colder than the ground): heavy and sticky, loading wires and trees. Moist, 8–12:1. Dry, 12–18:1: light, blowing about in a fresh wind. Very dry, 18:1 and above: powder that drifts.
It does not allow for compaction, melting on warm ground or sublimation in dry air. UK snow often falls near 0 °C, where it gives about 8–10:1. Cobb and Waldstreicher's (2005) method was tried and dropped: weighted by the strongest lift, it gave ratios near 30:1 and more for UK model profiles, far outside UK experience. If the type below is not snow, the ratio is shown only as a guide.
Graupel in showers
A forecaster's rule of thumb in the Winter summary, not a calibrated probability. Possible (amber) when showers are deep and cold enough for plenty of ice and riming, the most unstable parcel's Parcel Top at −20 °C or colder with at least 25 J/kg of CAPE, and the wet-bulb freezing level is below 1500 m above the ground set, so small graupel reaches the ground before it melts. Likely (orange) with a top at −30 °C or colder, 100 J/kg or more and the freezing level below 1000 m. The temperature is the environment's at the Parcel Top; a top above the profile counts as cold enough.
Precipitation type (Bourgouin 2000)
What reaches the ground if anything falls, from the area between the temperature curve and the 0 °C isotherm (Rd ∫ (T − 0 °C) d ln p, in J/kg, the same as the area on a tephigram). A warm area aloft melts falling snow; a cold area under it refreezes the drops.
- Warm aloft, cold at the ground: under 2 J/kg of melting, snow. Otherwise, refreezing of more than 66 + 0.66 × melting gives ice pellets, less than 46 + 0.66 × melting gives freezing rain, and between them both. Under 13.2 J/kg of melting the snow is only partly melted.
- Warm at the ground only: under 5.6 J/kg, snow; 5.6 to 13.2, rain and snow; above 13.2, rain.
- Warm at the ground with a warm layer aloft: pellets refrozen in the cold layer between melt again by the surface layer's rule.
The column is read up to 500 hPa above the ground, and only the lowest warm layer aloft is used. Freezing rain risk: high for freezing rain, moderate with ice pellets, low for ice pellets after melting aloft, or for rain on ground just above 0 °C, where cold surfaces may still freeze it.
Freezing drizzle
Most freezing drizzle does not come from melted snow: it forms in shallow supercooled cloud, where droplets grow by collision because too few ice crystals form to take the water (Rauber et al. 2000; Cobb and Waldstreicher 2005 note snow grains below −10 °C cloud tops and freezing drizzle above −8 °C). The table ticks off the ingredients in the lowest cloud (relative humidity at least 90%):
- Needed: ground at or below 0 °C, and the cloud's base within 1500 m of it.
- Cloud top warmer than −10 °C, so the cloud stays liquid. −10 to −12 °C is marginal, and caps the risk at low.
- Adding to it: a cloud at least 300 m deep; no ice cloud (−10 °C or colder, saturated over ice) within 3 km above to seed it with ice; and ascent in the cloud (UKV points only).
High with every ingredient, moderate with one missing, low otherwise. It is a guide from the ingredients, not a probability.
Bourgouin, P. (2000) ‘A method to determine precipitation types’, Weather and Forecasting, 15(5), pp. 583–592. doi:10.1175/1520-0434(2000)015<0583:AMTDPT>2.0.CO;2.
Cobb, D.K. and Waldstreicher, J.S. (2005) ‘A simple physically based snowfall algorithm’, 21st Conference on Weather Analysis and Forecasting, Washington DC. American Meteorological Society, paper 2A.2.
NOAA (n.d.) ‘Numerical methods for determining snow accumulation’, Forecast Operations Guide, NOAA Virtual Lab. Available at: https://vlab.noaa.gov/web/forecast-guide/fog?page=numerical-methods-for-determining-snow-accumulation.
Rauber, R.M., Olthoff, L.S., Ramamurthy, M.K. and Kunkel, K.E. (2000) ‘The relative importance of warm rain and melting processes in freezing precipitation events’, Journal of Applied Meteorology, 39(7), pp. 1185–1195.
Fog
Fog mode opens with no construction on. Turn on Saunders' fog point with key f or the second button under Constructions, then drag the handles on the surface isobar to set the afternoon maximum and dew point.
- Fog point (Saunders)
- The temperature at which fog forms overnight, found from the maximum, the dew point and the type of sounding (I, II, IIIa, IIIb, chosen automatically unless you pick one). Fog is likely if the night's minimum falls to the fog point. Labels A, B, C and Tc follow the original figure.
- Cross-checks
- The screen dew point less 2 °C, and the dew point 60 hPa above the ground.
- Forecast for comparison
- UKV's minimum over the next 18 hours and maximum for the day, the screen dew-point depression (T − Td) at the ascent time, screen visibility, fog fraction and 10 m wind. Badges in the risk colours: dew-point depression amber at 0.5 °C or less, orange at 0.25 °C or less, red at 0 °C; RAF / NATO colour states, in their own colours:
The visibility state is from UKV's screen visibility. The cloud base state is from the lowest height above ground where UKV's cloud fraction on its height levels reaches 3/8 (37.5%), interpolated between levels and always in feet above ground; "none below" means no level up to 6 km reaches it. A level's cloud fraction is not quite the oktas an observer counts (they add up the layers below), so the base can sit a little high where broken cloud lies in several thin layers. The overall colour state, larger and outlined, is the poorer of the two. UKV points only; fog fraction amber from 10%, orange from 40%, red from 70%.State Visibility Cloud base, 3/8 or more BLU 8000 m 2500 ft WHT 5000 m 1500 ft GRN 3700 m 700 ft YLO1 2500 m 500 ft YLO2 1600 m 300 ft AMB 800 m 200 ft RED below below - Wind and cloud
- Light winds and clear skies favour radiation fog; stronger winds tend to give low stratus instead.
UKV humidity is on pressure levels only, so the construction reads point B from those levels (see Data and limitations).
Temperatures
For how warm or cold it gets. On the plot, dashed purple lines bring the 925 and 850 hPa temperatures down the dry adiabat to the ground, with their values.
- Summary
- Under the plot: UKV's maximum and minimum, the screen values now, and how they feel, each with a badge: blues for cold (deeper for colder), amber, orange and red for heat, grey when there is nothing to flag. The maximum is badged on UK heat bands: amber from 27 °C, orange from 32 °C, red from 37 °C, and blue below 0 °C (an ice day). The minimum is badged by frost: air frost below 0 °C, hard frost from −5 °C, severe from −10 °C. The wet bulb is badged amber from 21 °C, orange from 24 °C and red from 28 °C, where cooling by sweating starts to fail. Open Diagnostics details for the tables below.
- Maximum and minimum
- UKV's own maximum for the UTC day and its lowest minimum in the next 18 hours, and the 925 and 850 hPa temperatures mixed down dry adiabatically. 925 hPa suits a short winter day or a shallow mixed layer; 850 hPa a long summer day with deep mixing. Cloud, a sea breeze or a strong inversion all keep the ground cooler.
- Feels like
- From the screen temperature and dew point and the 10 m wind, with no sunshine allowed for (it adds several degrees in strong sun):
- Apparent temperature (Steadman 1994, as the Australian Bureau of Meteorology uses it): T + 0.33 e − 0.70 V − 4.00, with e the vapour pressure in hPa and V the 10 m wind in m/s. It includes humidity and wind at any temperature, but its wind term (0.70 °C per m/s) runs far colder than wind chill in strong wind.
- Wind chill (JAG/TI 2001, used in Canada and the US, and the basis of most "feels like" in the cold): 13.12 + 0.6215 T − 11.37 V0.16 + 0.3965 T V0.16, V in km/h, at 10 °C or below with at least 4.8 km/h of wind. Environment Canada's bands: below −10 moderate, below −28 high (frostbite in 10–30 minutes), below −40 very high and worse.
- Heat index (US NWS: Steadman 1979, Rothfusz's regression and the NWS adjustments): from 27 °C. Caution 27, extreme caution 32, danger 41, extreme danger 54.
- Humidex (Environment Canada): T + 0.5555 (e − 10), with e from the dew point; from 20 °C. Some discomfort 30, great discomfort 40, dangerous 46.
- Wet bulb at the screen, the lowest temperature sweating can cool to, and WBGT in the shade, estimated from T and humidity as the Bureau of Meteorology does: 0.567 T + 0.393 e + 3.94. A measured WBGT needs the globe temperature, so in sun it runs higher.
- Level temperature
- On the standard pressure levels.
- Thickness
- Proportional to the layer's mean temperature: 1000–500, 1000–850 and 850–500 hPa, with the height of the 1000 hPa surface.
- Near the ground
- The mean temperature of the lowest 100 hPa and the 0 °C isotherm above the profile's ground.
- Plots
- Temperature and dew point, or potential temperature, which is upright where the air is well mixed and rises sharply at an inversion.
Aviation
Hazards to flight from the one ascent: Cb tops, icing, turbulence and contrails. Heights in this mode's tables are above sea level, as altitudes and flight levels are; choose feet, flight levels or metres with key u. Everything is worked out in the browser on a regular grid of heights, the same way for the model and the radiosonde. The map shows MSLP and precipitation rate, as in Convection.
Aviation summary
Under the plot, one row per hazard: a badge for the worst category, coloured as the Winter tab's risks (grey none, amber, orange, red), and the layers where it is, above sea level. Cb: None with no CAPE (or ECAPE with the switch on), Weak under 100 J/kg, Moderate 100–1000 J/kg, Strong 1000 J/kg and more, with the Parcel Top of the parcel with most. Icing: Light, Moderate or Severe. Turbulence: Possible or Likely. Lee waves: Trapped, or Steep where they may break or form rotors. Contrails (purple, as on the strip): Short-lived, outlined, or Persistent, filled. The last row gives the 0 °C level and the tropopause. Open Diagnostics details for the tables described below.
On the plot
- Hazard strip at the right of the plot: a column each for Icing (blue, deeper for Moderate and Severe), Turbulence (orange, deeper for Likely) and Contrails (purple, deeper for persistent), with the letters above the plot and a key under it. The green outline beside the turbulence column is the trapped lee wave layer: where the leading wave's up- and downdraughts are at least 70% of their strongest, so where wave turbulence and the largest height changes for aircraft are most likely; rotors, if any, form below it near the ground. Hover or tap a block for its heights.
- Contrail lines, labelled along the line like the 0 and −20 °C isotherms: Cont Sat Air (dashed), the Schmidt–Appleman threshold in air saturated over water, and Cont Unsat Air (dotted), the threshold in completely dry air. Where the temperature curve is left of Cont Unsat Air, contrails form whatever the humidity; right of Cont Sat Air, never; between the two, only if the air is moist enough. The labels move below the readouts in the corner when Normand's point or the measuring tool is on.
- Cb top: the most unstable parcel's Parcel Top, when it has at least 25 J/kg of CAPE. With ECAPE on, its entraining Parcel Top and ECAPE.
Cb tops
The Parcel Top is where a parcel rising through its EL has used up its CAPE in overshoot, so about where an active Cb's top would reach; the EL is nearer where an anvil spreads. Rows for the most unstable, surface and mixed-layer parcels, and for Normand's parcel once you move it. The line under the table reads the parcel with the most CAPE. The model's CAPE is only as good as its humidity, which is on pressure levels only.
The CAPE / ECAPE switch, shared with Convection, lifts each parcel while it entrains, as described under Entraining CAPE. Diluted parcels stop lower, so ECAPE's tops are usually lower, most in weak storm-relative inflow and dry mid levels, and nearer what ordinary showers reach.
Icing
Layers between 0 and −20 °C that are in cloud, in 100 m steps. For the UKV the model's own cloud fraction is used, up to 6 km above ground where it stops, so humid but cloud-free air is not flagged; a radiosonde has no cloud data, and there, and in the UKV above 6 km, relative humidity over water stands in. The model's cloud fraction counts ice falling into dry air, which holds no supercooled water, so the air must also be humid. Each step gets a score: cloud (0 at 5% cloud fraction, 1 at 50%, times humidity, 0 at 60%, 1 at 80%; or, with no cloud data, humidity alone, 0 at 75%, 1 at 95%) × temperature (1 from −3 to −12 °C, where supercooled water is most common, falling to 0.5 at 0 and −20 °C) × vertical motion (1 in still air, up to 2 with 0.4 m/s of ascent, 0.7 in descent). Light above 0, Moderate from 0.6, Severe from 1.1, which needs saturation and ascent together. It is a guide to where supercooled water is likely, in the spirit of the usual temperature and humidity rules, not a calibrated icing algorithm: it knows nothing of droplet size, and freezing rain under a warm nose is not included.
Turbulence
The Richardson number, Ri = N²/S², the stability (from the dry potential temperature) against the square of the shear, over 500 m layers every 250 m from 300 m above ground. Layers are flagged where the shear is at least 3 kt per 1000 ft and Ri is below 1 (Possible) or below 0.5 (Likely). Kelvin–Helmholtz billows need Ri below 0.25 in thin layers that model and sonde levels smooth out, so the thresholds are raised to match. Negative Ri is a layer unstable in itself: in the lowest kilometre or two by day, that is the convective boundary layer. Mechanical turbulence near the ground and convective turbulence in cloud are not assessed.
Lee waves: Shutts' finder, as in Lee Waves. The leading trapped wave's layer (where its w is at least 70% of its peak) is outlined beside the turbulence column, with its wavelength and strongest w under the table. A wave flagged as steep may break or form rotors.
Contrails
The Schmidt–Appleman criterion (Schumann 1996). Exhaust mixing into the air follows a straight line on a vapour pressure–temperature diagram, with slope set by the pressure and the engine's propulsive efficiency (0.3 here, a modern turbofan; older engines need colder air). A contrail forms where that line reaches saturation over water, which happens below a threshold temperature: at 250 hPa, about −42 °C in saturated air (Cont Sat Air) and −51 °C in dry air (Cont Unsat Air). It is persistent where the air is also saturated over ice, so the trail can grow into cirrus; otherwise short-lived. The Met Office's mintra line is an older form of the same idea. Humidity high up is the weak point. UKV relative humidity is over water at all temperatures (in cloud below about −10 °C it settles at ice saturation, 70–90% over water), so over ice it can pass 100%; the model holds ice supersaturation up to about 145–150% near −45 °C, where droplets freeze on their own. Radiosonde humidity sensors struggle in very cold, dry air.
Plots and through the run
Beside the plot: the Richardson number with height, and humidity over ice and over water. Through the run's time–height section shades icing risk by default; Indices shows Cb tops, the lowest base and highest top of any icing, the depth of low-Ri layers (shaded where a trapped lee wave is found) and the lowest level at which contrails form.
Tools on the plot
- The buttons
- Under Constructions: the first icon is Normand's construction, the second Saunders' fog point, the ruler the measuring tool. Reset view returns to the whole ascent.
- Normand's point (key n, any mode)
- Lift a parcel from any level: drag the red T handle and the green Td handle, up or down to change the starting level. The readout in the top right corner of the plot gives T, Td and the wet-bulb Tw, the LCL (Normand's point), LFC, EL, Parcel Top, and CAPE and CIN. CAPE and CIN use virtual temperature, so the shading sits a little warmer than the T and parcel lines. Reset returns it to the surface. A new location (a map click, a place search, a station, a pin or an upload), a new time or a new run snaps it back to the new profile's surface T and Td, even once moved. Once a handle has been moved, the Convection summary and the Parcels and Parcel levels tables under Diagnostics details gain a User modified row for this parcel, and the Bulk shear table an Effective, user modified row (from the parcel's start to half way to its EL), worked out in the browser the same way as the others.
- Fog point (key f, any mode)
- Saunders' construction on the surface isobar; Fog mode's tables use it once it is on. A new location, time or run snaps it to the new profile's screen T and Td.
- Measure (key m)
- Drag on the plot from one level to another; drag either edge to adjust. Each edge shows its pressure at the left and its height as a tag on the height axis at the right, in the axis's units (above sea level, or FL); the readout gives heights above ground. The readout gives:
- Mean: the average wind through the layer, as u and v at every 100 m from bottom to top. Each 100 m counts equally: the mean is weighted by height, not by mass or density, so the thinner air near the top counts as much as the denser air near the bottom. It can be taken as the steering flow for a cell occupying that part of the profile. It is drawn on the hodograph as the steering level, and Measured under Storm motion uses it as the storm motion.
- Shear: the bulk wind difference across the layer, the length of the vector from the bottom wind to the top wind. Only the two ends count, so a wind that turns one way and back again between them can show little shear.
- EBWD, effective bulk wind difference: a parcel is lifted from the bottom of the layer with the profile's temperature and dew point there (in virtual temperature, as Normand's point). If it has CAPE, EBWD is the vector difference between the wind at the bottom of the layer and the wind half way up to that parcel's EL. The depth comes from the parcel, not from the top of the layer, so EBWD can be larger or smaller than the layer's own shear; with no CAPE it reads "no CAPE". It does not use Normand's point. For the SPC's effective bulk shear, put the bottom of the layer at the base of the effective inflow layer (the bracket at the left in Convection), where the most unstable parcel usually starts.
- RH: mean relative humidity over water, by mass (even steps of pressure), where the profile has humidity.
- Left strip (key a)
- Chosen with the switch just above it. w: derived vertical velocity in ft/s (m/min with metres), filled at the point, open the strongest within about 0.5°. T adv: thermal advection inferred from the thermal wind in 50 hPa layers; veering with height is warm (red), backing cold (blue). It assumes geostrophic balance, so the lowest kilometre is faded and it is unreliable near fronts and convection.
- Previous run
- The button beside Reset view overlays the run before the selected one, at the same place and valid time, in grey: how has the forecast changed? It joins the barbs (its own column), the hodograph and the comparison table like a pin, and stays on as you move through places and times until you turn it off. It needs an earlier run that covers the valid time; with the oldest run, or past the end of the earlier run, the status line says so.
- Surface ob
- Beside Export: fit the ascent shown to a surface observation. Paste a METAR or a SYNOP (AAXX, section 1) and the boxes fill in, or type the temperature, dew point and wind; an empty box keeps the ascent's own value, shown faintly. Warmer than the ascent at the screen: the boundary layer is taken as mixed, so the air goes onto the dry adiabat from the observed temperature up to where it meets the ascent (the afternoon maximum-temperature modification), with the observed mixing ratio mixed through the same layer. As cold or colder: the change at the screen fades out over the lowest 50 hPa, as a surface inversion, and the dew point likewise. The wind replaces the 10 m wind only. Every index and table is then worked out again in your browser, the ascent as it was is drawn dashed, and a note above the tables gives the main changes, with Back to the model to undo it. The ob stays with that place and time: step away and back and it returns. Nothing is sent anywhere.
- Export PNG
- Saves the plot in the theme on screen, with the graph shown or all graphs in the mode, and the diagnostics if ticked.
Data and limitations
- UKV open data
- Met Office UKV, 2 km, from the AWS open data archive: free and unsupported, not for critical use. Each point takes the nearest 2 km model column; nothing is interpolated between grid cells. Points are every 0.5° over the whole domain and every 0.25° over land and within about 0.1° of the coast.
- Levels
- Temperature and wind come from height levels near the ground and pressure levels above (1000, 975, 950, 925, 900, 850 hPa and every 50 or 25 hPa to 100 hPa). The plot stops at 100 hPa; above that the air is taken as isothermal for Parcel Top. Runs built before mid-November 2026 stop at 150 hPa and have no spread.
- Spread around a point
- The lines and every diagnostic come from the nearest 2 km column alone. The Spread button adds the lowest and highest values in the 5 × 5 block of model cells around it (about 10 km across): pale bands behind temperature and dew point on the skew-T, and ellipses on the hodograph. A narrow band means the column speaks for its neighbourhood; a wide one means a front, a shower or the coast is close and a small shift in place would change the sounding. It is the spread across neighbouring cells of one run, not an ensemble.
- Humidity on standard levels only
- The open data has relative humidity only at screen level and on the pressure levels, not on the height levels. The dew point between levels is interpolated by mixing ratio in log pressure, so thin moist or dry layers, and the sharp drying at an inversion top, are smoothed out. The dew point is also capped at the temperature, so it never shows supersaturation.
- Vertical velocity is derived
- The open data has no w field. It is worked out from continuity: the divergence of the pressure-level winds is integrated up from the ground, the slope of the pressure surface is added (important over hills), and the residual at the top is removed (O'Brien). Winds are smoothed over about 10 km first because direction is stored in whole degrees. It is an estimate and has not been checked against the model's own w.
- Thermal advection
- Inferred in the browser from how the wind turns with height (the thermal wind), assuming geostrophic balance. Friction, fronts and convection all break that assumption.
- Radiosondes
- From the University of Wyoming, at their reported levels; an ascent Wyoming never received is filled in two or three days later from NOAA's IGRA2 archive. Ascents are kept for 30 days. They have no vertical velocity, and humidity sensors become unreliable in very cold, dry air high up.
- Uploaded ascents
- Read in the browser and never sent to the server. The levels are thinned and gaps in humidity filled as for a radiosonde, and the indices come from
indices.js, a copy of the pipeline's code that the tests check against it on example files; parcels can differ from the pipeline's by a few J/kg, as its moist adiabats come from a table. A WMO number is placed from NOAA's IGRA2 station list. - Indices
- The tables come from the same code for model and radiosonde. The effective inflow layer, growth zones, measuring tool and storm-relative values are worked out in the browser. Many thresholds quoted are US-derived and only a rough guide in the UK.
UKV: © Crown copyright, Met Office, CC BY-SA 4.0. Ascents: University of Wyoming and NOAA NCEI IGRA2. Place names: GeoNames, CC BY 4.0. Station list for uploads: NOAA NCEI IGRA2. Created by Nicholas Silkstone; to cite the viewer, see Getting started.