Meteorology

Meteorology studies atmospheric structure, motion and the processes producing weather. Uneven heating generates pressure gradients, air motion transports heat and water vapor, and condensation or freezing releases latent heat. Rotation and surface friction help determine how that motion develops.

A weather station measures local conditions, while comparing observations over space reveals evolving atmospheric systems. Wind, temperature, relative humidity and pressure constrain different parts of the physical state; no one sensor completely describes it. Forecasts combine these observations with equations for fluid motion, energy and moisture. Climatology instead emphasizes statistics over longer periods, using many of the same physical mechanisms and observing systems.

Wind is motion of atmospheric gas relative to the underlying surface. A pressure gradient accelerates air toward lower pressure, while the Coriolis force deflects moving air in a rotating frame and friction alters flow near the surface. Consequently, large-scale wind need not point directly from high to low pressure: approximate geostrophic balance can produce flow nearly along pressure contours.

An anemometer measures the local motion. A stated wind observation needs its height, averaging interval and directional convention, because gusts and surface obstacles produce variability. Vertical air motion is also physically important: an updraft transports heat and moisture and can support growing hail.

A region of upward air motion. In a thunderstorm, an updraft can support and transport precipitation particles and supply the liquid-water environment in which hail grows.

Ice precipitation grown within thunderstorms, chiefly by collection and freezing of supercooled liquid droplets. A strong updraft can retain stones in a growth region; changing growth conditions can create clear and opaque layers without requiring one complete up-and-down trip per layer.^[ref]

Frost is ice deposited or formed on a surface under sufficiently cold conditions. Hoar frost grows directly from water vapor, whereas frozen dew begins as liquid droplets that subsequently freeze. The frost point is the saturation condition relative to ice; it is distinct from the dew point defined relative to liquid water.

At night a surface can lose thermal radiation faster than surrounding air replaces the energy, so the surface temperature can fall below the air temperature measured at standard height. Frost can therefore occur even when that air observation remains above freezing.^[ref] Clouds and stronger mixing alter this energy exchange and hence the likelihood of a surface deposit.

The temperature at which the given water-vapour partial pressure equals saturation vapour pressure over ice. It is defined relative to ice, whereas the usual dew point uses equilibrium with liquid water.

Frozen dew is a surface ice deposit made by first condensing liquid water and then freezing those droplets. It differs in its sequence of phase changes from hoar frost, which grows directly from water vapor. A surface can first cool to its dew point while still above freezing and then cool further.

The distinction matters when interpreting visible deposits: the presence of ice alone does not establish direct vapour deposition. The relevant temperature is that of the surface, which may differ from a nearby sheltered air measurement because of thermal radiation and local heat exchange.^[ref]

An ice-crystal deposit formed directly from atmospheric water vapour on a sufficiently cold surface. It is distinguished from frozen dew, in which liquid water first condenses and subsequently freezes.^[ref]

A rotating atmospheric column over water, with visible mist and sometimes surface spray. Fair-weather waterspouts and thunderstorm-associated tornadic waterspouts have distinct formation settings.^[ref]

An atmospheric optical feature produced by interactions with ice crystals. A common $22^{\circ }$ halo comes from a $60^{\circ }$ prism path in hexagonal crystals; minimum deviation concentrates rays near its inner angular boundary. This differs from the liquid-water path of a rainbow.^[ref]

A weather station is a set of instruments measuring atmospheric conditions at a specified site. A thermometer records temperature, a barometer records pressure, an anemometer records wind, and a rain gauge measures collected rainfall. Moisture observations may be reported as relative humidity or dew point.

Instrument exposure is part of the measurement. Direct sunlight can heat a temperature sensor above the surrounding air; nearby buildings alter wind; the height and averaging interval affect comparability. A long record therefore needs instrument and site metadata as well as numbers. Consistent observations support both immediate meteorology and the statistics used in climatology.

A rain gauge measures precipitation collected over a known horizontal area. The reported rainfall depth is collected liquid volume divided by collector area: $h=V/A$. A tipping-bucket gauge converts repeated fixed-volume tips into accumulated depth; a weighing gauge measures collected mass.

Wind can deflect drops around the opening, and evaporation, splashing or freezing can bias the result. A gauge samples one location, so its reading need not represent a whole catchment during spatially uneven rain. The instrument geometry and exposure are therefore part of interpreting the observation.

An anemometer measures air motion. In a cup anemometer, moving air exerts different drag on the two sides of the cups, turning a shaft; a calibration converts its rotation rate into wind speed. A separate directional measurement is needed to recover a horizontal velocity vector. A sonic anemometer instead compares sound travel times along different paths and can operate without rotating parts.^[ref]

The instrument measures the flow at its own location, so obstacles, mounting height and gust averaging affect the result. The reported wind is therefore not simply an intrinsic property of an entire town or atmospheric layer. Comparing observations requires a consistent exposure and a stated sampling procedure.

A hot-wire anemometer infers fluid motion from the cooling of an electrically heated fine wire. Faster transverse flow generally changes the convective heat loss, changing either the wire's electrical resistance or the power needed to maintain its temperature. The output is converted into flow speed through an instrument calibration, not by assuming that electrical voltage itself is velocity.

A primary manufacturer describes a constant-temperature hot-wire anemometer with the wire in one arm of a Wheatstone bridge. A servo adjusts heating current to keep the wire resistance approximately fixed. Representative probes use tungsten wires about $1\mathrm{mm}$ long and $5\mu \mathrm{m}$ in diameter supported between prongs.^[ref] Maintaining resistance approximately fixes wire temperature only over the characterized resistance-temperature relation and control bandwidth.

A lumped thermal balance illustrates the mechanism:Here $C_w$ is the wire's heat capacity, $h$ a heat transfer coefficient, $A_w$ its exposed area and $T_a$ the surrounding fluid temperature. Conduction to supports and radiation are explicitly retained as losses. In a quasi-steady constant-temperature regime, the electrical heating must track the flow-dependent heat loss. In a constant-current arrangement, wire temperature and resistance instead vary as the cooling changes.

An empirical calibration often has the formwhere $E$ is a specified bridge output and $A,B,n$ are fitted for the probe, fluid and operating conditions. The coefficient units depend on the chosen units of $U$ and the exponent. For an original model with $A=1{\mathrm{V}}^2$, $B=0.5{\mathrm{V}}^2(m/s{)}^{-1/2}$ and $n=1/2$, $U=16m/s$ gives $E=\sqrt{3}\mathrm{V}$. Differentiation gives $dU/dE\approx 55.43(m/s)/V$ there, so a small $0.010\mathrm{V}$ error corresponds to about $0.554m/s$ before other uncertainty contributions.

A proposed experiment places the probe in a controlled reference flow, records bridge output over several speeds, measures ambient temperature and repeats the calibration after the experiment. Rotating the probe reveals directional sensitivity; a single straight wire does not uniquely resolve an arbitrary three-dimensional velocity or the sign of reversed crossflow. Hot-wire anemometry therefore needs a stated probe orientation, frequency response and data-reduction procedure as well as a steady calibration.

A constant-temperature hot-wire anemometer is a hot-wire anemometer whose electrical feedback adjusts wire heating to keep a chosen wire resistance approximately constant. The resistance acts as a temperature indicator; increasing flow cooling calls for increased electrical power. The primary manufacturer's arrangement places the wire in a Wheatstone bridge controlled by a servo amplifier.^[ref]

A simple resistance-temperature calibration illustrates the setpoint. If the cold resistance at fluid temperature $T_a$ is $R_a$, approximateFor stipulated $R_a=5\Omega$, $R_w=7.5\Omega$ and ${\alpha }_R=0.004{\mathrm{K}}^{-1}$, the overtemperature is $125\mathrm{K}$. At ambient $20{}^{\circ }\mathrm{C}$, that corresponds to wire temperature $145{}^{\circ }\mathrm{C}$ in this linear model. These parameters are illustrative, not universal tungsten specifications. A fixed resistance setpoint and a fixed overtemperature require different adjustments when ambient temperature changes.

At steady state, the heating balances convective, support-conduction and radiative losses. The observed bridge output must be calibrated against known flow with the same probe, fluid and operating conditions. A bridge output squared may follow an empirical relation $E^2=A+B{U}^n$ over a limited range, but circuit ratios and losses matter before identifying that output with wire power. The manufacturer's technical account separately treats velocity, directional and temperature calibration.^[ref]

Finite-bandwidth thermal feedback in hot-wire anemometry can be explored with an original local model. Let a temperature perturbation satisfywhere $C_w>0$ is wire heat capacity, $K>0$ a linearized thermal-loss coefficient and $F>0$ the incremental cooling per speed change. An ideal proportional power controller applies $\delta P=-K_c\delta T$, with $K_c>0$. HenceA cooling perturbation first lowers temperature, causing the controller to raise power. Finite feedback gain leaves a residual temperature perturbation and gives DC power sensitivity $F{K}_c/(K+K_c)$.

For $C_w=10^{-8}J/K$, $K=10^{-4}W/K$ and $K_c=9\times 10^{-4}W/K$, the model's thermal time constant changes from $100$ to $10\mu \mathrm{s}$ and its half-power frequency from about $1.59$ to $15.9\mathrm{kHz}$. The closed-loop DC response is $0.9F$. This calculation assumes an instantaneous controller; real amplifier and cable dynamics can add phase lag and limit usable gain.

A proposed characterization calibrates the steady speed response, then applies a small electrical test perturbation and examines settling and ringing while measuring ambient temperature. Check the response over the intended flow range and verify that the measuring electronics do not clip. The primary practical guide describes the feedback bridge as part of the complete measuring chain.^[ref] A fast electrical response alone does not establish spatial resolution or uniquely determine the direction of a reversing flow.

Hot-wire anemometry is the measurement method that turns a hot-wire anemometer's heat-transfer signal into a flow velocity time series and its statistics. It is especially useful for resolving fluctuations, but the interpreted quantity depends on probe direction, spatial extent and dynamic response. The manufacturer's primary method requires known-flow calibration and separately describes directional and temperature sensitivity.^[ref]

The order of processing matters because a nonlinear calibration does not commute with averaging. For an original relation $E^2=1+0.5\sqrt{U}$ in numerical volts and metres per second, equal-duration speeds $4$ and $16$ give outputs $\sqrt{2}$ and $\sqrt{3}$. Converting their average output giveswhereas converting each output first and then averaging gives the true two-level mean $10m/s$. A proposed processing chain therefore applies the validated instantaneous conversion before calculating statistics, subject to its bandwidth and direction limitations.

For equally spaced converted samples, define the record mean and root-mean-square fluctuation byThe stipulated record $8,10,12,10m/s$ has mean $10m/s$ and $u_{\mathrm{rms}}=\sqrt{2}m/s$. Its ratio $u_{\mathrm{rms}}/\overline{U}$ is about $14.14\%$. This is the RMS fluctuation of that finite record, not automatically an unbiased estimate based on four independent observations; turbulence samples can be strongly correlated.

The measurement has both temporal and spatial filters. An electronic anti-aliasing filter must sufficiently attenuate unresolved fluctuations before digitization; a sampling rate of $10\mathrm{kHz}$ has Nyquist frequency $5\mathrm{kHz}$ regardless of the wire's faster response. Finite wire length produces hot-wire spatial averaging, so increasing sampling rate alone cannot recover spatial scales already averaged by the sensor. Flow direction changes can also be misread as speed changes by a single-wire calibration.

A useful proposed grid-turbulence experiment keeps the wire transverse to a positive mean flow, performs a reference-flow calibration before and after acquisition, records ambient temperature, and documents wire geometry, bridge settings, sample rate and filtering. Compare means and fluctuation spectra at different positions and repeat with a shorter wire or a changed electronic bandwidth. A changed high-frequency spectrum can reflect instrumental filtering as well as changed flow physics.

The electrical control loop should also be characterized rather than assigning a universal bandwidth to the wire alone; the primary description explicitly connects frequency response to wire and servo behavior.^[ref] A measured time series is evidence for the flow only within the direction, spatial, thermal and frequency ranges actually tested.

Hot-wire spatial averaging is the loss or alteration of small-scale flow information because a heated sensing wire has finite spatial extent. Even a perfectly fast electronic readout observes a spatially distributed heat-transfer response. The manufacturer's practical guide relates spatial resolution to wire dimensions, flow and instrument frequency response.^[ref]

An original small-perturbation model treats the inferred velocity fluctuation as a uniform line average over wire length $\ell$. With $s$ measuring distance along the wire,For a spatial sinusoid $u(s,t)=\hat{u}\cos (ks-\omega t+\varphi )$, direct integration givesThis is a sinc function transfer factor. It follows from the stipulated uniform weighting and local linearity; a real wire's nonuniform temperature and nonlinear heat transfer require a more detailed spatial kernel.

For spatial wavelength ${\lambda }_s=2\pi /|k|$, a fluctuation with ${\lambda }_s=4\ell$ has amplitude factor about $0.9003$ and variance or power factor $|H{|}^2\approx 0.8106$. At ${\lambda }_s=2\ell$, the factors are $2/\pi \approx 0.6366$ and $4/{\pi }^2\approx 0.4053$. At ${\lambda }_s=\ell$, one complete sinusoidal cycle lies along the wire and its line average is exactly zero. Raising the sample rate cannot restore a component that this spatial average has canceled.

For a $1\mathrm{mm}$ wire, these illustrative wavelengths are $4$, $2$ and $1\mathrm{mm}$. The result depends on the component of the spatial wavevector along the wire, rather than only on a temporal frequency. Orientation-dependent spatial filtering by a hot wire matters because a moving three-dimensional structure can have different projections along the wire and the mean flow. Under a frozen-advection model, temporal frequency obeys $\omega =\mathbf{k}\cdot \mathbf{U}$; substituting $k=\omega /U$ into the wire-length factor requires an additional alignment assumption.

Spatial and temporal filters can both reduce measured fluctuation variance. In this single-sinusoid model, the measured variance is $|H{|}^2$ times the true variance even with noiseless electronics. A correction obtained by dividing by $H$ becomes unstable near its zeros, amplifying measurement noise without recovering independently lost information.

A proposed test compares probes of different active lengths in the same statistically repeatable flow, after separate steady calibrations and matched electronic filtering. Record wire orientation, support geometry, ambient temperature and traverse location; compare fluctuation spectra as well as mean velocity. A difference that tracks wire length supports a spatial-filtering interpretation, but changed thermal response or probe disturbance must also be checked. The observation establishes the filtered response of the actual apparatus over its characterized range, not an unqualified point measurement at every turbulent scale.

A cup anemometer infers wind speed from the rotation rate of cups mounted around a vertical axis. The cup geometry produces different aerodynamic forces on the two sides of the rotor, giving a net driving torque.

A calibration relates rotation frequency to flow speed over the useful range. Bearing friction affects low speeds, and rotor inertia limits response to rapid fluctuations. The cup speed is not simply equal to wind speed. Comparing the output with a reference flow establishes the transfer relation needed for quantitative observations.

A sonic anemometer measures wind from acoustic travel times between transducers. For path length $L$ aligned with velocity component $u$, the idealized transit times are $t_{+}=L/(c+u)$ and $t_{-}=L/(c-u)$, givingThe reciprocal-time difference removes the common sound speed in this model. Several paths recover a velocity vector. Transducer wakes, path averaging and precipitation can affect the measurement, so geometry and calibration matter when using it for turbulent fluxes.

The temperature at which an air parcel with a specified water vapor partial pressure first becomes saturated on cooling at fixed pressure. It solves $E(T_D)=e_v$ with saturation pressure $E$. A falling dew point during ascent does not by itself imply loss of water vapor.

Fixed water vapor mole fraction gives $d\ln e_v=d\ln p$. Combining hydrostatic balance with the Clausius-Clapeyron relation yields $d\ln T_D/dz=-(g{R}_v/L_v{R}_d)(T_D/T)$. If $T_D/T\simeq 1$, then $T_D(z)\simeq T_D(0)e^{-z/H}$. All ratios use absolute temperature; constant latent heat, fixed composition and approximate parcel–environment temperature equality are required.

The difference between air temperature and its dew point. For unsaturated air it is positive. Near terrestrial surface conditions, a closed ascending parcel's lifting condensation level is approximately $125\mathrm{m}{\mathrm{K}}^{-1}$ times this difference, because its temperature decreases about $8\mathrm{K}\mathrm{k}{\mathrm{m}}^{-1}$ faster than its dew point.

Atmospheric circulation redistributes heat, water vapour and momentum through the atmosphere. Unequal radiative heating creates temperature and pressure contrasts; gravity, planetary rotation, friction and the evolving flow constrain the response. A simple direct overturning circulation is modified strongly by the Coriolis force on a rotating planet.

The tropical Hadley cell has rising air in strongly heated regions and a returning lower branch. Midlatitude eddies transport heat and momentum and help maintain the Ferrel cell. These statistical cells summarize a time-varying flow containing waves and storms. They explain why a local radiation imbalance can be balanced by horizontal transport rather than by local radiation alone.

The Southern Oscillation is a large-scale variation of tropical atmospheric pressure and circulation, associated with the coupled El Niño–Southern Oscillation. Its pressure pattern is commonly monitored through the contrast between the eastern and western tropical Pacific.

Changes in trade winds affect ocean upwelling and surface temperatures; those temperature changes can feed back on convection and pressure. The atmospheric oscillation and the oceanic El Niño pattern are therefore connected components of a coupled system. A pressure index is an observational summary, not a complete dynamical model of the phenomenon.

Named after William Ferrel, the thermally indirect middle-latitude mean atmospheric overturning: near-surface flow is poleward and the upper return equatorward. It is maintained largely by eddy forcing and surface stress. Its Eulerian mean thermal transport can oppose the poleward eddy contribution; it is not a second thermally direct Hadley cell.

A tropical atmospheric overturning circulation with rising air near the equator, poleward flow aloft, subtropical descent and an equatorward lower return. In its ideal upper branch, conservation of angular momentum gives $u=\Omega a{\sin }^2\varphi /\cos \varphi$ for air initially at rest relative to the equatorial surface. Eddies and friction modify this idealization.

A comparatively narrow band of strong atmospheric flow. Tropospheric jets are associated with large horizontal temperature gradients through the thermal wind relation; strong potential vorticity gradients can guide Rossby waves. The background speed and its spatial variation control Doppler shifting and refraction of wave packets. Not every atmospheric jet has the same origin or altitude.

Correlated atmospheric variability between geographically separated regions. Wave propagation, mean-flow changes and coupled atmosphere–ocean forcing can produce physical connections, while statistical correlation alone does not identify a unique cause. A Rossby-wave teleconnection is an important large-scale pathway rather than an exhaustive definition of all teleconnections.

A spatially localized forcing, such as persistent tropical heating, can excite Rossby waves that propagate wave activity into distant regions. Mean winds Doppler-shift the waves, and background potential vorticity gradients and jets guide or refract the packets. A stationary pattern can thus connect remote ridges and troughs even though the underlying wave energy propagates. Group velocity, rather than phase velocity alone, diagnoses packet propagation; damping, seasonality and eddy feedbacks affect the response.

The level where an initially unsaturated air parcel first reaches saturation during an approximately dry adiabatic ascent. Before saturation, expansion cools the parcel while its water-vapour mixing ratio remains approximately constant. At saturation, further ascent can condense water and release latent heat; the subsequent parcel lapse rate differs from the dry rate. Mixing with surrounding air changes this ideal parcel construction.

A relatively warm, dry downslope wind on the lee side of a mountain. Condensation and condensate removal during ascent can produce precipitation-driven foehn warming; isentropic foehn drawdown, turbulent mixing and radiative warming can also contribute. Rainfall is therefore not a necessary condition for every foehn event. The atmosphere's upstream stratification, moisture and flow over or around the mountain determine the contributing mechanisms.^[ref]

Mountain blocking can prevent low-level upstream air from crossing the barrier while air from higher levels passes over and descends. In stable stratification that elevated source air has higher potential temperature; adiabatic descent can bring it into a lee valley warmer than the blocked low-level air. This source-air change permits a dry foehn without requiring condensation and rainout along the arriving parcel's path.

In the classical idealized parcel path, ascend dry adiabatically to condensation height $z_c$, continue at constant moist lapse rate ${\Gamma }_m$ to height $H>z_c$, lose condensate, then descend at dry lapse rate ${\Gamma }_d$ to the initial height. Adding the temperature changes gives the title. Complete condensate removal and an unsaturated descent are essential assumptions. A reversible dry ascent and descent with the same source air instead gives no net temperature increase at the same pressure.

The temperature a parcel would have after reversible adiabatic compression to a reference pressure. For a dry ideal gas, $\theta =T(p_0/p{)}^{R/c_p}$; it is conserved for dry adiabatic motion.

For a dry fixed-composition ideal gas with constant specific heats, $ds=c_{p}d\ln T-R_{d}d\ln p$. Defining $\theta =T(p_0/p{)}^{R_d/c_p}$ gives the displayed identity. Reversible adiabatic parcel motion conserves entropy and theta simultaneously. Irreversible entropy production or moist latent heating can change ordinary dry theta even without heat exchanged through the parcel boundary.

Dry potential temperature is conserved for a fixed-composition ideal gas undergoing a reversible adiabatic change. Adiabatic means no heat passes through the boundary; it does not by itself prohibit internal entropy production. With constant specific heat capacity at constant pressure,Positive irreversible entropy production can therefore increase $\theta$ even when the parcel receives no boundary heat.

For example, dissipation of mechanical energy by viscosity heats the gas, and a shock changes its entropy. Mixing also requires care because the resulting parcel is not simply the original unmixed material following one reversible path. This qualification prevents treating all unheated atmospheric motion as exactly conserving dry potential temperature.

The Met Office is the United Kingdom's national meteorological service, producing observations, forecasts and climate information. Its scientific work connects instruments and numerical models: measurements constrain the present atmospheric state, while fluid and thermodynamic equations predict its evolution.

A concrete example is wind observation. The Met Office describes how cup anemometers and sonic anemometers measure local flow and why exposure and averaging conventions matter.^[ref] Its descriptions of the North Atlantic Oscillation also distinguish a large-scale circulation index from a deterministic forecast of weather at one site.

A lower-atmospheric high-pressure circulation associated with a cold dense air mass, often over a continent in winter. This does not imply the same circulation sign at all altitudes.

A lower-atmospheric low-pressure circulation associated with a warm air mass or heated land surface. Its strength and vertical structure depend on environmental constraints.

The National Oceanic and Atmospheric Administration is a United States scientific agency whose observations and research cover the atmosphere and oceans. Its measurement products connect meteorology, climatology and ocean circulation, with a stated observing method and reference period needed to interpret each product.

For example, its Arctic Oscillation index projects a measured atmospheric field onto a specified spatial pattern, while a climatological normal summarizes a different kind of information: statistics over a defined reference interval.^[ref][ref] Neither product is a direct measurement of one universal scalar called climate.

The World Meteorological Organization coordinates international cooperation on atmospheric observations and their interpretation. Shared measurement conventions allow observations made by different services to be compared: instrument exposure, averaging intervals and reporting definitions can change the meaning of otherwise similar numbers.

Its instrument guide documents methods for measuring quantities such as pressure at a weather station.^[ref] The International Cloud Atlas by World Meteorological Organization similarly provides definitions for cloud and precipitation phenomena. Such conventions support physical comparison across observing networks; they do not remove instrument uncertainty or local sampling effects.

Condensed water falling from clouds or deposited from the atmosphere as liquid or solid particles. It can efficiently scavenge soluble gases and aerosol particles. Gas conversion into sulfate and subsequent atmospheric precipitation removal are separate stages of an atmospheric sulfur budget.

Liquid-water precipitation reaching the surface. In a catchment water balance its rate is an input measured as water depth per time; only the part not absorbed or otherwise lost becomes rainfall excess feeding overland flow.

Nucleation creates the initial ice embryo, either within liquid or with assistance from a suitable interface. Deposition, riming and ice-crystal aggregation enlarge already existing ice. A list of efficient ice-growth processes does not by itself explain how the first crystals formed. Small-particle nucleation probabilities depend on thermodynamic driving, interfaces, volume and exposure time.

Ice particles collide and stick to form a larger aggregate, such as a snowflake. Temperature, habit and collision speed affect sticking. This differs from vapour deposition onto ice and from riming, in which collected supercooled liquid freezes on the ice particle. ^[ref]

The fraction of a specified condensate or moisture supply that appears as precipitation under a stated spatial, temporal and flux convention. Collection, fall-out, entrainment, evaporation and sublimation all affect it. A cloud can activate many droplets yet precipitate inefficiently; droplet formation, growth to precipitation size and survival to the surface are separate requirements.

Supercooled droplets collide with an ice particle and freeze onto it, increasing its mass. Deposition, riming and ice aggregation are distinct growth mechanisms.

Because ice has lower equilibrium vapour pressure than supercooled liquid, a mixed-phase environment can sustain deposition onto ice while liquid droplets evaporate. This is distinct from ice collecting droplets by riming.

An atmospheric electrical discharge. Its high-temperature chemistry can produce nitrogen oxides, influencing ozone and radical chemistry away from surface emission sources. The ozone response depends on subsequent transport and the local chemical regime.

Electric current in the ground after a nearby lightning discharge. A person can bridge points of different electric potential, creating indirect exposure without the main channel striking the body.

A suspension of liquid-water droplets, ice particles or both in an atmosphere. Clouds interact with solar and thermal radiation; their net radiative effect depends on height, water content, particle size distribution and coverage.

The local mass of suspended liquid water per unit volume, in $\mathrm{kg}{\mathrm{m}}^{-3}$. Integrating this quantity vertically gives liquid-water path; it is not the total moist-air density.

The vertically integrated liquid-water mass per horizontal area in a cloud. Its units are mass per area, unlike the local liquid-water content per volume. Droplet number, effective radius and water path are distinct properties; changing one does not specify the others without further assumptions.

Radiative cooling can promote condensation and turbulence; net heating can promote evaporation or sublimation. Cloud altitude alone does not determine lifetime, because optical thickness, environmental radiation, entrainment and moisture supply all matter.

A liquid-water particle suspended in a cloud. Number, size distribution and liquid-water content are separate properties; increasing number at fixed total water shifts the characteristic radius downward.

For droplet number density $N$ and radius distribution $n(r)$, let $r_v^3=N^{-1}\int r^{3}n(r)dr$ and $r_s^2=N^{-1}\int r^{2}n(r)dr$. The effective radius is $r_e=r_v^3/r_s^2$. With liquid-water content $w=c_{w}h$, $r_v=[3c_{w}h/(4\pi {\rho }_{w}N){]}^{1/3}$. The shape factor $k=r_v^3/r_e^3$ gives $r_s^2=k^{1/3}{r}_v^2$; these are distribution moments, not a monodisperse assumption.

The number of cloud droplets per volume in each radius interval, often represented by $n(r)dr$. Its width controls differential terminal speeds and collection opportunities; condensational growth with a common $d(r^2)/dt$ narrows relative radius differences.

The ratio of the third to the second radius moment determines projected extinction area per liquid mass. For size-independent extinction efficiency $Q$, the local particle extinction coefficient is $3Qw/(4{\rho }_w{r}_e)$. It is not generally the mean radius or the mean-volume radius. In adiabatic cloud droplet moments, the shape factor gives $r_e=r_v/k^{1/3}$.

Growth of liquid cloud droplets by collisions followed by merging. Differences in droplet fall speeds and collection efficiency affect the process. Increasing droplet number at fixed liquid-water content can reduce typical sizes and delay drizzle, but cloud mixing and water-content adjustments prevent a universal precipitation response.

If a spherical collector falls with $u_R=k_{R}R$ through constant liquid-water content $L$ and efficiency $E$, neglecting condensation and small-droplet fall speed, then $\dot{R}={\alpha }_{R}R$ with ${\alpha }_R=k_{R}LE/(4{\rho }_l)$. A uniform updraft gives $h(t)=Ut-(k_R{R}_0/{\alpha }_R)(e^{{\alpha }_{R}t}-1)$. The formula is valid only while the assumed cloud conditions hold.

A gravitational collision kernel multiplies geometric encounter area, relative fall speed, collection efficiency and coalescence efficiency. Its units are volume per time. The number of encounters also depends on the droplet size distribution. Hydrodynamic deflection, breakup and turbulence change the simple settling picture. Condensational growth alone tends to reduce relative size differences, so a precipitation-growth calculation must explain how efficient collectors arise.

A meteorologist studies the atmosphere and weather using observations, physical theory and numerical models. Fluid mechanics, thermodynamics and radiative processes connect measured pressure, temperature and moisture with changing winds and clouds.

Forecasting requires an estimated initial state as well as evolution equations. Uncertainty in observations and model approximations can grow, so an ensemble of forecasts is useful for expressing a range of plausible outcomes.

Joseph Smagorinsky developed numerical experiments on atmospheric circulation and was a central figure in the early modeling work at the Geophysical Fluid Dynamics Laboratory. Its institutional history credits his work with Syukuro Manabe in building atmospheric models for climate research during the 1950s and 1960s.^[ref]

His 1963 General Circulation Experiments with the Primitive Equations: I. The Basic Experiment by Joseph Smagorinsky integrated a simplified atmospheric model over an extended period and examined its momentum and heat transport. The model represented winds at two internal levels and a single temperature, with specified radiative forcing and dissipative mechanisms. Its lateral exchange coefficient depended on the local horizontal strain rate.^[ref]

The motivating difficulty is scale. A computational grid can represent large circulation patterns while failing to resolve smaller eddies that still transport momentum and heat. The Smagorinsky model represents some effects of those unresolved motions by an eddy viscosity, linking the closure to the resolved deformation rather than tracking every small eddy directly.

The original experiment also illustrates scientific restraint: Smagorinsky reported aspects of agreement with observed circulation while judging the modeled small-scale diffusion to play too large a role. A model can reproduce broad patterns yet misallocate the processes producing them. Comparing transport mechanisms and energy budgets therefore supplies a stronger test than visual resemblance alone.^[ref]

Edward Norton Lorenz showed how a deterministic low-dimensional model of convection can produce bounded, nonperiodic evolution with strong sensitivity to its initial state. His 1963 paper became a central example of chaos: the equations determine a trajectory, while finite initial precision can still limit long-time prediction.

The model retains three coupled variables from a simplified convection description rather than attempting to simulate the entire atmosphere. This distinction matters: the result demonstrates a mechanism of unpredictability without making every weather pattern random or every forecast impossible.^[ref]

Jeff Haby authored forecasting notes that connect the formation of surface condensation and frost with surface cooling, humidity and wind. The notes distinguish ice deposited from vapour from frozen dew.^[ref]

A surface exposed to a clear night sky can become colder than the air measured above it. Forecasting a freezing surface therefore requires its energy balance and moisture conditions, not only a standard-height minimum air temperature.

Alfred Wegener was a meteorologist and geophysicist who developed continental drift into a coordinated scientific argument. He presented the proposal publicly in 1912 and published its first book-length treatment in 1915. His investigations of the atmosphere grew alongside his polar fieldwork: observations from Greenland expeditions contributed to his work on atmospheric thermodynamics.^[ref]

The matching continental outlines were a starting clue. Alfred Wegener assembled evidence from geophysics, geology and paleontology, including related rock formations and past organisms on opposite sides of the Atlantic. He also used isostasy to challenge the idea that vanished land bridges explained these connections: the density contrast between continents and their substrate mattered to his physical argument. He attempted to test motion through geodesy in Greenland, but the inferred displacement rates were much too large. This distinguishes the broad evidence for continental drift from an individual attempted measurement of it.^[ref]

William Ferrel analysed atmospheric circulation on a rotating Earth. The Ferrel cell carries his name and describes the middle-latitude branch of the idealized meridional circulation.^[ref]

Its modern interpretation requires care: this mean circulation is thermally indirect and strongly affected by eddy transport. It is not simply a second thermally direct Hadley cell. The distinction links the historical three-cell picture to the conservation and transport arguments used in geophysical fluid dynamics.

Gilbert Walker used correlations among atmospheric observations to identify large-scale relationships, notably the opposed tropical pressure variations called the Southern Oscillation.^[ref]

The Walker circulation is named in his honour, but a pressure index and a circulation mechanism are different objects. Correlation identifies a recurring pattern; interpreting it requires the coupled physics of trade winds, ocean temperature and atmospheric motion. That distinction remains useful when a climate index is used for prediction.

Isaac Held develops physical explanations of atmospheric circulation and its remote responses. His demonstration of teleconnections uses localized tropical heating in a rotating-sphere model to generate a stationary Rossby wave response extending into other latitudes.^[ref]

This supplies a dynamical pathway connecting a tropical forcing with distant pressure and circulation patterns. It goes beyond observing a statistical correlation by testing how the atmospheric flow transports the disturbance.

Sebastian Bathiany coauthored an analysis explaining why increasing carbon dioxide can cool the middle atmosphere while contributing to warming elsewhere. The work compares radiative regimes rather than treating a gas as having one temperature effect independent of its surroundings.^[ref]

The physical interpretation depends on the balance of absorbed and emitted radiation, the atmospheric structure and the level at which energy escapes. This is a concrete application of radiative transfer to climate.

Helge F. Goessling coauthored work on the radiative response of the middle atmosphere to increased carbon dioxide. The paper develops a model perspective that distinguishes local emission and absorption from the whole-column energy budget.^[ref]

It illustrates why the effect of changing an absorber cannot be inferred from its absorption strength alone. Optical depth, temperature and the surrounding radiation field jointly determine whether a layer gains or loses energy.

Piers Forster works on the planetary energy budget, radiative forcing, climate feedback and climate sensitivity. He is a coordinating author of an assessment chapter bringing those quantities into one quantitative framework.^[ref]

The distinctions matter physically: a forcing perturbs the energy budget, while feedback describes how the radiative response changes as the climate state changes. The inferred sensitivity also depends on the response timescale and on how those quantities are defined and estimated.

S. Twomey analysed how changing the number of cloud particles can alter shortwave reflection. At fixed liquid-water amount, distributing the water among more, smaller droplets can increase cloud optical depth and albedo.^[ref]

The fixed-water assumption separates this optical mechanism from changes in cloud lifetime, water content and atmospheric precipitation. It therefore identifies one contribution to a cloud's radiative response, rather than a universal numerical relation between aerosol abundance and total climate feedback.

Richard T. Wetherald and Syukuro Manabe calculated atmospheric radiative-convective equilibrium with a prescribed relative humidity distribution. Allowing the water-vapour content to adjust with temperature changed the sensitivity compared with fixing absolute humidity.^[ref]

The calculation demonstrates why a humidity constraint is a physical modelling assumption: it controls part of the radiative feedback as the atmosphere warms.

Syukuro Manabe and Richard T. Wetherald modelled an atmosphere relaxing toward radiative-convective equilibrium and studied its response to changes in factors including carbon dioxide. Their prescribed relative humidity lets water vapour change with temperature.^[ref]

This couples radiative transfer with vertical thermal adjustment, making the feedback assumptions explicit. A simpler dry or fixed-water-content atmosphere need not predict the same equilibrium response.

Weather forecasting estimates future meteorological conditions from observations and physical models. Numerical models use an approximate initial condition and equations of fluid mechanics to simulate future conditions; forecasters also compare model outputs and apply experience and research.^[ref] The forecast horizon describes how far ahead a prediction remains useful at a specified accuracy; sensitive dependence on initial conditions can limit detailed predictions under deterministic evolution.

An ensemble runs multiple forecasts with perturbed initial states, model parameters or physical formulations to represent uncertainty. Its distribution can indicate probabilities and sensitivity, but finite sampling or shared model errors can make the ensemble underdispersive. Growing initial perturbations can be selected using right singular vectors of a finite-time propagator.

Combining observations with a dynamical prior estimates an initial atmospheric state and its uncertainty. In a linear Gaussian update, analysis increments are $K(y-H{x}_b)$ with $K=B{H}^T(HB{H}^T+R{)}^{-1}$, where $B$ and $R$ are prior and observation error covariance matrices. Observation noise, sparse coverage and incorrect model or covariance assumptions can all leave forecast errors.