Atmospheric Stability & Skew-T Lab · Guide · glidepathscience.org
GuideAtmospheric Stability & Skew-T Lab
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a meteorology course teaches them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about twenty minutes to read. Every worked example uses numbers that are not one of the sim's own problems, and every sounding in the figures is constructed, not observed.
Part 1 · The essentials
Math you’ll need: algebra, rates in °C per kilometer, square roots and natural logarithms; calculus only in a box marked “With calculus.”
A sounding on a skew-T log-P diagram
Twice a day, at hundreds of stations, a weather balloon carries a radiosonde up through the atmosphere. Its readings of pressure, temperature and humidity make a sounding, which forecasters read on a diagram built so that the physics of rising air can be done with a pencil: in North America, the skew-T log-P diagram.
Pressure falls upward from 1,000 hPa near the bottom to 100 hPa, about 16 km up, on a logarithmic scale that makes equal steps on the page roughly equal steps of height. The lines of equal temperature, the isotherms, lean 45° up and to the right: every point is shifted right by as much as it is high on the page, and that shift is the skew. The red line is the temperature T, the green line the dewpoint $T_d$, the temperature to which the air would have to be cooled to saturate. Behind them run five families of lines: isobars, isotherms, tan dry adiabats for unsaturated rising air, dashed teal moist adiabats for saturated air, and dotted green mixing-ratio lines of constant water vapor.
The usual mistake is to read a temperature straight down to the bottom axis, as on an ordinary graph: because the isotherms lean, that is too warm, and more so the higher the point.
The first thing to read from a sounding is how fast its air cools with height.
The environment's lapse rate
How fast the temperature falls with height decides whether a lifted parcel keeps rising and whether a layer is a lid. It is the lapse rate, with a minus sign so that cooling with height counts as positive:
$\Gamma = -\Delta T/\Delta z$
Here ΔT is the change of temperature across a layer and Δz its thickness. The ICAO standard atmosphere cools 6.5 °C/km up to 11 km. Real soundings range from nearly 10 °C/km in a sunlit afternoon layer to less than zero in an inversion, where the air warms with height. Two slips are common: the sign, and dividing by a thickness in meters, which gives a rate a thousand times smaller than the °C/km a forecaster quotes.
That is how the air around a parcel cools; next, the parcel itself.
Rising air cools: the dry adiabat
Rising air cools without losing heat. A parcel, a blob of air with one temperature, expands as it is lifted into lower pressure, and the work it does on the air around it comes out of its internal energy. With no heat exchanged (an adiabatic process) and no cloud forming, the first law and hydrostatic balance give a fixed cooling rate, the dry adiabatic lapse rate:
$\Gamma_d = g/c_p \approx 9.8$ °C/km $T = T_0 - \Gamma_d\,\Delta z$
Here $c_p$ is the specific heat of dry air at constant pressure (1,005.7 J kg⁻¹ K⁻¹, giving 9.75 °C/km), $T_0$ the starting temperature and Δz the rise. The parcel follows a tan dry adiabat; a sinking parcel warms at the same rate. The usual mistake is to use the environment's lapse rate, often the standard 6.5 °C/km, for the parcel: the parcel's rate depends only on g and $c_p$.
Its water vapor, meanwhile, does not change at all.
What a rising parcel keeps
Until cloud forms, a rising parcel keeps its mixing ratio r, the mass of water vapor per kilogram of dry air, set by the vapor's partial pressure e and the total pressure p:
$r = \varepsilon e/(p - e)$ $\varepsilon = 0.622$
where ε is the ratio of the molar masses of water and dry air. The vapor pressure is the saturation vapor pressure at the dewpoint, which a table gives (12.3 hPa at 10 °C). An unsaturated parcel's dewpoint slides along a mixing-ratio line as it rises. The dewpoint itself is not conserved: as p falls, so does e, and the dewpoint drops about 1.8 °C/km. The relative humidity climbs and the dewpoint spread $T - T_d$ closes. The dewpoint can reach the temperature but never pass it: at $T_d = T$ the air is saturated.
Where the closing spread reaches zero, cloud begins.
The cloud base: the lifting condensation level
Rising air saturates where the dry adiabat up from T meets the mixing-ratio line up from $T_d$: the lifting condensation level (LCL), the flat base of a cumulus. The temperature falls about 9.8 °C/km and the dewpoint about 1.8, so the spread closes at about 8 °C/km, and each degree of spread is worth about 125 m:
$z_{LCL} \approx 125\,\mathrm{m/°C} \times (T - T_d)$
Pilots use 400 ft per °C. The rule assumes air mixed from the ground to the cloud base, as on a sunny afternoon; under a night-time inversion, or with cloud from moisture aloft, it fails. The slips are multiplying the temperature instead of the spread, and mixing feet with meters.
Above the LCL the parcel is a cloud, and it no longer cools at the dry rate.
Cloud slows the cooling: the moist adiabat
Vapor condensing into cloud releases its latent heat, about 2.5 MJ per kilogram, so a saturated parcel cools at the moist adiabatic lapse rate $\Gamma_m$, less than the dry rate by an amount that grows with the vapor: about 4 °C/km in warm air near the ground, close to 9.8 °C/km high and cold. The dashed teal moist adiabats show it, upright near the warm bottom of the diagram and leaning over aloft. Thinner air, the cloud's shade or a slower ascent play no part: the dry rate is the same at every pressure.
With both adiabats drawn, a sounding's layers can be sorted by what they do to a lifted parcel.
Stable, conditionally unstable, unstable
A nudged parcel that ends up colder than its surroundings sinks back (stable); one that ends up warmer keeps rising (unstable). Compare the environment's lapse rate Γ with the parcel's, $\Gamma_d$ while clear and $\Gamma_m$ once saturated:
$\Gamma > \Gamma_d$: absolutely unstable $\Gamma_m < \Gamma < \Gamma_d$: conditionally unstable $\Gamma < \Gamma_m$: absolutely stable
An inversion is the most stable layer of all; steeper than dry adiabatic happens only just above hot ground. The mistake is comparing with only one adiabat: a layer cooling 8 °C/km is stable for dry parcels, yet saturated ones rise freely in it.
A stable layer low down can hold back an unstable atmosphere above it. That is the lid.
The lid and the level of free convection
A lifted parcel often stays colder than its surroundings for a kilometer or two, held back by warm, dry air above the moist layer, until it turns warmer at the level of free convection (LFC). The energy it would have to be given to get there is the convective inhibition (CIN), the blue area. With no buoyant layer below the LFC, the parcel needs a starting speed
$\tfrac{1}{2}w_0^2 > |\text{CIN}|$ $w_0 > \sqrt{2\,|\text{CIN}|}$
No thermal rises that fast, so a front, a hill or the afternoon sun must break the lid. Sunshine heats and stirs the lowest layer until its temperature follows a dry adiabat up from the ground. The afternoon high at which the CIN is gone is the convective temperature. By hand: follow the surface dewpoint's mixing-ratio line up to the temperature curve, at the convective condensation level (CCL), the base of the cumulus heating alone makes; the dry adiabat from there meets the ground at the convective temperature. Expecting a large CAPE to win at once is the usual mistake: until the lid breaks, nothing happens.
Above the LFC the parcel gains energy, and the diagram shows how much.
CAPE and CIN are areas
The work buoyancy does on each kilogram of a parcel from the LFC to the equilibrium level (EL), where it is no longer warmer, is the convective available potential energy (CAPE). The diagram is built so that an area on it is an energy per kilogram: the red area is the CAPE, and CIN the negative area below the LFC. For one unbroken positive layer a rectangle gives an estimate,
$\text{CAPE} \approx R_d\,\Delta T \ln(p_{LFC}/p_{EL})$
with ΔT the parcel's average excess and $R_d$ = 287 J kg⁻¹ K⁻¹ the gas constant of dry air. Strictly, buoyancy depends on density, so the temperatures compared are virtual temperatures $T_v$: the temperature dry air would need for the moist air's density, a little above T.
A storm's anvil spreads near the EL. The slips are using the pressure ratio itself, or log base 10, for the natural log of the ratio.
Energy per kilogram has the units of a speed squared, and the next section turns it into one.
How fast? Parcel theory's updraft
If all the CAPE became kinetic energy, $\tfrac{1}{2}w^2$ = CAPE at the EL, parcel theory's upper limit:
$w_{max} = \sqrt{2\,\text{CAPE}}$
Real updrafts often reach about half of it, slowed by mixing with drier air around them (entrainment), cloud water and the air above. Even so, $w_{max}$ ranks soundings well, and the updrafts that hold large hail aloft come from the soundings with the most CAPE. The slips are dropping the 2 and forgetting the square root.
An older number does the same job at one level.
The lifted index
The lifted index compares a lifted surface parcel with its surroundings at 500 hPa, about halfway up the troposphere:
$\text{LI} = T_{env} - T_{parcel}$ at 500 hPa
A parcel warmer than the air around it gives a negative LI, so negative means unstable: roughly 0 to −3 marginal, −3 to −6 moderate, below −6 very unstable (illustrative categories). It uses temperature, as originally defined; programs that use virtual temperature get it slightly more negative. One level cannot see a lid below it. The classic slip is the sign, which turns a very unstable sounding into an apparently stable one.
For pilots, where the temperature crosses freezing matters too.
The freezing level and icing
Supercooled cloud droplets, liquid below 0 °C, freeze onto whatever they hit. Most structural icing occurs in cloud between 0 and about −20 °C (a rule of thumb), so a pilot reads the freezing level, where the temperature curve crosses the 0 °C isotherm, as carefully as the cloud base. In air cooling at a steady Γ from $T_s$ at the ground,
$z_0 = T_s/\Gamma$
When a warm nose above freezing lies over sub-freezing air at the ground, snow melts in the nose and the drops refreeze on contact: freezing rain, with two freezing levels aloft. The slips are dividing the lapse rate by the temperature, forgetting kilometers to meters, and reading height above sea level where the scale gives height above the ground.
Freezing rain is one pattern a forecaster knows at a glance.
What a sounding says about the weather
Radiation fog: saturated only at the ground under a surface inversion. A marine layer: a shallow saturated layer under a strong inversion, stratus that cannot grow. Freezing rain: the warm nose. The inverted V: temperature and dewpoint far apart near the ground and close near 500 hPa, where rain evaporating in the dry layer chills it and the cold air plunges in violent gusts. The tropical sounding: deep moisture, weak lapse rates and almost no CIN, tall but rarely severe showers. And the loaded gun: moisture at the ground, a lid of warm, dry air and steep lapse rates aloft add up to severe storms, if the lid breaks. Read the whole shape: a large spread at the ground can mean a fine afternoon or a microburst.
Part 2 · Going further
Math you’ll need: derivatives, integrals and exponentials; one differential equation, in outline.
Potential temperature and Poisson's equation
For a parcel that exchanges no heat, the first law gives $c_p\,\mathrm{d}T = (R_d T/p)\,\mathrm{d}p$, and integrating to 1,000 hPa gives Poisson's equation for the potential temperature θ, the temperature the air would have if brought to 1,000 hPa without exchanging heat:
$\theta = T(1000/p)^{\kappa}$ $\kappa = 0.2854$
with T in kelvins and κ = $R_d/c_p$ = 287.04/1,005.7. Every point on a dry adiabat has the same θ, which labels the tan lines, so θ is conserved in dry adiabatic motion. A layer whose θ increases with height is stable for dry parcels; constant θ means well mixed. The slips are leaving T in degrees Celsius and inverting the ratio: air brought down to 1,000 hPa is compressed and warms, so θ exceeds T above that level.
The diagram's coordinates explain why the dry adiabats and isotherms lie as they do.
Why skewed, why log P: area is energy
The diagram plots x linear in T and y linear in ln p, with the skew added as a shear:
$x = s\,(T - T_0) + y\tan\varphi$ $y = v\ln(p_{bot}/p)$
where s and v are the scales and φ = 45°. The Jacobian of the map from (T, ln p) to (x, y) is the constant $s v$, so an area anywhere on the page is $s v \int T\,\mathrm{d}\ln p$, and $R_d \int T\,\mathrm{d}\ln p$ is a work per kilogram: that is why CAPE and CIN are areas. The shear $x \to x + y\tan\varphi$ never changes an area; it opens the angle between isotherms and dry adiabats to nearly 90°, so lapse rates can be compared by eye. At 0° the diagram is an emagram; with p in place of ln p it would no longer be equal-area, and the upper levels would count for too little.
The vertical coordinate also carries the heights.
Thickness and the hypsometric equation
Combine the hydrostatic equation, $\mathrm{d}p = -\rho g\,\mathrm{d}z$, with the gas law for moist air, $p = \rho R_d T_v$, and integrate between two pressures: the thickness of a layer depends only on its mean virtual temperature,
$\Delta z = (R_d \bar{T}_v/g)\ln(p_1/p_2)$ $\bar{T}_v = g\,\Delta z/(R_d \ln(p_1/p_2))$
A warm layer is thick, a cold one thin. The 1,000–500 hPa thickness is a thermometer for the lower troposphere: about 5,400 m, the "540 line", roughly separates rain from snow, a rule of thumb that terrain and a warm layer near the ground often break. The sim computes its height scale this way. The slips are log base 10 and mixing kelvins with degrees Celsius.
The virtual temperature depends on the air's water, and that depends on saturation.
Saturation and the mixing ratio
The saturation vapor pressure $e_s$, the vapor pressure in equilibrium with flat liquid water, rises steeply with temperature: by the Clausius–Clapeyron equation, $\mathrm{d}\ln e_s/\mathrm{d}T = L_v/(R_v T^2)$, from under 6 % per kelvin at 30 °C to 7.3 % at 0 °C. Radiosonde dewpoints are reported over liquid water, even below 0 °C; the sim uses Bolton's fit for $e_s$, and the saturation mixing ratio follows:
$e_s = 6.112\exp(17.67\,T/(T + 243.5))$ $r_s = \varepsilon e_s/(p - e_s)$
with T in °C and $e_s$ in hPa. The dotted mixing-ratio lines are lines of constant $r_s$. The slips are giving r in kg/kg when g/kg is asked, and leaving out ε.
With $e_s$ in hand, the LCL can be found exactly.
The exact LCL
The LCL is where the parcel's temperature on its dry adiabat equals the dewpoint of its fixed mixing ratio. Their difference falls steadily with height, so there is one root, which the sim finds by Brent's method. Bolton's eq. 15 gives the LCL temperature directly, in kelvins (it depends on T and $T_d$ alone, not on the starting pressure), and the dry rate estimates the height:
$T_L = 1/(1/(T_d - 56) + \ln(T/T_d)/800) + 56$ $\Delta z \approx c_p(T - T_L)/g$
Bolton's formula agrees with the exact solution within about 0.03 K; the height estimate takes the parcel's cooling as exactly $g/c_p$ per meter and is good to about 1 %. The LCL is a little colder than the starting dewpoint, because the dewpoint falls as the parcel rises. Forgetting to turn Bolton's kelvins back into degrees Celsius is the usual slip.
Above the LCL the parcel follows the moist adiabat.
The moist adiabatic lapse rate
For saturated ascent with the condensate falling out as it forms, the pseudo-adiabatic lapse rate is, in the AMS Glossary's form,
$\Gamma_m = g\,\frac{(1 + r_s)(1 + L_v r_s/R_d T)}{c_{pd} + r_s c_{pv} + L_v^2 r_s(\varepsilon + r_s)/R_d T^2}$
with T in kelvins, $L_v$ = 2.501 × 10⁶ − 2,370 T(°C) J/kg and $c_{pv}$ = 1,870 J kg⁻¹ K⁻¹ for the vapor. Set $r_s$ = 0 and it returns the dry rate $g/c_{pd}$. The sim integrates it in ln p by fourth-order Runge–Kutta; the formula, not the integration, limits the accuracy, to about 0.2 K up to 150 hPa. Using the dry rate in saturated air is the slip.
A saturated parcel conserves something along this curve, and it labels the moist adiabats.
Equivalent potential temperature
Condense all of a parcel's vapor, drop it out, and bring the parcel down dry to 1,000 hPa: its temperature then is the equivalent potential temperature $\theta_e$. It is conserved in dry and in pseudo-adiabatic ascent, so each moist adiabat is a line of constant $\theta_e$, labeled by $\theta_w$, its temperature at 1,000 hPa. Bolton's eq. 39, with r in kg/kg:
$\theta_e = \theta_{DL}\exp[(3036/T_L - 1.78)\,r\,(1 + 0.448\,r)]$ $\theta_{DL} = T(1000/(p - e))^{\kappa}(T/T_L)^{0.28\,r}$
$\theta_e$ moves about 0.3 K for every 0.1 g/kg, so round the mixing ratio late. Reporting θ, or $\theta_{DL}$ without the exponential, are the slips.
Vapor also changes a parcel's density, and so its buoyancy.
Virtual temperature
Water vapor (18 g/mol) is lighter than dry air (29 g/mol), so moist air is the less dense at the same T and p. The virtual temperature is the temperature dry air would need to have the same density:
$T_v = T(1 + r/\varepsilon)/(1 + r)$ $T_v \approx T\,(1 + 0.61\,r)$
The approximation keeps the first order in r; the next term, −0.61 r² T, is a tenth of a kelvin in very moist air. Buoyancy depends on density, so CAPE and CIN should use $T_v$ (Doswell and Rasmussen, 1994); with plain T, a moist parcel under a dry lid looks too cold, so the CIN comes out too strong, the LFC too high and the CAPE too small. The slips are r in g/kg and T in degrees Celsius.
Which air should be lifted in the first place depends on the sounding.
Which parcel?
Three parcels are standard. The surface-based (SB) parcel is the air at the ground. The mixed-layer (ML) parcel takes the mass-weighted mean θ and mixing ratio of the lowest 100 hPa and lifts it from the surface, which suits an afternoon when the sun has stirred that layer. The most unstable (MU) parcel starts at the level with the highest $\theta_e$ in the lowest 300 hPa. At night, with a shallow cold layer at the ground under moist, unstable air, storms feed on the air above the inversion: they are elevated, the MU parcel represents them, and the SB parcel can show almost no CAPE.
Parcels are lifted one at a time; fronts and mountains lift whole layers.
Lifting a layer: potential instability
When a front or a mountain range lifts a whole layer that is moist at the bottom and dry at the top, the bottom saturates first and then cools at the moist rate while the top keeps the dry rate. The layer stretches and its lapse rate steepens until, saturated throughout, it can be unstable although it began stable. The test is how $\theta_e$ changes with height:
$\partial\theta_e/\partial z < 0$: potentially (convectively) unstable
Judging such a layer by its lapse rate alone is the mistake.
So far the condensate has left the parcel at once. Keeping it changes the answer.
Pseudo-adiabatic or reversible ascent
A pseudo-adiabatic parcel drops its condensate as soon as it forms; a reversible parcel keeps it. The water warms the parcel a little through its heat capacity, but its weight pulls it down. With the total water $r_t$ fixed, the reversible parcel conserves its moist entropy, and its buoyancy uses the density temperature, which counts the liquid's weight:
$s = (c_{pd} + r_t c_l)\ln T - R_d\ln(p - e) + L_v r_v/T$ $T_\rho = T(1 + r_v/\varepsilon)/(1 + r_v + r_l)$
Without the weight, reversible ascent gives the most CAPE; with it, the least; pseudo-adiabatic CAPE, the usual convention, sits between. Thinking the condensate's heat outweighs its weight is the mistake.
In stable air a parcel does not rise freely, but it does move: it swings.
Stable air oscillates
Push a dry parcel up where θ increases with height and it is colder than its surroundings, so it sinks back, overshoots, and rises again. For small displacements the restoring acceleration is proportional to the displacement, and the parcel oscillates at the Brunt–Väisälä frequency N:
$N^2 = (g/\theta)\,\mathrm{d}\theta/\mathrm{d}z$ period $2\pi/N$
Tropospheric periods are typically 5 to 15 minutes, faster in inversions, and the same clock spaces gravity waves and lee-wave clouds downwind of mountains. The slips are leaving dθ/dz in K/km, which makes N about 30 times too large, and forgetting the 2π.
Parcel theory treats the parcel as sealed. The last section opens it.
Where parcel theory ends
A real updraft mixes with the air around it. In the sim's simple model of this entrainment, the saturated parcel's $\theta_e$ relaxes toward the environment's at a rate λ per kilometer of rise:
$\mathrm{d}\theta_{e,parcel}/\mathrm{d}z = -\lambda\,(\theta_{e,parcel} - \theta_{e,env})$
The environment's $\theta_e$ is usually lowest in the dry middle troposphere, so mixing cools the parcel and shrinks its CAPE; rates for deep convection are often quoted near 0.1–0.2 per km (illustrative). Ice, the condensate's weight, drag and pressure perturbations pull real storms further from parcel theory, so undiluted CAPE is an upper bound, though still the best single ranking of soundings.
Reference
Common mistakes
Part 1
- Reading straight down. The isotherms lean 45°; follow them.
- The lapse rate's sign and unit. $\Gamma = -\Delta T/\Delta z$ is positive for cooling, in °C/km.
- The environment's rate for a parcel. A clear parcel cools at $\Gamma_d$, about 9.8 °C/km.
- The dewpoint as conserved. The mixing ratio is; the dewpoint falls about 1.8 °C/km.
- One adiabat for stability. Compare with both $\Gamma_m$ and $\Gamma_d$.
- CAPE always wins. The CIN must be removed first.
- The logarithm. CAPE and thickness take the natural log of a pressure ratio.
- The LI backward. Environment minus parcel: negative is unstable.
Part 2
- Units. Poisson's equation and $T_v$ need kelvins, and formulas take r in kg/kg.
Equation sheet
| Part 1 | |
|---|---|
| lapse rate | $\Gamma = -\Delta T/\Delta z$ |
| dry adiabat | $\Gamma_d = g/c_p \approx 9.8$ °C/km; $T = T_0 - \Gamma_d\,\Delta z$ |
| mixing ratio | $r = \varepsilon e/(p - e)$ |
| cloud base | $z_{LCL} \approx$ 125 m/°C × $(T - T_d)$ |
| stability | $\Gamma_m < \Gamma < \Gamma_d$: conditionally unstable |
| the lid | $w_0 > \sqrt{2\,|\text{CIN}|}$ |
| CAPE | $\text{CAPE} \approx R_d\,\Delta T \ln(p_{LFC}/p_{EL})$, $R_d$ = 287 J kg⁻¹ K⁻¹ |
| updraft | $w_{max} = \sqrt{2\,\text{CAPE}}$ |
| lifted index | $\text{LI} = T_{env} - T_{parcel}$ at 500 hPa |
| freezing level | $z_0 = T_s/\Gamma$ |
| Part 2 | |
|---|---|
| Poisson | $\theta = T(1000/p)^{\kappa}$, $\kappa = 0.2854$ |
| thickness | $\bar{T}_v = g\,\Delta z/(R_d \ln(p_1/p_2))$ |
| saturation | $e_s = 6.112\exp(17.67\,T/(T + 243.5))$, $r_s = \varepsilon e_s/(p - e_s)$ |
| LCL (Bolton) | $T_L = 1/(1/(T_d - 56) + \ln(T/T_d)/800) + 56$ |
| moist rate | $\Gamma_m$ (AMS Glossary), with $L_v = 2.501 \times 10^6 - 2370\,T$ |
| virtual | $T_v = T(1 + r/\varepsilon)/(1 + r)$ |
| density | $T_\rho = T(1 + r_v/\varepsilon)/(1 + r_v + r_l)$ |
| oscillation | $N^2 = (g/\theta)\,\mathrm{d}\theta/\mathrm{d}z$, period $2\pi/N$ |
| entrainment | $\mathrm{d}\theta_e/\mathrm{d}z = -\lambda\,(\theta_e - \theta_{e,env})$ |
Symbols: T temperature (K in formulas, except $e_s$ and $L_v$, which take °C) · $T_d$ dewpoint · p pressure (hPa) · Γ lapse rate · r, $r_s$ mixing ratio and its saturation value (kg/kg) · e, $e_s$ vapor pressure and its saturation value · θ, $\theta_e$ potential and equivalent potential temperature · $T_v$, $T_\rho$ virtual, density temperature · $R_d$ = 287.04 J kg⁻¹ K⁻¹, $c_{pd}$ = 1,005.7 J kg⁻¹ K⁻¹, ε = 0.622, g = 9.80665 m/s² (Bolton's constants).
Further reading
J. M. Wallace and P. V. Hobbs, Atmospheric Science: An Introductory Survey, 2nd ed. (Academic Press, 2006), chapter 3; K. A. Emanuel, Atmospheric Convection (Oxford, 1994). The formulas follow D. Bolton, Monthly Weather Review 108, 1046 (1980), and the AMS Glossary of Meteorology; virtual temperature in CAPE, C. A. Doswell III and E. N. Rasmussen, Weather and Forecasting 9, 625 (1994). Open textbooks: OpenStax, University Physics Volume 2, §3.6 (adiabatic processes), and Chemistry 2e, §10.3 (vapor pressure). For the pilots' rules of thumb, the FAA's Pilot's Handbook of Aeronautical Knowledge and Aviation Weather Handbook. The Field Manual lists every source.
Atmospheric Stability & Skew-T Lab v1.3 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org