Meteorology

Meteorology is the scientific study of the Earth’s atmosphere, including its physical and chemical properties, its motions, and its interactions with land and oceans. Its central concerns include understanding and predicting weather, from local cloud formation and thunderstorms to large-scale circulation systems. Meteorology overlaps with climatology, which emphasizes the statistical characteristics and variability of climate, rather than only particular atmospheric conditions and events. (glossary.ametsoc.org)

The discipline combines observations, physical theory, laboratory and field investigations, and mathematical modelling. Weather stations, balloons, radar, aircraft, ocean platforms, and satellites supply measurements that support atmospheric analysis and forecasting. Computer-based prediction and ensemble methods estimate both future conditions and their uncertainty, providing information used in transport, agriculture, water management, energy production, and warnings of hazardous weather. (community.wmo.int)

Meteorology developed from accumulated weather observations and early natural philosophy into an instrument-based, internationally coordinated science. Its history encompasses the emergence of national forecasting services, the formulation of numerical prediction, and the integration of atmospheric research into a broader Earth-system approach. Its principal subjects therefore range from atmospheric thermodynamics and dynamics to observing systems, forecast methods, environmental interactions, and public services. (metoffice.gov.uk)

Scope, Branches, and Scales

Meteorology investigates the composition, structure, physical processes, and motion of the Earth’s atmosphere, including its exchanges with land and oceans. It is a major component of atmospheric science, a broader designation encompassing atmospheric physics, chemistry, and related disciplines. Weather forecasting is one application, not the discipline’s entire scope: explaining cloud formation, measuring atmospheric radiation, and investigating turbulent transport are meteorological activities even when they do not produce forecasts. Climatology emphasizes the statistics, variability, and long-term evolution of climate, drawing on many of the same observations and physical principles; the distinction is therefore one of emphasis rather than separate governing laws. For example, climatological standard normals describe conditions over 30-year periods, whereas an individual weather analysis represents atmospheric conditions at a specified time. (glossary.ametsoc.org)

The discipline’s boundaries follow exchanges of matter and energy rather than the atmosphere alone. Hydrology connects atmospheric precipitation and evapotranspiration with terrestrial water storage and flow; oceanography examines the oceanic side of exchanges of heat, moisture, and momentum with the atmosphere. Physical geography supplies the spatial setting, including terrain, vegetation, and land cover, that modifies atmospheric conditions. Thus rainfall over a drainage basin is simultaneously an atmospheric process, an input to the water cycle, and a geographically distributed phenomenon. These connections motivate coupled studies without making meteorology synonymous with the whole Earth system. (wmo.int)

Branches differ partly by their questions and partly by their methods. Physical meteorology examines processes such as radiative absorption and scattering, cloud-particle growth, atmospheric electricity, and changes of water phase, using physical theory, measurements, and experiments. Dynamical meteorology asks how atmospheric motions arise and evolve, expressing conservation principles through equations of fluid motion and studying their analytical or numerical solutions. Synoptic meteorology analyzes observations assembled for common reference times, especially weather charts that reveal the distribution and evolution of pressure systems and associated weather. These branches overlap: understanding a cyclone requires both its circulation and the physical processes operating within its clouds. (glossary.ametsoc.org)

Mesoscale meteorology focuses on intermediate-sized phenomena, including thunderstorms, squall lines, sea breezes, and mountain-generated disturbances. It combines detailed observations, including radar measurements, with theoretical analysis and high-resolution simulations to investigate their initiation, organization, and interaction with their surroundings. Applied meteorology is distinguished by purpose rather than scale: it uses atmospheric observations, analyses, and forecasts for practical problems in agriculture, transportation, pollution management, and industry. A study of storm dynamics and an aviation assessment of the same storm can therefore share data while addressing different questions. (glossary.ametsoc.org)

Scale classifications usually refer to characteristic horizontal dimensions, with associated—not rigidly prescribed—durations. The microscale, approximately 2 km or less in one common convention, includes turbulence and localized circulations. The mesoscale spans a few kilometres to several hundred kilometres in the American Meteorological Society’s general definition. Isidoro Orlanski’s 1975 classification instead gives a precisely partitioned mesoscale: meso-gamma, 2–20 km; meso-beta, 20–200 km; and meso-alpha, 200–2,000 km. The synoptic scale covers several hundred to several thousand kilometres, characteristic of migrating high- and low-pressure systems. The planetary scale concerns the largest circulations and long waves, with dimensions of order 10,000 km or more. These ranges express conventions, not physical discontinuities. (glossary.ametsoc.org)

A phenomenon can occupy several categories depending on the dimension considered: an elongated front has a much smaller cross-front width than its along-front extent. Nor are scales dynamically isolated. Synoptic flows establish environments in which mesoscale storms develop, while cloud-scale heating and momentum transport can modify larger circulations. Such interactions are studied through combinations of observations, idealized experiments, and multiscale numerical models rather than by treating each size class as an independent atmospheric system. (twister.caps.ou.edu)

Historical Development

Weather knowledge developed in different societies through seasonal calendars, practical rules, and records of unusual events. Around 650 BCE, Babylonian forecasters interpreted clouds and optical phenomena such as haloes; by about 300 BCE, Chinese astronomers had developed a calendar dividing the year into 24 seasonal periods associated with characteristic weather. These traditions connected recurring observations with expectations rather than calculating atmospheric evolution. In the fourth century BCE, Aristotle composed Meteorologica, treating rain, winds, thunder, and other natural phenomena within a common explanatory framework. Its scope also included comets, earthquakes, and the sea. “Meteorology” derives from Greek terms meaning the study or treatment of “things aloft,” not specifically the study of meteorites. (science.nasa.gov)

Instrumental measurement gradually made atmospheric descriptions quantitative. Evangelista Torricelli’s mercury barometer of 1643 enabled measurements of atmospheric pressure, while improvements in thermometers and hygrometers supported records of temperature and humidity. Standardized descriptions also made observers’ reports more comparable. Luke Howard’s On the Modification of Clouds (1803) introduced Latin cloud names, including cirrus, cumulus, and stratus, that became the basis of subsequent cloud classification. Francis Beaufort developed his wind-force scale in 1805 and recorded it in his journal in January 1806; subsequent revisions related wind force to the sail a frigate could carry. Such classifications converted otherwise variable verbal accounts into repeatable observational categories. (science.nasa.gov)

Mid-nineteenth-century telegraph networks allowed observations from separated stations to reach a central office quickly enough to inform forecasts. Reports collected for corresponding times supported synoptic charts showing the geographical arrangement of pressure, winds, and storms. In Britain, Robert FitzRoy became head of the newly established Meteorological Department in 1854. Coastal stations subsequently transmitted morning observations to London, where they were assembled into daily reports. Following the Royal Charter disaster of October 1859, FitzRoy developed a storm-warning service, which began in February 1861 and communicated warnings through coastal signals. His first public newspaper forecast appeared in The Times on August 1, 1861, covering the following 24–48 hours. (science.nasa.gov)

In 1904, Vilhelm Bjerknes formulated forecasting as a problem in mechanics and physics: a sufficiently complete description of the initial atmosphere, together with its governing laws, should permit calculation of its subsequent state. After Bjerknes moved to Bergen in 1917, the Bergen school, including his son Jacob, developed frontal analysis and the polar-front interpretation of midlatitude cyclones. Dense observations helped identify narrow boundaries between warm and cold air and relate them to cyclone development. Lewis Fry Richardson pursued the computational approach in Weather Prediction by Numerical Process (1922), representing atmospheric quantities on a grid and calculating their changes numerically. His proposed “forecast factory” envisaged 64,000 human calculators coordinating computations to keep pace with the weather. (bjerknes.uib.no)

Electronic computing made this programme practicable after the Second World War. In 1950, Jule Charney, Ragnar Fjørtoft, and John von Neumann published an experiment using ENIAC to integrate the barotropic vorticity equation—a deliberately simplified representation of atmospheric flow. Rather than attempting Richardson’s more comprehensive calculation, this experiment demonstrated successful numerical prediction with a tractable dynamical model. (ecmwf.int) Observation expanded alongside computation. Wartime radar operators recognized precipitation echoes, and postwar researchers adapted radar specifically for tracking weather systems. Space-based observation followed: TIROS-1, launched on April 1, 1960, tested television imaging of Earth’s cloud cover. Its 78-day mission demonstrated that satellites could reveal large-scale weather patterns from orbit, extending observation beyond the geographically uneven surface network. (training.weather.gov)

Atmospheric Composition, Structure, and State

Dry air in the lower atmosphere of Earth consists, by volume, of approximately 78.08% nitrogen, 20.95% oxygen, and 0.93% argon. The remainder includes carbon dioxide, neon, helium, methane, ozone, and other trace gases. “Dry air” is a compositional reference that excludes water vapour, not a description of ordinary atmospheric conditions. Water vapour varies from extremely small concentrations in cold or dry environments to approximately 4% by volume in humid tropical air; its presence correspondingly reduces the fractions occupied by other gases. Atmospheric composition also includes aerosols: suspended solid particles and liquid droplets such as mineral dust, sea salt, smoke, and sulfate particles. These constituents must be distinguished from the physical quantities—temperature, pressure, density, and motion—that describe the atmosphere’s changing state. (noaa.gov)

The atmosphere’s principal thermal layers are defined by how temperature changes with altitude, rather than by sharp changes in chemical composition. In the troposphere, temperature generally decreases upward, although local inversions interrupt that pattern. This lowest layer contains most atmospheric water vapour and is the principal setting of clouds, precipitation, and familiar weather. Its upper boundary, the tropopause, varies with latitude, season, and atmospheric conditions: it is generally higher in the tropics, sometimes reaching 18–20 km, and lower toward the poles. Above it, the stratosphere extends to approximately 50 km. Temperature increases through much of this layer because ozone absorbs solar ultraviolet radiation; the resulting stratification inhibits vigorous vertical overturning without eliminating vertical motion. (forecast.weather.gov)

Above the stratopause, temperature decreases again through the mesosphere, which extends to approximately 85 km. The mesopause marks the transition to the thermosphere, where absorption of high-energy ultraviolet radiation and X-rays produces increasing temperatures with height. These high temperatures describe molecular kinetic energy in exceptionally tenuous gas, not the heat content associated with an equally hot, dense gas near the surface. Farther outward, the exosphere grades into space as collisions become increasingly infrequent and some particles escape Earth’s gravitational binding. These thermal boundaries are approximate and variable rather than fixed spherical shells. (noaa.gov)

Meteorological state variables are represented as fields that vary with position and time. Standard reporting units include kelvin or degrees Celsius for temperature and hectopascals for pressure, with $1\ \mathrm{hPa}=100\ \mathrm{Pa}$; density is mass per unit volume, normally expressed in $\mathrm{kg\,m^{-3}}$. Humidity requires a specified measure: specific humidity is the mass of water vapour divided by the total mass of moist air, commonly reported in $\mathrm{g\,kg^{-1}}$. Wind is a vector, expressed through speed and direction or through velocity components. Locations may be specified using latitude, longitude, and geometric height, but upper-air analyses frequently replace height with pressure as the vertical coordinate. (sites.stat.washington.edu)

The approximate vertical balance underlying this pressure coordinate is hydrostatic balance:

where $p$ is pressure, $z$ geometric height, $\rho$ air density, and $g$ gravitational acceleration. The upward pressure-gradient force balances the downward force of gravity, so pressure decreases with height as the weight of overlying air diminishes. This is an excellent approximation for large-scale atmospheric structure, but strong vertical accelerations in thunderstorms and mountain waves require departures from it. A constant-pressure surface is therefore not a constant-height surface: its elevation varies geographically and with time. For example, a reported 500-hPa height of 5,600 m locates that pressure surface above the reporting station, rather than assigning 500 hPa a universal altitude. Upper-air charts commonly express these elevations as geopotential heights, which closely approximate geometric heights while accounting for variations in gravity. The separation between pressure surfaces increases in warmer atmospheric columns and decreases in colder ones. (glossary.ametsoc.org)

Radiation and Atmospheric Energy

Incoming electromagnetic radiation from the Sun supplies most of the energy driving weather. Solar radiation is concentrated at visible and near-infrared wavelengths, with a smaller ultraviolet component; the cooler Earth emits mainly thermal infrared, conventionally called longwave radiation. Absorption transfers radiant energy to matter, whereas emission converts internal energy into radiation. Scattering redirects radiation, producing diffuse skylight; reflection returns part of the incident radiation from surfaces or atmospheric particles. At Earth’s mean orbital distance, solar irradiance on a surface perpendicular to the rays is approximately $1{,}361\ \mathrm{W\,m^{-2}}$. Averaging over the spherical Earth, including its night side, divides this by four, giving approximately $340\ \mathrm{W\,m^{-2}}$. About $100\ \mathrm{W\,m^{-2}}$ is reflected to space and $240\ \mathrm{W\,m^{-2}}$ absorbed by the atmosphere and surface. Outgoing longwave radiation is of comparable magnitude; their difference determines whether the planet gains or loses energy. (earth.gsfc.nasa.gov)

The atmospheric greenhouse effect arises because absorption and emission depend on wavelength. Water vapour, carbon dioxide, and other greenhouse gases absorb terrestrial infrared in characteristic spectral bands and emit infrared both upward and downward. Some surface radiation escapes directly through relatively transparent atmospheric “windows”; much of the radiation reaching space is emitted by atmospheric gases and clouds. Downward atmospheric emission reduces the surface’s net radiative cooling, allowing a warmer surface than would occur without infrared-absorbing gases. This does not reverse the net transfer of heat from warmer to colder regions: upward and downward radiation coexist, while their difference determines net exchange. Atmospheric temperature and the height from which radiation escapes therefore matter alongside gas concentrations. (science.nasa.gov)

Clouds affect both sides of the radiation budget. Their droplets and ice crystals reflect sunlight, but also absorb and emit longwave radiation. Low, optically thick clouds generally exert a strong solar cooling influence; high, cold clouds can substantially reduce outgoing infrared emission. The balance depends on cloud height, thickness, particle properties, and the underlying surface. Aerosols likewise scatter and absorb radiation: sulphate particles predominantly scatter sunlight, whereas black carbon absorbs it and heats the air containing it. Aerosols also modify cloud droplet populations and thus cloud reflectivity. Surface albedo, the fraction of incident solar radiation reflected, provides another control: snow-covered ground generally absorbs less sunlight than dark vegetation or open water under comparable illumination. (ipcc.ch)

For a land surface with negligible energy storage in its thin interface, a useful local budget is

where all terms are fluxes in $\mathrm{W\,m^{-2}}$. Net radiation $R_n$ is downward minus upward shortwave and longwave radiation; $H$ is sensible heat transferred to the air, $LE$ is latent heat transferred through evaporation or transpiration, and $G$ is heat conducted downward into the ground. These signs are conventions, and individual fluxes can reverse. More complete budgets include storage, melting, and horizontal energy transport. Sensible heat exchange changes air temperature directly; evaporation consumes surface energy, while subsequent condensation releases latent heat elsewhere in the atmosphere. Ground heat storage delays the surface temperature response and can supply energy back toward the surface at night. (ecmwf.int)

Atmospheric energy budgets combine radiative heating and cooling with surface sensible heat, latent heating, and energy transport by moving air. Heating is uneven: low latitudes receive a net radiative surplus over the annual cycle, while high latitudes lose more radiation than they absorb. Atmospheric circulation and ocean currents transport energy poleward, reducing this contrast. Rotation imposes a daily solar cycle; axial tilt changes solar elevation and daylight duration seasonally. Land–ocean differences in heat storage further alter the timing and magnitude of heating. These contrasts generate temperature and pressure differences, making part of the unevenly distributed thermal energy available for conversion into kinetic energy of atmospheric motion; friction ultimately dissipates that motion back into internal energy. (gml.noaa.gov)

Thermodynamics, Moisture, and Stability

The equation of state connects atmospheric pressure, density, and temperature. For dry air, $p=\rho R_dT$, where $R_d\approx287\ \mathrm{J\,kg^{-1}\,K^{-1}}$; for moist air without suspended condensate, replacing $T$ with virtual temperature $T_v$ accounts for water vapour’s lower molecular mass. The first law of thermodynamics, expressed per unit mass for dry air, is $\delta Q=c_p\,dT-\alpha\,dp$, where $\alpha=1/\rho$ and $\delta Q$ is heat supplied. During adiabatic ascent, falling ambient pressure allows expansion: the parcel does work and cools without exchanging heat with its surroundings. Adiabatic descent produces compression and warming. Potential temperature,

is the temperature a parcel would attain if brought dry-adiabatically to $p_0=1000\ \mathrm{hPa}$. Its conservation during dry-adiabatic motion permits comparisons between parcels at different pressures. (eoas.ubc.ca)

Several humidity measures describe distinct properties. Vapour pressure $e$ is water vapour’s partial pressure; saturation vapour pressure $e_s(T)$ describes equilibrium with a liquid-water surface, with a separate value for ice. Relative humidity, conventionally referenced to liquid water, is $100e/e_s(T)$ percent. Specific humidity $q$ is the mass of water vapour divided by the total mass of moist air, commonly reported in $\mathrm{g\,kg^{-1}}$. Dew point $T_d$ is the temperature reached by cooling at constant pressure and unchanged water-vapour content until saturation occurs. Because $e_s$ increases rapidly with temperature, cooling can raise relative humidity without adding vapour. Saturation therefore concerns phase equilibrium, not a fixed moisture-storage capacity of air. (eoas.ubc.ca)

An unsaturated rising parcel cools at the dry-adiabatic lapse rate, $\Gamma_d=g/c_p\approx9.8\ \mathrm{K\,km^{-1}}$. Once saturation is reached, condensation releases latent heat, partially offsetting expansion cooling. The saturated, or moist-adiabatic, lapse rate consequently depends on temperature and pressure rather than having one constant value. For example, National Weather Service guidance gives approximate saturated lapse rates through the 700–500-hPa layer of $6.5\ \mathrm{K\,km^{-1}}$ when the 700-hPa temperature is $0^\circ\mathrm{C}$, and $5.6\ \mathrm{K\,km^{-1}}$ when it is $10^\circ\mathrm{C}$. Warmer saturated parcels generally experience greater latent heating because more vapour condenses during ascent. (forecast.weather.gov)

Atmospheric stability concerns a displaced parcel’s buoyancy relative to its environment. Writing the environmental lapse rate as $\Gamma_e=-dT/dz$, a layer is absolutely stable when $\Gamma_e<\Gamma_m$, absolutely unstable when $\Gamma_e>\Gamma_d$, and conditionally unstable when $\Gamma_m<\Gamma_e<\Gamma_d$, where $\Gamma_m$ is the local saturated-adiabatic rate. Conditional instability means that unsaturated ascent is resisted, whereas a saturated parcel can become buoyant after sufficient lifting. A temperature inversion—temperature increasing with height—is strongly stable. Under the dry approximation, increasing potential temperature with height indicates stability, constant potential temperature indicates neutrality, and decreasing potential temperature indicates instability. (eoas.ubc.ca)

Atmospheric soundings display observed temperature and dew point against pressure on thermodynamic diagrams such as the skew-T–log-pressure chart. Close temperature and dew-point traces indicate near-saturation; their separation identifies drier layers. A hypothetical parcel follows a dry adiabat to its lifting condensation level and a moist adiabat above it. Comparing this trajectory with the environmental profile identifies stable caps, the level of free convection, and the equilibrium level. The calculation depends on whether the parcel represents surface air, an average boundary-layer mixture, or an elevated layer. (eoas.ubc.ca)

Convective available potential energy (CAPE) integrates positive parcel buoyancy between the level of free convection and the equilibrium level, in $\mathrm{J\,kg^{-1}}$. Density differences are commonly evaluated using virtual temperature. Convective inhibition (CIN) measures the negative-buoyancy energy barrier below free convection; it may be reported as a negative integral or its positive magnitude. Large CAPE can coexist with a strong cap and no deep convection: lifting, heating, or moisture changes must first enable parcels to overcome inhibition. CAPE is an energy diagnostic, not a direct prediction of storm occurrence or severity. (weather.gov)

Clouds, Precipitation, and Atmospheric Optical Phenomena

A cloud consists of suspended liquid droplets, ice crystals, or both, rather than visible water vapour. Droplets usually form on cloud condensation nuclei: aerosol particles, including sea salt and soluble material in smoke, that take up water. Rising air cools by expansion and can develop supersaturation, in which vapour pressure exceeds its equilibrium value over a flat water surface. Sufficient supersaturation activates suitable particles into growing droplets; particle size and composition influence this threshold. Ice-nucleating particles instead facilitate ice formation. Liquid droplets can remain supercooled below $0\,^\circ\mathrm{C}$, coexisting with ice in mixed-phase clouds; without effective ice-nucleating particles, homogeneous freezing becomes important near $-38\,^\circ\mathrm{C}$. Condensation and freezing are therefore distinct processes, not interchangeable consequences of reaching saturation. (noaa.gov)

Cloud classification combines appearance with altitude. The Latin-based nomenclature originated with Luke Howard’s 1803 classification; the World Meteorological Organization recognizes ten genera. In temperate regions, high clouds typically occupy about 5–13 km: fibrous cirrus, sheetlike cirrostratus, and small-element cirrocumulus. Middle-level clouds, commonly at 2–7 km, include altocumulus and altostratus; low clouds, generally below 2 km, include stratus and stratocumulus. These ranges overlap and vary with latitude. Nimbostratus commonly extends across several levels, while cumulus and cumulonimbus have low bases but potentially great vertical extent. Layered clouds commonly accompany widespread ascent or cooling within stable layers; cumulus develops in buoyant rising parcels, and deeper convection produces cumulonimbus. Orographic lifting can generate clouds where moist air ascends mountain slopes. (public.wmo.int)

Cloud particles must grow substantially to produce precipitation: a representative cloud droplet is about 0.02 mm across, compared with roughly 2 mm for a raindrop—a millionfold difference in volume. In collision–coalescence, larger droplets fall faster than smaller ones, collide with them, and sometimes merge, accelerating growth. This process can produce rain without ice. In mixed-phase clouds, saturation vapour pressure is lower over ice than over supercooled liquid water. Consequently, ice crystals can grow by vapour deposition while nearby droplets evaporate, the Bergeron–Findeisen process. Ice particles also grow by aggregation of crystals and by riming, in which collected supercooled droplets freeze onto them. (weather.gov)

The thermal and moisture structure beneath a cloud determines what reaches the surface. Ice crystals and aggregates surviving descent form snow, whereas sufficient melting produces rain. A warm layer aloft can melt falling snow; a sufficiently deep, cold layer below can refreeze the drops into ice pellets, termed sleet in United States usage. With insufficient cooling time to freeze before impact, supercooled drops reach exposed surfaces as freezing rain and form an ice coating. Surface temperature alone therefore does not determine precipitation type. During a December 1, 2007 event documented by the National Weather Service, variations in the intervening warm layer produced rain, freezing rain, sleet, and snow across the affected region. (weather.gov)

Fog forms when droplets develop in air near the surface. Radiation fog follows nocturnal surface cooling, especially under clear skies and light winds; advection fog forms when moist air moves over a colder surface. Evaporation from warm water followed by mixing with colder air can produce steam fog. Dew differs because condensation occurs directly on a surface cooled below the adjacent air’s dew point, rather than creating suspended droplets. Frost can form by direct vapour deposition onto a surface below its frost point and is not necessarily frozen dew. (weather.gov)

Cloud and precipitation particles also produce optical phenomena. A primary rainbow results from wavelength-dependent refraction, internal reflection, and renewed refraction of sunlight in droplets, forming an arc roughly $42^\circ$ from the antisolar direction. Ice-crystal halos instead surround the Sun or Moon: the common halo has an angular radius near $22^\circ$. Crystal orientation affects the display; refraction through suitably oriented crystals produces sundogs beside the Sun, whereas reflection from falling ice crystals produces vertical sun pillars. (weather.gov)

Atmospheric Dynamics and Governing Equations

Atmospheric models express conservation of mass, momentum, and energy, together with budgets for water vapour and condensed water. In a rotating reference frame, compact forms of the mass and momentum equations are

Here $\rho$ is density, $p$ pressure, $\mathbf{u}$ velocity, $\boldsymbol{\Omega}$ Earth’s rotation vector, $\mathbf{g}$ effective gravity including centrifugal acceleration, and $\mathbf{F}$ frictional acceleration. The material derivative $D/Dt=\partial/\partial t+\mathbf{u}\cdot\nabla$ follows moving air, incorporating both local change and transport. An equation of state closes the relationship among pressure, density, and temperature. Hydrostatic formulations replace vertical acceleration with pressure–gravity balance; nonhydrostatic formulations retain it. (ecmwf.int)

Energy conservation is commonly represented through a thermodynamic equation. For dry ideal air,

where $c_p$ is specific heat at constant pressure and $\dot Q$ is heating per unit mass. This distinguishes temperature changes caused by compression or expansion from diabatic heating. A schematic moisture equation, $Dq_v/Dt=S_v$, describes water-vapour transport, with $S_v$ representing phase changes and unresolved mixing. Condensation removes vapour while releasing latent heat; evaporation reverses these transfers. Separate cloud-water, ice, and precipitation budgets account for exchanges among water species and sedimentation, rather than treating condensation as destruction of total water. (ecmwf.int)

The pressure-gradient force accelerates air toward lower pressure, while gravity acts predominantly downward. The Coriolis effect deflects horizontal motion rightward in the Northern Hemisphere and leftward in the Southern Hemisphere without directly changing its speed. For slowly evolving, large-scale flow away from the equator, horizontal pressure-gradient and Coriolis accelerations approximately cancel:

This geostrophic wind flows parallel to isobars on a horizontal surface. Gradient-wind balance additionally accounts for the centripetal acceleration of curved flow. The difference between actual and geostrophic velocity is the ageostrophic wind, important in accelerating flows and circulations associated with jet streams. Surface drag and turbulent momentum exchange disrupt geostrophic balance, allowing near-surface winds to cross isobars toward lower pressure. (weather.gov)

Vorticity and divergence diagnose different aspects of circulation. Relative vorticity, $\boldsymbol{\omega}=\nabla\times\mathbf{u}$, measures local rotation; divergence, $\nabla\cdot\mathbf{u}$, measures local spreading. Mass continuity links horizontal convergence and divergence to vertical transport and density changes. Potential vorticity combines rotation with stratification: dry Ertel potential vorticity is $P=\rho^{-1}(\boldsymbol{\omega}+2\boldsymbol{\Omega})\cdot\nabla\theta$, where $\theta$ is potential temperature. It is materially conserved in adiabatic, frictionless flow, making its anomalies useful tracers of disturbances. Under suitable balance assumptions and boundary conditions, potential-vorticity inversion diagnoses associated wind and temperature fields. Thermal-wind balance, obtained by combining hydrostatic and geostrophic balance, relates vertical geostrophic shear to horizontal temperature gradients; “thermal wind” denotes that shear, not an additional wind. (ocw.mit.edu)

Internal gravity waves occur when buoyancy restores vertically displaced air in a stably stratified atmosphere. Their propagation transports energy and momentum; flow over mountains provides one source. Rossby waves instead depend on a background potential-vorticity gradient, commonly associated with the latitude dependence of $f$. Their phase propagates westward relative to a uniform background current, although sufficiently strong westerlies can carry their pattern eastward over the ground. Their group velocity, describing wave-packet and energy propagation, need not share the phase velocity’s direction. (ocw.mit.edu)

Instabilities allow small disturbances to extract energy from a background circulation. Baroclinic instability draws on available potential energy associated with horizontal temperature contrasts and vertically sheared flow. Growing disturbances transport heat poleward and convert available potential energy into eddy kinetic energy, providing a central mechanism for midlatitude cyclone development. Eric Eady’s 1949 idealized model demonstrated this growth in a uniformly sheared, stably stratified layer. Barotropic instability, by contrast, extracts kinetic energy from horizontal shear; atmospheric disturbances can involve both mechanisms. (ocw.mit.edu)

Boundary Layers and Local Circulations

The atmospheric boundary layer is the lowest part of the atmosphere directly influenced by the underlying surface. Its depth ranges from a few metres in strongly stable conditions to several kilometres under vigorous mixing. Surface drag transfers atmospheric momentum to the ground, while wind shear generates turbulence that redistributes this influence vertically. Buoyant thermals provide another source of turbulence when the surface heats the air above it. Together, these motions exchange sensible heat, water vapour, and momentum between the surface and the atmosphere; evaporation also transfers energy as latent heat. Boundary-layer turbulence therefore controls how surface conditions affect temperature, humidity, wind, and pollutant concentrations above the ground. (metoffice.gov.uk)

Over land on sunny days, surface heating produces a convective mixed layer in which potential temperature becomes approximately uniform above the shallow surface layer. Rising thermals penetrate the stable air at its upper boundary and incorporate some of that air into the mixed layer, a process called entrainment. This both modifies the layer’s properties and helps it deepen. Around sunset, declining surface heat input removes the buoyant energy source. Radiative cooling of the ground then cools the lowest air, producing a stable layer, commonly with temperature increasing upward. The former daytime mixed layer remains above as a residual layer, increasingly disconnected from surface exchanges. Stability suppresses vertical mixing, although wind shear can maintain turbulence. After sunrise, renewed heating erodes the near-surface inversion and reconnects the growing mixed layer with the residual air. (weather.gov)

Sea and land breezes arise from differential heating between land and water. During daytime, warmer air over land supports a circulation with near-surface flow from sea to land, ascent inland, and a return flow aloft. The advancing sea-breeze front concentrates convergence and can initiate clouds or thunderstorms. In Florida, prevailing winds strongly influence how far this boundary penetrates across the peninsula and where thunderstorms develop. At night, faster cooling of land can reverse the thermal contrast and produce an offshore land breeze, generally weaker and shallower than its daytime counterpart. These circulations are most distinct when larger-scale pressure gradients are weak; strong background winds can displace or suppress them. (noaa.gov)

Mountainous terrain produces both thermally driven circulations and mechanically forced flows. Daytime heating drives upslope winds, while broader valley circulations commonly carry air up-valley; nocturnal cooling produces downslope drainage and down-valley flow, allowing cold air to accumulate in sheltered basins. These cold drainage winds differ from warm, dry downslope winds such as the Alpine foehn and Rocky Mountain chinook, in which descending air warms through adiabatic compression. Strong cross-mountain flow in a stable atmosphere can also generate nearly stationary mountain waves extending downstream. Clouds may mark ascending portions while evaporating in descending portions: satellite imagery on October 18, 2004, showed an extensive wave-cloud field over the Four Corners region of the United States. Wave breaking and low-level rotors can produce intense turbulence. (repository.library.noaa.gov)

Surface characteristics alter these exchanges even without pronounced coastal or mountain circulations. Buildings and forests create aerodynamic drag and mechanically generated turbulence, while urban street geometry can shelter particular locations from ventilation. An urban heat island develops through several interacting mechanisms: reduced vegetation and evaporation, absorption and storage of solar energy by construction materials, restricted radiative cooling between buildings, and heat released by human activities. Stored energy is released after sunset, often making the urban–rural air-temperature contrast especially apparent at night. Vegetation provides shade and evapotranspirational cooling; soil moisture determines how much water is available for this energy transfer. Consequently, replacing moist vegetation with dry, impervious surfaces changes both the local heat budget and the moisture supplied to boundary-layer air. Clear skies and light winds commonly favour stronger urban heat islands. (metoffice.gov.uk)

Global Circulation and Tropical Meteorology

The Hadley, Ferrel, and polar cells describe idealized, latitude–height overturning circulations averaged around the globe, rather than three continuous conveyor belts. In the Hadley circulation, air rises within tropical rain belts, flows poleward aloft, and descends in the subtropics, typically near 20°–35° latitude. Its equatorward surface return flow is deflected into northeasterly trade winds in the Northern Hemisphere and southeasterly trades in the Southern Hemisphere. Their convergence helps maintain the intertropical convergence zone (ITCZ), a seasonally migrating, discontinuous belt of deep convection. The Ferrel cell occupies the middle latitudes, with poleward flow near the surface and equatorward flow aloft; the polar cell has descending air over high latitudes and an equatorward surface branch associated with polar easterlies. Actual circulation departs substantially from this pattern because of continents, oceans, seasons, and individual weather systems. (noaa.gov)

Midlatitude surface westerlies and upper-tropospheric jet streams accompany this circulation. The subtropical jets occur near the poleward margins of the Hadley cells, where poleward-moving upper-level air acquires a strong eastward velocity relative to the surface through approximate conservation of angular momentum. Polar-front jets are associated with strong horizontal temperature contrasts farther poleward. Both jet systems meander, shift seasonally, and sometimes merge; their positions are not fixed boundaries between cells. (noaa.gov)

Monsoons are seasonal reorganizations of winds and rainfall, not simply episodes of heavy rain. Summer continental heating changes pressure patterns and draws moist oceanic air inland; convergence and ascent then support widespread convection. In the North American monsoon, for example, summer circulation transports moisture from the Pacific Ocean and Gulf of California toward northwestern Mexico and the southwestern United States. (climate.gov) The Walker circulation instead describes east–west tropical overturning. Across the equatorial Pacific, surface easterlies push warm water westward, supporting deep convection over the western warm pool, eastward return flow aloft, and subsidence over the cooler eastern Pacific. Wind-driven upwelling and a shallower eastern thermocline help preserve the sea-surface temperature contrast, coupling atmospheric circulation to the ocean. (pmel.noaa.gov)

Tropical disturbances become tropical cyclones only when convection organizes around a persistent low-level rotating circulation. Favourable environments combine moist middle-tropospheric air, weak vertical wind shear, a pre-existing disturbance, and sufficient planetary rotation; formation is therefore uncommon within roughly 5° of the equator. Ocean temperatures around 26.5°C over a sufficiently deep warm layer provide a conventional indicator of favourable conditions, rather than a guarantee of development. (aoml.noaa.gov) Ocean-to-atmosphere heat and moisture transfer, followed by latent heat release in convection, sustains the circulation. Intense storms develop spiral rainbands and an eyewall containing their strongest winds, surrounding a relatively calm eye. (aoml.noaa.gov) Environmental winds steer their movement, often westward on the equatorward side of subtropical ridges and poleward or eastward where steering currents change. Strong shear, dry-air intrusion, and land interaction can weaken storms; eyewall replacement can also temporarily reduce intensity even over warm water. (aoml.noaa.gov)

Tropical circulation also varies within and between seasons. The Madden–Julian Oscillation, first described in 1971, is an eastward-moving disturbance of convection, winds, and pressure with a typical recurrence interval of 30–60 days; it changes tropical rainfall and influences weather beyond the tropics. (psl.noaa.gov) El Niño–Southern Oscillation involves coupled changes in equatorial Pacific temperatures and atmospheric circulation. During El Niño, weakened trades and reduced cold-water upwelling accompany warming farther east and displacement of tropical convection; La Niña generally strengthens the trades and Walker circulation. Changes in tropical heating alter remote upper-level circulation and jet streams, linking Pacific variability to regional rainfall and storm-track changes rather than prescribing identical weather in every event. (weather.gov)

Synoptic Weather Systems and Organized Convection

Air masses are extensive bodies of air with relatively uniform temperature and humidity, acquired over source regions: continental polar air is characteristically cold and dry, whereas maritime tropical air is warm and moist. Fronts are transition zones between contrasting air masses. A cold front advances as colder air replaces warmer air; a warm front advances as warmer air replaces colder air; a stationary front has little movement. An occluded front forms when a cold front overtakes a warm front, lifting the intervening warm air away from the surface. Surface charts conventionally show cold fronts with blue triangles, warm fronts with red semicircles, stationary fronts with these symbols on opposite sides, and occlusions with purple triangles and semicircles on the same side. Temperature contrasts, wind shifts, and pressure troughs help locate these boundaries. (noaa.gov)

An extratropical cyclone develops through interaction between a low-level temperature boundary and an upper-air disturbance. In the idealized Norwegian cyclone model, formulated during the 1910s and 1920s, a frontal wave develops into a low with distinct warm and cold fronts, followed by occlusion. Surface maps reveal the low and its fronts, while upper-air charts show the accompanying trough and jet stream; these are parts of a three-dimensional system rather than independent features. Cyclones commonly intensify near and downstream of upper-level troughs, and the large-scale flow helps determine their storm tracks. Anticyclones are relative pressure maxima associated with descending air and reduced cloud formation. Atmospheric blocking occurs when a persistent high or amplified ridge disrupts the usual eastward progression of weather systems. An omega block has a high between two lows; such configurations can maintain dry conditions beneath the high and repeated precipitation around its margins for days or weeks. (prod-01-alb-www-noaa.woc.noaa.gov)

The organization of thunderstorms depends on instability, moisture, lifting, and vertical wind shear—the change of wind with height. An ordinary single-cell storm typically lasts 30–60 minutes: an initial updraft develops precipitation, then a downdraft whose outflow eventually cuts off the warm, moist supply. Multicells persist through successive updrafts initiated along the advancing gust front, even though individual cells remain short-lived. A supercell instead contains a persistent rotating updraft, or mesocyclone. Strong vertical shear supports this organization, allowing the storm to sustain its inflow rather than being rapidly overwhelmed by its own precipitation-cooled outflow. Supercells can produce large hail, damaging winds, and tornadoes, but rotation does not imply that a tornado will occur. (nssl.noaa.gov)

Squall lines arrange multiple cells along an elongated boundary and are especially associated with damaging straight-line winds. Their rain-cooled outflow lifts fresh air at the leading edge, continually regenerating convection. A mesoscale convective system (MCS) is a broader organizational category: a connected collection of thunderstorms acting as a system, often extending at least 100 km in one direction. It may be linear or clustered and persist for several hours; a mesoscale convective complex is a particular large, long-lived, approximately circular form identified in satellite imagery. Thus, “squall line,” “supercell,” and “MCS” describe different organizational properties and are not wholly exclusive categories. (nssl.noaa.gov)

Lightning results from electrical charge separation and discharge. Collisions between graupel and smaller ice particles in the presence of supercooled droplets transfer charge; different particle motions then separate charged regions. The detailed initiation of lightning remains incompletely resolved. Hail grows when supercooled droplets freeze onto ice particles retained within strong updrafts, falling when support weakens or the stones become too large. Damaging convective winds usually arise from downdrafts spreading outward at the ground, rather than from tornadoes. (nssl.noaa.gov)

A tornado is a rotating column of air connected to a convective cloud and the ground. In supercells, vertical shear supplies horizontal rotation that the updraft can tilt and stretch, helping establish the mesocyclone. Tornadogenesis nevertheless requires additional near-ground processes: the interaction of circulation, downdrafts, buoyancy, and stretching remains an active research problem, and a mesocyclone does not simply extend downward to become a tornado. Tornadoes also occur without supercells, demonstrating that a persistent deep rotating updraft is not a universal prerequisite. (noaa.gov)

Instruments and Observing Networks

Surface stations measure atmospheric conditions with instruments whose exposure is as important as their sensing principle. Platinum-resistance thermometers infer temperature from electrical resistance; ventilated screens shield them from direct sunlight while allowing air circulation. The UK Met Office places screen thermometers 1.25 m above ground. Electronic barometers measure pressure, while capacitive humidity probes respond to water-vapour-dependent changes in a polymer’s electrical properties. Cup anemometers measure wind speed through rotation, vanes indicate direction, and ultrasonic anemometers determine airflow from differences in acoustic travel time. Standard surface wind observations commonly use instruments at 10 m, away from significant obstructions. These heights provide comparable reference exposures rather than measurements representative of every nearby surface. (metoffice.gov.uk)

Precipitation gauges collect water over a known aperture: tipping buckets register successive increments, whereas weighing gauges record changes in collected mass. One Met Office tipping-bucket design records 0.2 mm per tip; this is its measurement increment, not a statement of overall accuracy. Snow blockage, evaporation and exposure can affect collected totals. Pyranometers measure incoming solar irradiance on a plane, normally expressed in watts per square metre, and require an unobstructed exposure. Calibration establishes relationships between instrument indications and reference values, but does not eliminate field errors. Measurement uncertainty also reflects drift, response time, environmental influences and sampling; a precisely calibrated sensor beside a heated wall can still provide an unrepresentative air-temperature observation. (metoffice.gov.uk)

Balloon-borne radiosondes transmit temperature, humidity and pressure observations during ascent, with winds derived from tracking their position, commonly using the Global Positioning System. Routine soundings are coordinated around 00:00 and 12:00 UTC. A flight can exceed two hours, reach above 35 km and drift more than 300 km, so its profile is neither instantaneous nor vertically above one location. Commercial aircraft complement soundings with temperature and wind measurements at cruising altitude and during ascent and descent, although coverage follows flight routes and airport activity. Ships and moored or drifting buoys supply marine observations, including atmospheric pressure and sea-surface temperature. Research aircraft add targeted sampling: NOAA hurricane missions deploy parachuted dropsondes to measure descending profiles of pressure, temperature, humidity and wind. (weather.gov)

Weather radar distinguishes several observables. Reflectivity characterizes returned signal strength and depends on the number, size and electromagnetic properties of scattering particles; converting it to rainfall requires additional assumptions. Doppler velocity measures motion toward or away from the antenna, not the complete wind vector. Dual-polarization radar compares horizontally and vertically polarized signals: differential reflectivity indicates particle shape tendencies, correlation coefficient measures similarity between the returns, and specific differential phase describes differential phase change along the propagation path. Together these help distinguish rain, snow, hail and non-meteorological echoes. The United States completed the dual-polarization upgrade of its operational WSR-88D fleet in spring 2013. (noaa.gov)

Satellite remote sensing extends coverage over oceans and sparsely instrumented land. Geostationary satellites repeatedly view the same broad region, whereas polar-orbiting satellites sample successive swaths, including high latitudes. Visible imagery records reflected sunlight; thermal-infrared imagery detects emitted radiation from surfaces, clouds and atmospheric layers. Sounders use multiple spectral channels with different atmospheric sensitivities to infer vertical structure. NOAA’s Cross-track Infrared Sounder and Advanced Technology Microwave Sounder provide complementary measurements: microwave channels can obtain information through clouds that obstruct infrared observations. Radio occultation instead measures changes in navigation-satellite signals as their paths cross the atmospheric limb; signal delays and bending yield profiles from which temperature and humidity can be retrieved. These are indirect measurements requiring physical interpretation, not thermometer readings at discrete heights. (noaa.gov)

Observational coverage therefore depends on both geography and measurement geometry. Radiosondes resolve vertical structure along a drifting trajectory but leave large intervals between launches; aircraft concentrate observations along transport corridors. Radar supplies frequent regional measurements, yet its beam broadens and rises with distance, potentially overshooting shallow precipitation or low-level circulation. Terrain blockage and anomalous atmospheric refraction further alter coverage or create misleading echoes. Satellite coverage is much broader, but individual channels integrate radiation over atmospheric layers, and retrievals depend on cloud treatment and assumptions about atmospheric and surface properties. Direct and remotely sensed observations consequently constrain different aspects of the atmosphere rather than serving as interchangeable measurements. (weather.gov)

Atmospheric Analysis, Data Assimilation, and Reanalysis

Observation quality control checks whether reports are physically plausible, internally consistent, and compatible with neighbouring stations, earlier measurements, and short-range forecasts. NOAA’s Meteorological Assimilation Data Ingest System (MADIS), for example, attaches flags recording validity, temporal-consistency, and spatial-consistency tests rather than treating every report as equally reliable. Station metadata—including coordinates, elevation, observing schedules, relocations, and measurement techniques—help distinguish atmospheric changes from changes in observing practice. Systematic errors require additional treatment: variational bias correction can estimate satellite-radiance biases alongside the atmospheric state, although model bias can contaminate these estimates without suitable observational constraints. (madis.ncep.noaa.gov)

Atmospheric analysis organizes observations and estimated fields into representations of weather structure. Surface and upper-air maps show horizontal distributions, while vertical cross-sections expose changes in temperature, moisture, and wind across a front or other feature. Sounding analyses commonly use skew-$T$, log-pressure diagrams to plot temperature and dew point against pressure, with winds alongside; these reveal inversions, moist layers, and vertical wind shear. Cross-sections assembled from multiple profiles or gridded analyses reveal structures that a single sounding cannot locate horizontally. Such displays are diagnostic representations, not additional independent observations. (weather.gov)

Data assimilation combines observations with a model-based background, usually a short forecast from the preceding analysis. The resulting analysis estimates the atmospheric state while accounting for uncertainties in both information sources. Observation operators translate model variables into measured quantities—for example, calculating satellite radiances from atmospheric profiles—so assimilation need not first convert every measurement into a retrieved temperature or humidity. Error covariance matrices determine how strongly observations modify the background and how corrections extend between locations and variables. Repeated analysis–forecast cycling carries information forward: each analysis initializes a forecast that supplies the next background, rather than reconstructing the atmosphere independently at every observation time. (ecmwf.int)

Variational methods formulate this estimation as mathematical optimization, minimizing weighted departures from observations and the background. Three-dimensional variational assimilation, or 3D-Var, estimates a state at an analysis time; four-dimensional variational assimilation, or 4D-Var, adjusts a model trajectory over an assimilation window and compares observations at their respective times. ECMWF introduced operational 4D-Var on November 25, 1997, initially with a six-hour window, subsequently extended to twelve hours. This treatment is particularly important for irregularly timed aircraft and satellite reports. Strong-constraint 4D-Var assumes a perfect model within the window; weak-constraint formulations additionally represent model error. (ecmwf.int)

Ensemble methods instead estimate evolving error relationships from differences among multiple model realizations. Ensemble Kalman filters use these sampled covariances to update the atmospheric state, making corrections dependent on the current flow rather than exclusively on fixed statistical relationships. Limited ensemble size introduces sampling errors, commonly controlled through covariance localization, which reduces spurious long-distance relationships. Hybrid systems combine ensemble-derived information with variational methods or more stable climatological covariance estimates. A concrete operational example was the Met Office’s introduction of a hybrid 4D-Var/Ensemble Transform Kalman Filter system in July 2011. (ecmwf.int)

Atmospheric reanalysis applies a substantially consistent forecasting and assimilation system retrospectively to historical observations, reducing discontinuities caused by operational model upgrades. ECMWF’s ERA5 provides hourly estimates from 1940, with approximately 31-kilometre horizontal spacing and 137 atmospheric levels; its 1940–1958 extension was released on March 7, 2023. Reanalysis is nevertheless a model-constrained reconstruction, not a direct observational record. Early Southern Hemisphere observations were sparse, and changing observational coverage alters how strongly the model is constrained. Model assumptions and assimilation errors can also produce artificial discontinuities: ERA5.1 was specifically produced for 2000–2006 to improve a temperature discontinuity in the stratosphere and uppermost troposphere. (ecmwf.int)

Numerical Weather Prediction

Numerical weather prediction treats forecasting as an initial-value problem: an estimated atmospheric state at time $t_0$ is advanced using equations for motion, mass conservation, thermodynamics, and moisture. Data assimilation supplies the initial state; the forecast model then calculates its evolution. Spatial discretization replaces continuous partial differential equations with a finite computational representation. Grid-point models use methods such as finite differences, whereas spectral-transform models also represent fields through spatial basis functions. Time integration advances these quantities through successive time steps. For example, the Met Office’s ENDGame dynamical core uses finite differences on a latitude–longitude grid, semi-Lagrangian transport, and semi-implicit time stepping; ECMWF’s Integrated Forecasting System combines grid-point calculations with spectral transforms. These choices affect accuracy, numerical stability, and computational cost. (metoffice.gov.uk)

Global models cover the entire atmosphere and therefore require no lateral boundary conditions. Limited-area models concentrate computational resources on a region but require time-varying information at their lateral boundaries, usually from a larger-domain forecast. Nesting places a finer grid within a coarser parent domain: one-way nesting transfers information inward, while two-way nesting also feeds the fine-grid solution back to the parent. Surface conditions provide the lower boundary, either prescribed or calculated through interacting surface models. The Met Office introduced a 1.5 km convection-permitting UK model in 2010, illustrating how regional grids can represent terrain and convective structures more explicitly than coarser global configurations. (www2.mmm.ucar.edu)

Hydrostatic and nonhydrostatic formulations differ in their treatment of vertical momentum. Hydrostatic models approximate vertical pressure-gradient force as balancing gravity, neglecting vertical acceleration in that balance. This is effective for large-scale atmospheric motions. Nonhydrostatic models retain vertical acceleration and can represent dynamically important departures from hydrostatic balance in convective updrafts and small-scale mountain flows. The distinction is not identical to global versus regional coverage: the Unified Model’s nonhydrostatic equations serve both configurations, whereas ECMWF’s IFS atmospheric dynamical core uses a hydrostatic formulation. (ecmwf.int)

Parameterizations calculate the effects of processes not explicitly represented by the model’s resolved dynamics. Turbulence schemes estimate exchanges of momentum, heat, and moisture; convection schemes represent collective cloud transports; cloud microphysics schemes calculate transformations among vapour, droplets, and precipitation particles. Radiation schemes calculate atmospheric heating and cooling, while surface schemes represent exchanges involving soil, vegetation, snow, and water. Increasing resolution changes this division of labour rather than eliminating parameterization. Kilometre-scale models can represent much deep convection through resolved motion, but still require treatments of cloud microphysics, small turbulent eddies, and subgrid variability. Radiation also requires a separate physical calculation rather than emerging automatically from finer fluid-dynamical grids. (ecmwf.int)

Coupled prediction evolves the atmosphere alongside ocean, land, and sea-ice components, exchanging surface conditions and fluxes. Soil moisture and snow influence surface energy and water exchanges; ocean mixing changes sea-surface temperature and consequently atmospheric heat supply. In ECMWF experiments involving Typhoons Haiyan and Neoguri, coupling produced different responses because their underlying warm ocean layers differed: Neoguri depleted a shallow warm layer, producing stronger surface cooling and reduced ocean-to-atmosphere heat flux than an uncoupled forecast. Such feedbacks make the lower boundary an evolving part of the prediction rather than a fixed input. (ecmwf.int)

Operational design balances grid spacing, vertical resolution, physical complexity, forecast duration, ensemble size, and delivery deadlines. Finer grids increase the number of calculations and can require shorter time steps; additional interacting components and ensemble members add further integrations. Parallel computing distributes this workload, although communication between processors also affects efficiency. On June 27, 2023, ECMWF increased its medium-range ensemble’s horizontal resolution from approximately 18 km to 9 km. In the same upgrade, extended-range forecasts used a separate, coarser configuration, allowing more members and daily production without imposing medium-range resolution throughout the longer integration. (metoffice.gov.uk)

Forecast Products, Predictability, and Verification

Forecast ranges describe overlapping operational categories rather than fixed limits of predictability. Nowcasting covers approximately 0–6 hours ahead, emphasizing local conditions and rapidly updated observations. Persistence assumes that present conditions continue; extrapolation moves observed features, such as radar rainfall patterns, along estimated trajectories. These methods are useful initially but cannot fully represent storm initiation, growth, or decay, so nowcasting increasingly blends observations with numerical guidance. Short-range forecasts generally cover the next 1–3 days, while medium-range forecasts extend roughly from day 3 to day 10, sometimes farther depending on the service. Both rely heavily on numerical predictions interpreted and processed into location-specific products. (space.oscar.wmo.int)

Subseasonal prediction commonly addresses about two weeks to two months ahead; seasonal products describe monthly or three-month averages and category probabilities. Their targets differ from a forecast of tomorrow’s rain: they may estimate whether a coming season’s mean temperature will fall in the upper third of its historical distribution, rather than specify individual hot days. Slowly varying components of the Earth system provide predictive information after detailed atmospheric trajectories become uncertain. For example, El Niño–Southern Oscillation alters tropical heating and atmospheric circulation, shifting regional temperature and rainfall probabilities without determining every weather event. (wmo.int)

Atmospheric predictability is limited by chaotic error growth: initially small differences can amplify through nonlinear dynamics, at rates that depend on the circulation and spatial scale. Initial-condition uncertainty arises from incomplete, imperfect observations and their interpretation; model uncertainty arises from approximations in dynamics, numerical methods, and unresolved physical processes. These sources interact because models also help construct the initial atmospheric analysis. Martin Leutbecher and Tim Palmer’s 2007 review Ensemble Forecasting describes prediction as estimating a flow-dependent distribution of possible outcomes, rather than treating one calculated trajectory as certain. Loss of skill in locating individual storms does not necessarily eliminate skill in predicting larger-scale patterns or time-averaged statistical tendencies. (ecmwf.int)

An ensemble forecast samples possible evolutions using members with perturbed initial conditions and representations of model uncertainty. A raw event probability can be estimated by the fraction of members exceeding a specified threshold: illustratively, 12 of 40 members predicting more than 10 mm of rain in 24 hours gives 30%. Ensemble spread measures disagreement, often through a standard deviation or quantile range; it is an uncertainty indicator, not a guaranteed bound on the observed outcome. Statistical post-processing uses archived forecast–observation pairs, including retrospective forecasts made with a consistent model version, to correct systematic bias and adjust predictive distributions. Probabilistic calibration means that, across sufficiently many comparable cases assigned 30% probability, approximately 30% experience the event. Calibration does not mean that each individual outcome can be predicted correctly. (ecmwf.int)

Forecast verification compares predictions with suitable reference observations or analyses, matching location, valid time, accumulation period, and spatial scale. Reference uncertainty and differences between point measurements and grid-cell quantities affect interpretation. For $N$ paired forecasts $f_i$ and observations $o_i$, mean bias is $N^{-1}\sum_i(f_i-o_i)$, while root-mean-square error is $\sqrt{N^{-1}\sum_i(f_i-o_i)^2}$. Bias measures systematic overprediction or underprediction but permits positive and negative errors to cancel; RMSE measures overall error magnitude and gives greater weight to large errors. Threshold events are evaluated using hits, misses, false alarms, and correct rejections. The critical success index, $H/(H+M+F)$, excludes correct rejections, avoiding the misleadingly high accuracy obtainable by always predicting that a rare event will not occur. (cawcr.gov.au)

Probability forecasts require complementary diagnostics. Reliability diagrams compare issued probabilities with observed event frequencies; the Brier score, $N^{-1}\sum_i(p_i-y_i)^2$, measures squared probability error, with $y_i=1$ for occurrence and $0$ otherwise. Lower values are better. Skill expresses improvement over an explicit baseline, such as climatological probabilities or persistence: for a lower-is-better score $S$, $1-S/S_{\mathrm{ref}}$ is positive when the forecast outperforms the reference and negative when it performs worse. Evaluation must therefore specify the variable, region, season, lead time, and reference forecast; no single score establishes universal forecasting superiority. (confluence.ecmwf.int)

Atmospheric Chemistry and Earth-System Connections

Atmospheric constituents are redistributed by winds, turbulent mixing, and convection while undergoing chemical reactions and removal at the surface. Ground-level ozone is a secondary pollutant: it forms through sunlight-driven chemistry involving nitrogen oxides and volatile organic compounds rather than being emitted directly. Its concentrations therefore depend on radiation and atmospheric transport as well as precursor emissions. Sulfur dioxide and nitrogen oxides undergo oxidation to produce sulfuric and nitric acids and associated particles. Wet deposition removes soluble gases and particles through precipitation; dry deposition transfers them to vegetation, soil, water, and other surfaces without precipitation. These processes connect air pollution to the chemistry of terrestrial and aquatic ecosystems. (epa.gov)

Mineral dust, sea salt, wildfire smoke, and particles formed from gaseous precursors constitute atmospheric aerosols. Their effects depend on size, composition, altitude, and the surrounding meteorological conditions. Particles scatter or absorb sunlight, and some act as nuclei for cloud droplets or ice crystals, connecting atmospheric chemistry to cloud microphysics and the Earth’s energy budget. Absorbing smoke can warm the atmospheric layer containing it while reducing sunlight at the surface. Transport can extend these effects far beyond emission regions: in June 2020, an exceptionally large Saharan dust plume crossed the Atlantic and affected air quality in the Caribbean and the United States. Its evolution illustrates why local pollutant concentrations cannot be explained from local emissions alone. (science.nasa.gov)

The water cycle couples atmospheric motion to water storage on land and in the ocean. Evaporation and plant transpiration, together termed evapotranspiration, return water to the atmosphere; transport and condensation subsequently redistribute it as precipitation. Rain and snow replenish soil moisture, groundwater, rivers, and seasonal snowpacks. Snow accumulation delays water release until melting, separating the timing of precipitation from that of runoff and river discharge. Vegetation, radiation, humidity, wind, and soil-water availability regulate evapotranspiration. Consequently, dry soils can restrict evaporative cooling, directing more available surface energy into sensible heating and reinforcing hot conditions. Conversely, moist land can sustain evaporation and supply atmospheric moisture, although the resulting precipitation also depends on circulation and atmospheric stability. (science.nasa.gov)

Exchanges with other Earth-system components operate over different timescales. The ocean exchanges heat, moisture, and momentum with the atmosphere: winds drive surface currents, while sea-surface temperature influences evaporation and atmospheric heating. Ocean heat storage also moderates the response to sustained changes in the planetary energy budget; the ocean has absorbed more than 90% of the excess heat associated with global warming. Vegetation and soils regulate moisture exchanges, while the cryosphere influences both water storage and surface reflectivity. Snowfall or melting can immediately alter near-surface temperatures and runoff. Over longer periods, declining snow and ice cover can lower albedo, increase absorbed solar energy, and amplify warming. These feedbacks differ from the short-lived atmospheric response to a particular warm ocean surface or snow-covered landscape. (pfeg.noaa.gov)

Weather research and climate modelling share conservation laws and representations of radiation, clouds, turbulence, and surface exchanges. The Met Office Unified Model, for example, uses a common dynamical core and, where possible, common parameterizations across weather and climate applications. Their questions nevertheless differ: an event forecast estimates the evolving atmospheric state from specified initial conditions, whereas a climate projection estimates statistical changes under specified forcing scenarios, not the weather on a particular distant date. Extreme-event attribution asks another question: how human or natural influences changed an event’s probability or intensity. It combines observations, models, and statistical analysis, often comparing climates with and without human influence. Such a comparison does not replace an event’s meteorological explanation; it quantifies how the background climate altered the conditions or likelihood of its occurrence. (metoffice.gov.uk)

Applied Meteorology and Weather Services

Aviation meteorology supplies information organized around airports, flight levels, routes, and operating periods. Terminal aerodrome forecasts describe expected airport conditions, particularly wind, visibility, and cloud ceiling; en-route products identify turbulence, icing, thunderstorms, and other hazards. Pilots’ reports supplement observations by documenting conditions actually encountered. World Area Forecast System products provide gridded winds and significant-weather information for flight planning, allowing routes and cruising levels to be assessed against forecast hazards rather than surface weather alone. (connect.aviationweather.gov) Marine services similarly combine high-wind warnings with forecasts of wind waves, swell, visibility, and sea state. Wave height alone is insufficient: direction and period distinguish, for example, short, choppy seas from long-period swell. Significant wave height conventionally represents the mean height of the highest one-third of waves, not the maximum individual wave a vessel may encounter. (weather.gov)

Agricultural services translate atmospheric conditions into crop-specific risks and water requirements. Frost assessment considers crop development and local temperature differences: buds and flowers can become colder than the measured air during radiative cooling, while cold-air drainage creates frost pockets. (fao.org) For irrigation, Richard G. Allen, Luis S. Pereira, Dirk Raes, and Martin Smith’s 1998 FAO manual standardized a widely used crop-coefficient approach, $ET_c=K_cET_o$. Here $ET_o$ is reference evapotranspiration, $K_c$ varies with crop characteristics and growth stage, and $ET_c$ estimates crop evapotranspiration under standard conditions. (fao.org) Hydrometeorological services connect precipitation forecasts with catchment models whose initial conditions include soil moisture, snow-water storage, and river flow. Forecast-informed reservoir operations use predicted inflows and watershed conditions to support retention or release decisions while balancing water supply, environmental flows, flood control, and dam safety. (weather.gov)

Weather services for renewable energy address both generation and demand. Wind and solar forecasts support estimates of electricity production, while weather-dependent demand forecasts help operators manage the electrical grid. The UK Met Office, for example, has worked with the electricity system operator to improve solar-radiation forecasts and supplies information for renewable-generation modelling and consumer-demand forecasting. Site-specific wind and wave forecasts also identify access windows for offshore construction and maintenance. (metoffice.gov.uk) Transport services often require forecasts of the operating surface rather than the surrounding air: road-surface temperature and road-state predictions inform winter salting because pavement temperature can differ substantially from air temperature. (metoffice.gov.uk) Longer-term services combine meteorological records and projections with asset locations to inform building design, drainage, land use, and urban planning; the Met Office’s Climate Data Portal, introduced in 2023, enables users to combine climate information with their own geospatial datasets. (metoffice.gov.uk)

Impact-based warnings distinguish the atmospheric hazard from exposure—the people, property, and activities in its path—and vulnerability, their susceptibility to damage. Consequently, identical wind speeds need not imply identical consequences for a crane, a road vehicle, or an unoccupied landscape. WMO’s 2015 guidelines formalized this distinction for warning services. (etrp.wmo.int) The forecast-to-action chain therefore includes estimating hazard likelihood, combining forecasts with exposure and vulnerability information, communicating expected consequences, and linking warning levels to decisions agreed with emergency managers and other users. WMO’s expanded guidance emphasizes partnerships between forecasters, disaster managers, community representatives, and decision-makers. (wmo.int) An institutional public-health application is the NOAA–CDC HeatRisk service, presented nationally on April 22, 2024: its seven-day outlook combines weather forecasts with health-related evidence, considering the seasonal unusualness of heat, its duration, and both daytime and nighttime temperatures. (cdc.gov)

International Cooperation and Professional Practice

The International Meteorological Organization (IMO) originated in the 1873 International Meteorological Congress in Vienna, which established a permanent committee under the Dutch meteorologist Christophorus Buys Ballot to organize international cooperation. Its rules were subsequently developed at Utrecht in 1878 and its organization consolidated at the Rome congress in 1879. This largely nongovernmental association of meteorological services promoted standardized observations and cross-border reporting. Its intergovernmental successor, the World Meteorological Organization (WMO), was established when the WMO Convention entered into force on 23 March 1950; it became a specialized agency of the United Nations in 1951. The change placed international commitments to meteorological cooperation on a governmental rather than principally professional footing. (public.wmo.int)

Responsibilities remain distributed rather than concentrated in a single world forecasting service. National Meteorological and Hydrological Services collect and analyse observations, maintain national records, and provide forecasts, warnings, and authoritative advice within their respective mandates. International centres pool resources for computationally demanding prediction: the European Centre for Medium-Range Weather Forecasts (ECMWF), for example, supplies global guidance and boundary conditions used by national limited-area models. Universities and research institutes contribute scientific research, model development, and education; commercial providers participate in producing and delivering observations, information, and specialized services. These roles overlap, and public–private–academic cooperation operates within differing national legislative and institutional arrangements. (public.wmo.int)

Launched in 1963, the World Weather Watch organized cooperation around three complementary functions: observing the atmosphere, exchanging observations, and processing them into analyses and predictions. Its original core systems were the Global Observing System, Global Telecommunication System, and Global Data-processing and Forecasting System. The facilities belong to participating countries and organizations, not to a centrally owned WMO network. The modern WMO Integrated Global Observing System (WIGOS) provides an overarching framework for observing systems, including common metadata requirements. The WMO Information System (WIS) supports data discovery and exchange; WIS 2.0 entered its operational phase on 1 January 2025 using open standards and web technologies. The WMO Integrated Processing and Prediction System (WIPPS), formerly the Global Data-processing and Forecasting System, coordinates operational centres that share defined analysis and prediction products. These arrangements connect national observing capabilities with international computing resources and national service delivery. (wmo.int)

Common measurement practices make exchanged observations interpretable across borders. WMO’s Guide to Instruments and Methods of Observation addresses instrument operation, sampling, quality management, calibration, and intercomparison. WIGOS metadata standards provide the accompanying documentation needed to interpret measurements rather than treating a reported value as independent of its observing circumstances. At the policy level, the WMO Unified Data Policy, approved in October 2021, reaffirmed free and unrestricted international exchange and extended its scope across weather, climate, hydrology, oceans, atmospheric composition, the cryosphere, and space weather. Standardization thus concerns both measurement quality and the practical availability of data to other services. (community.wmo.int)

Observational capacity is uneven, particularly in least developed countries and small island developing states. The Global Basic Observing Network (GBON), approved in 2021, establishes requirements for essential surface-based observations and their international exchange. The associated Systematic Observations Financing Facility (SOFF) provides grant finance and technical assistance aimed at sustained compliance. Its emphasis on continuing operations addresses a recurrent infrastructure problem: installing equipment alone does not secure a reliable data stream without maintenance, trained personnel, communications, and long-term institutional support. Gaps affect both forecasts in poorly observed regions and the global models that assimilate their observations. (wmo.int)

Professional preparation distinguishes foundational education from competence in a particular job. WMO’s Basic Instruction Package for Meteorologists (BIP-M) and its counterpart for meteorological technicians (BIP-MT) specify educational outcomes; these do not by themselves establish all competencies needed for operational duties. The 2023 edition of WMO’s education and training guide preserves this distinction. Continuing professional development includes specialized forecasting, observing-system operation, and instrument calibration—for example, WMO training in calibration procedures and measurement uncertainty. Such training supports both the scientific interpretation of observations and the routine technical work required to keep international networks functioning. (etrp.wmo.int)

Research Frontiers and Unresolved Problems

Cloud microphysics, convection, and turbulent mixing remain major sources of model uncertainty because their controlling processes span scales far smaller than operational forecast grids. Even models with grid spacings of a few kilometres must approximate droplet and ice-particle populations, shallow clouds, and small-scale turbulence. Increasing resolution therefore changes rather than eliminates the parameterization problem: explicitly simulated convective motions must interact consistently with unresolved mixing and precipitation formation. Kilometre-scale Earth-system experiments represent storm organization and ocean fronts more directly, but models still differ substantially in turbulence near their grid scale. Resolving additional scales also increases computational and data-storage costs, constraining simulation length and ensemble size. (wmo.int)

Localized heavy precipitation illustrates the difficulty of connecting these processes across scales. An important unresolved problem is representing how convection organizes and precipitation moves, rather than merely producing realistic rainfall totals. ECMWF’s Integrated Forecasting System (IFS) Cycle 50r1, implemented on May 12, 2026, addressed a persistent tendency for convective precipitation to remain stationary: part of the precipitation generated by the convection scheme is transferred to the cloud scheme, where it can be transported and evaporate. This improves simulated inland propagation from coasts without requiring a finer horizontal grid. Rapid tropical-cyclone intensification presents a related forecasting limit; NOAA identifies it as a principal source of large operational intensity errors. Research combines atmospheric observations with measurements of upper-ocean structure to improve coupled hurricane forecasts. (ecmwf.int)

Machine learning offers both alternatives to numerical integration and components that can be embedded within physical models. In 2024, Dmitrii Kochkov and colleagues presented NeuralGCM, which combines a numerical solver for atmospheric dynamics with learned representations of unresolved processes. Their experiments produced competitive weather forecasts and stable multidecadal simulations with prescribed sea-surface temperatures. Such results demonstrate a hybrid modelling method, not an operational replacement for a fully coupled Earth-system model. The study also found that the system did not extrapolate to substantially different future climates, exposing a central generalization problem: reproducing historical atmospheric behaviour does not establish validity under unfamiliar forcing. (nature.com)

Dependence on training data extends to changes in the analyses used to initialize forecasts. ECMWF tested several machine-learning models using prototype Cycle 50r1 analyses for January 21–March 20, 2026. Some models lost skill when their initial conditions changed, despite improvements in the physical forecasting system. This provides a concrete example of distribution shift between training and operational inputs. Evaluation must consequently distinguish performance against model-based analyses from performance against independent measurements; ECMWF’s experiment additionally checked surface variables against SYNOP station observations. Its operational AIFS version 2 was adapted using prototype analyses from the upgraded system. (ecmwf.int)

Uncertainty estimation is another active frontier. A single accurate machine-learning forecast does not specify the range or probabilities of possible outcomes. The 2024 GenCast study instead generated probabilistic ensembles and reported greater skill than ECMWF’s comparison ensemble on 97.2% of 1,320 evaluated targets. That result concerns the specified experimental targets and comparison system, rather than establishing superiority for every hazard or location. Probabilistic forecasting research therefore examines extreme-event performance alongside ensemble reliability, while hybrid approaches investigate how learned components can coexist with physical dynamics and remain stable during repeated integration. (doi.org)

Coupled initialization and improved observation use are already producing operational changes, although comprehensive coupling remains an ongoing research programme. Cycle 50r1 uses coupled atmosphere–ocean trajectories when comparing forecasts with observations, while calculating atmospheric and ocean–sea-ice adjustments separately. Selected satellite radiances thereby constrain the ocean and sea ice as well as the atmosphere. Other changes include hourly geostationary observations and shorter observation-assignment intervals, reduced from 30 to 15 minutes. These implemented capabilities are distinct from experimental global kilometre-scale Earth-system simulations, whose initialization, unresolved processes, and evaluation of slowly evolving ocean and ice components remain substantial research challenges. (ecmwf.int)

References

1. Weatherwmo.int
2. Overviewwmo.int
3. Climatewmo.int
4. Weather Forecasting Through the Agesscience.nasa.gov
5. The Internet Classics Archive: Meteorology by Aristotleclassics.mit.edu
6. Factsheet 20 — From Sorcery to Supercomputersmetoffice.gov.uk
7. On The Modification Of Cloudsmetoffice.gov.uk
8. Admiral Beaufort’s ‘Scale of wind force’metoffice.gov.uk
9. Robert FitzRoy and the early Met Officemetoffice.gov.uk
10. Pioneers in modern meteorology and climate researchbjerknes.uib.no
11. Celebrating Lewis Fry Richardson and his legacymetoffice.gov.uk
12. Development of the ECMWF forecasting systemecmwf.int
13. History of RADARtraining.weather.gov
14. TIROSscience.nasa.gov
15. 60 Moments in NASA Historynasa.gov
16. The Atmospherenoaa.gov
17. Aerosol Size - NASA Sciencescience.nasa.gov
18. NOAA's National Weather Service - Glossary: Troposphereforecast.weather.gov
19. Layers of the Atmospherenoaa.gov
20. Temperature - Dry Bulb/Web Bulb/Dew Pointweather.gov
21. Guide to Meteorological Instruments and Methods of Observation: Chapter 1, Generalsites.stat.washington.edu
22. WMO OSCAR: Specific humidityspace.oscar.wmo.int
23. Isoplethsweather.gov
24. NOAA's National Weather Service - Glossary: Geopotential Heightforecast.weather.gov
25. Temperature Gradients...Heights and Thicknessesweather.gov
26. Pressureweather.gov
27. Climate and Earth’s Energy Budgetscience.nasa.gov
28. Satellite and Ocean Data Reveal Marked Increase in Earth’s Heating Raterepository.library.noaa.gov
29. Chapter 7: The Earth’s Energy Budget, Climate Feedbacks, and Climate Sensitivityipcc.ch
30. Chapter 6: Short-lived Climate Forcersipcc.ch
31. Part IV: Physical Processesecmwf.int
32. Physical relations in observations and modelsecmwf.int
33. Global Radiation and Aerosolsgml.noaa.gov
34. The Sea Breezenoaa.gov
35. Practical Meteorology, Chapter 3: Thermodynamicseoas.ubc.ca
36. Practical Meteorology, Chapter 4: Water Vaporeoas.ubc.ca
37. Met Office Hadley Centre observations datasets: HadISDHmetoffice.gov.uk
38. NOAA’s National Weather Service Glossary: Adiabaticforecast.weather.gov
39. Environmental Parameters and Indicesweather.gov
40. UBC ATSC 201: Atmospheric Soundings and Stability—Tutorial Aeoas.ubc.ca
41. How Clouds Formnoaa.gov
42. ESCI 241 – Meteorology: Lesson 10 – Precipitationweather.gov
43. Production of ice-nucleating particles (INPs) by fast-growing phytoplanktonacp.copernicus.org
44. Classifying cloudspublic.wmo.int
45. Generacloudatlas.wmo.int
46. Definitions of cloudscloudatlas.wmo.int
47. Cloud Developmentweather.gov
48. Freezing Rain and Sleetweather.gov
49. What Causes A Wintry Mix Of Precipitation?weather.gov
50. How Fog Formsweather.gov
51. Dew and Frost Developmentweather.gov
52. How Do Rainbows Form?weather.gov
53. Overview of the Numerics of the ECMWF Atmospheric Forecast Modelecmwf.int
54. Atmospheric dynamicsecmwf.int
55. A parametrization scheme for the ECMWF gridpoint modelecmwf.int
56. Atmospheric physicsecmwf.int
57. Weather Termsweather.gov
58. Forecasters’ Reference Bookweather.gov
59. Divergence/Convergenceweather.gov
60. Dynamics of the Atmosphereocw.mit.edu
61. 810 (S23) Lecture Slides 4: Large-Scale Flow with Rotation and Stratificationocw.mit.edu
62. 810 (S23) Introductory Notes: Internal Gravity Wavesocw.mit.edu
63. 810 (S23) Lecture Slides 6: E-P Fluxes and Instability in Midlatitudesocw.mit.edu
64. The atmospheric boundary layer - Met Officemetoffice.gov.uk
65. Surface and sub-surface processes - Met Officemetoffice.gov.uk
66. Presentazione di PowerPointweather.gov
67. TN-98/002aoml.noaa.gov
68. On the Observation and Modeling of the Slope Winds of the Upper Current River Valley of Southeast Missouri and Their Relationship to Air-Mass Thunderstorm Formationrepository.library.noaa.gov
69. Mountain Wave Activity Over the Southern Rockiesweather.gov
70. Guide to Reducing Heat Islands - US EPAepa.gov
71. What Are Heat Islands? - US EPAepa.gov
72. Global Atmospheric Circulationsnoaa.gov
73. Global circulation patterns - Met Officeweather.metoffice.gov.uk
74. The Jet Streamnoaa.gov
75. The North American Monsoonclimate.gov
76. The Tropical Ocean-Global Atmosphere observing system: A decade of progresspmel.noaa.gov
77. Responses of the Tropical Atmosphericrepository.library.noaa.gov
78. TCFAQ A15) How do tropical cyclones form ?aoml.noaa.gov
79. TCFAQ A16) Why do tropical cyclones require 80°F (26.5°C)aoml.noaa.gov
80. Tropical Cyclone Structurenoaa.gov
81. Hurricane FAQ - NOAA/AOMLaoml.noaa.gov
82. Dynamics and Physics - NOAA/AOMLaoml.noaa.gov
83. L0311: Hurricane Readinessnhc.noaa.gov
84. Air Massesnoaa.gov
85. Identifying and Forecasting Long Wave Weather Patternsweather.gov
86. High and low pressureweather.metoffice.gov.uk
87. Blocking patternsweather.metoffice.gov.uk
88. Observationsmetoffice.gov.uk
89. Frequently Asked Questionsweather.gov
90. How we measure sunshine and radiation - Met Officeweather.metoffice.gov.uk
91. Guide to Instruments and Methods of Observation (WMO-No. 8)community.wmo.int
92. Radiosonde Observationweather.gov
93. Global Observing System (GOS)public.wmo.int
94. Hurricane Research Division - NOAA/AOMLaoml.noaa.gov
95. How radar worksnoaa.gov
96. JetStream Max: Dual Polarizationnoaa.gov
97. Dual Polarization Radarweather.gov
98. NWS/OCLO/WDTD Dual-Polarization Radar Training for NWS Partnerstraining.weather.gov
99. Weather Satellitesnoaa.gov
100. MADIS Quality Controlmadis.ncep.noaa.gov
101. Recognition criteria for meteorological observing stationswmo.int
102. Bias correction methodsevents.ecmwf.int
103. Constrained adaptive bias correction for satellite radiance assimilation in the ECMWF 4D-Var systemecmwf.int
104. Skew-T Parameters and Indicesweather.gov
105. The skew-T log-P diagramweather.gov
106. Meteorological Techniquesweather.gov
107. Data assimilationecmwf.int
108. Global reanalysis: goodbye ERA-Interim, hello ERA5ecmwf.int
109. Fifty years of data assimilation at ECMWFecmwf.int
110. 20 years of 4D-Var: better forecasts through a better use of observationsecmwf.int
111. Improving the handling of model bias in data assimilationecmwf.int
112. Numerical weather prediction modelsmetoffice.gov.uk
113. Even Newer Dynamics for General atmospheric modelling of the environment (ENDGame)metoffice.gov.uk
114. Running WRF — WRF Model documentationwww2.mmm.ucar.edu
115. Nested Model Options — WRF Model documentationwww2.mmm.ucar.edu
116. Chapter 5: WRF Modelwww2.mmm.ucar.edu
117. History of numerical weather predictionweather.metoffice.gov.uk
118. The current forecasting equations are still valid near km-scale resolutionecmwf.int
119. High resolution model developmentmetoffice.gov.uk
120. Physics-dynamics couplingmetoffice.gov.uk
121. WMO OSCAR: Application Area 2.3 Nowcasting / Very Short-Range Forecastingspace.oscar.wmo.int
122. Radar Nowcasting Methodsetrp.wmo.int
123. How to use our 10-day weather forecastsweather.metoffice.gov.uk
124. Subseasonal to Seasonal Prediction: Research Implementation Planwmo.int
125. SEAS5 user guideecmwf.int
126. Ensemble Forecastingecmwf.int
127. Quantifying forecast uncertaintyecmwf.int
128. Addressing model uncertainty through statistical post-processing using reforecastsecmwf.int
129. B Statistical Concepts – Probabilistic Dataconfluence.ecmwf.int
130. Verification Methods for NWP Model Outputcawcr.gov.au
131. Verification of Precipitation Forecasts: A Survey of Methodologycawcr.gov.au
132. Common Verification Methodscawcr.gov.au
133. What is Ozone?epa.gov
134. What is Acid Rain?epa.gov
135. Aerosols: Tiny Particles, Big Impactscience.nasa.gov
136. Atmospheric heating in the US from saharan dust: Tracking the June 2020 event with surface and satellite observationsrepository.library.noaa.gov
137. The Water Cyclescience.nasa.gov
138. Modeling the Water Budget – Science Lessonjpl.nasa.gov
139. Chapter 8: Water Cycle Changesipcc.ch
140. Reporting on the State of the Climate in 2018content-drupal.climate.gov
141. Unified Modelmetoffice.gov.uk
142. Coastal Waters Forecasts With Wave Detailweather.gov
143. Frost protection: fundamentals, practice, and economics — Chapter 1fao.org
144. Frost protection: fundamentals, practice, and economics — Recommendationsfao.org
145. Crop evapotranspiration — Guidelines for computing crop water requirementsfao.org
146. Chapter 5 — Introduction to crop evapotranspirationfao.org
147. Missouri/Souris River Floods of May–August 2011weather.gov
148. Federal action plan focused on improving water prediction for Western U.S.noaa.gov
149. How does the Met Office work with the energy sector?metoffice.gov.uk
150. Energy sector servicesmetoffice.gov.uk
151. OpenSite: Road Surface Forecastsmetoffice.gov.uk
152. History of IMO and WMOpublic.wmo.int
153. WMO at 75: A Shining Example of the Benefits of International Cooperationpublic.wmo.int
154. Serving meteorologyecmwf.int
155. About PPEwmo.int
156. 20th centurywmo.int
157. World Weather Watchwmo.int
158. WMO Integrated Global Observing Systemwmo.int
159. WMO Information System 2.0 will transform sharing of Earth System Datawmo.int
160. WMO Integrated Processing and Prediction System (WIPPS)wmo.int
161. WMO overhauls data exchange policywmo.int
162. Kilometre-Scale Modelling of the Earth System: A New Paradigm for Climate Predictionwmo.int
163. Upgrade to IFS Cycle 50r1ecmwf.int
164. Significant update to ECMWF’s key forecasting systems IFS and AIFS goes liveecmwf.int
165. Hurricane Modeling and Prediction Programaoml.noaa.gov
166. Cyclone intensificationaoml.noaa.gov
167. Neural general circulation models for weather and climatenature.com
168. Farewell to the external AI modelsecmwf.int
169. Probabilistic weather forecasting with machine learningdoi.org