Skew-T Explorer

A Skew-T looks intimidating; it isn't. It's just a weather report turned on its side: the bottom of the chart is the runway, the top is airline cruise altitude, and everything in between is what the air is doing as you climb.

Ignore the busy background for a moment. Only two lines carry data:

• Red — temperature as you go up.
• Green — dewpoint, i.e. how much moisture is up there.

And one rule: where the two lines squeeze together, there's cloud; where they spread apart, the air is dry. That single idea is most of reading a Skew-T. The feathers on the right are wind at each height (full barb = 10 kt, flag = 50 kt).

The busy background? Just graph paper.

The skewed blue lines mark temperature (the 0 °C one is your freezing level), the brown and green curves are the paths rising air cools along, and the dashed purple lines track moisture. You never read them directly — they're rails that the interesting lines get compared against. Hover the diagram in section 4 and the boxes under it will do the comparing for you.

One bubble of air tells the future

Take air from the surface and imagine lifting it. It cools as it rises; where it gets as cool as its dewpoint, a cloud forms — that height is a good guess at the day's cumulus cloud base. Keep lifting: if the bubble ends up warmer than the air around it, it floats up on its own — that's CAPE, storm fuel. If it's colder, it sinks back — that's CIN, the lid that keeps a hot humid day quiet until it suddenly isn't.

Five things a pilot pulls off it

• Cloud bases — where red and green pinch together.
• Bumps or smooth — steep red line = thermals and bumps; red line leaning right (inversion) = smooth on top, haze below.
• Icing — moisture where the temperature sits between 0 and −20 °C.
• Freezing level — where the red line crosses the 0 °C line.
• Wind shear — barbs that twist or grow suddenly over a short height.

Two ways to get a sounding: a real balloon (section 2, twice a day at fixed sites) or a model's virtual column (section 4, any airport, any hour). Section 3 explains how the first becomes the second.

Twice a day, at 00Z and 12Z, about 92 U.S. National Weather Service offices release weather balloons. Each radiosonde climbs at roughly 1,000 ft/min for about two hours, radioing back pressure, temperature, and humidity while GPS drift gives the winds — one true measured column of atmosphere, up to ~100,000 ft. The SPC chart above is that raw ascent drawn by their SHARP analysis package, with the severe-weather math laid out around it. Our nearest launch site is Sterling VA (IAD), about 40 nm west of Annapolis.

Assimilation: the balloon teaches the model

Those observations don't stay on a chart. Every RAOB — along with satellite radiances, aircraft reports, surface stations, and radar — is fed into data assimilation, the step where a model's previous forecast is nudged toward what was actually measured. Radiosondes are disproportionately valuable here: they are one of the few direct, full-column truth measurements, so they anchor the satellite data that fills the gaps between launch sites. The corrected 3-D snapshot (the analysis) becomes the starting point from which the GFS, HRRR, ICON and friends integrate the equations of motion forward in time.

From grid to the diagram at the top of this page

The model's output is a grid — for the GFS, roughly 13 nm between points; for the HRRR, about 1.6 nm — with values stored at standard pressure levels (1000, 975, 950 … 100 hPa). When you type an airport code in section 4 below, the page looks up its coordinates, and Open-Meteo pulls the model column nearest that point: temperature, humidity, wind, and geopotential height at each level. This page then reconstructs dewpoint from humidity, lifts a surface parcel, and draws the same kind of diagram SPC draws — except the "balloon" is virtual, it exists everywhere, and it exists for tomorrow too.

Why they disagree — and which to trust

• Resolution — the balloon reports every few seconds of ascent and captures razor-thin inversions and moist layers; the model column is smoothed onto ~21 levels here. A 300-ft-thick freezing-drizzle layer can be invisible in the model sounding.
• Time and place — the RAOB is truth, but only at 00Z/12Z and only at the launch site (and the balloon drifts downwind as it climbs). The model column is an estimate, but for any hour and any point.
• Drift with lead time — a model sounding valid an hour from now is usually excellent; at 48 h, timing errors in fronts and moisture grow. Comparing the 12Z IAD balloon against the model's 12Z column (set the slider in section 4 to the same hour) is the honest way to judge how much to trust today's forecast soundings.

Good habit: check the morning IAD balloon here, note where the inversions and moisture really are, then watch how the forecast soundings evolve them through the day.