{
  "id": "precipitation",
  "title": "Precipitation",
  "lang": "en",
  "source_lang": "en",
  "summary": "Water falling from the atmosphere as rain, snow, or other liquid or solid forms, linking weather, climate, and the water cycle.",
  "category": "geography",
  "subcategory": "",
  "type": "concept",
  "aliases": [
    "Atmospheric precipitation",
    "Hydrometeorological precipitation"
  ],
  "generated_at": "2026-10-25",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/precipitation/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/precipitation/",
      "json": "https://aiwiki.page/zh/precipitation.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Precipitation is liquid or solid water that falls from clouds or other parts of the atmosphere, principally as rain, drizzle, snow, ice pellets, and hail. It forms when water droplets or ice particles grow sufficiently large to fall under gravity. In surface weather observations, precipitation refers to water reaching the ground; falling particles that evaporate or sublimate before arrival are called virga. Dew, frost, and the direct interception of fog are generally distinguished from precipitation.\n\nAs a major component of the water cycle, precipitation transfers atmospheric moisture to land and ocean, replenishing rivers, soils, glaciers, and groundwater. Its amount, timing, intensity, and physical form help determine climates, ecosystems, agricultural conditions, and water availability. Excesses and deficits contribute to floods and droughts, while snow, freezing rain, and hail create distinctive hazards.\n\nPrecipitation reflects interactions among cloud microphysics, atmospheric circulation, terrain, and temperature. Its distribution varies from individual storm cells to global climatic belts and from minute-scale showers to long-term climatic changes. Gauges, radar, and satellites provide complementary observations, while numerical models support forecasting and the assessment of changes associated with climate change."
    },
    {
      "heading": "Definition, quantities, and atmospheric water budgets",
      "text": "Precipitation consists of liquid or solid hydrometeors falling from the atmosphere, including rain, snow, and hail. Ordinary cloud droplets and small ice crystals remain suspended because their settling speeds are small relative to atmospheric motions; their presence does not by itself establish precipitation. Meteorological observations distinguish precipitation aloft from precipitation reaching the surface. Falling particles that evaporate or sublime completely before reaching the ground form virga. Dew and frost, by contrast, form through condensation or vapour deposition directly onto surfaces. Fog intercepted by vegetation or structures also supplies surface water, but is normally accounted for separately from falling precipitation.\n\nAccumulated precipitation is expressed as the depth of liquid water that would cover a horizontal surface if the received water were retained without runoff, infiltration, or evaporation. Thus, \\(1\\) millimetre over \\(1\\) square metre equals \\(0.001\\) cubic metres, or exactly \\(1\\) litre. Frozen precipitation is expressed as its liquid-water equivalent: the depth obtained after melting it, not the depth of the snow layer. Precipitation rate is accumulation per unit time, commonly millimetres per hour. An accumulation of \\(12\\) millimetres over three hours gives an average rate of \\(4\\) millimetres per hour, even if brief bursts were much more intense. Duration records how long precipitation occurs; frequency describes how often a specified occurrence criterion is met. Counts of “wet days,” for example, depend on whether the threshold is \\(0.1\\), \\(1\\), or another number of millimetres.\n\nWithin the water cycle, precipitation removes atmospheric water supplied by evaporation and evapotranspiration, the latter combining evaporation from land surfaces with plant transpiration. Winds transport moisture between regions, so local precipitation need not originate from local evaporation. For an atmospheric column, conservation of water can be written\n\n\\[\\frac{\\partial W}{\\partial t}+\\nabla\\cdot\\mathbf{Q}=E-P,\\]\n\nwhere \\(W\\) is vertically integrated atmospheric water storage, \\(\\mathbf{Q}\\) is horizontal water transport integrated through the column, and \\(E\\) and \\(P\\) are surface evaporation and precipitation fluxes. All terms must use consistent units. Water vapour dominates atmospheric storage and transport, although cloud liquid and ice also contribute. Moisture convergence, represented by negative \\(\\nabla\\cdot\\mathbf{Q}\\), can sustain precipitation substantially exceeding local evaporation.\n\nAtmospheric storage matters particularly over short periods: storms can draw down previously accumulated moisture while receiving continuing moisture inflow. The global atmospheric water inventory corresponds to roughly \\(25\\) millimetres spread over Earth’s surface, whereas global mean annual precipitation is approximately \\(1\\) metre. Their ratio implies an atmospheric water turnover time of about nine days, not a fixed lifetime for every water molecule. Precipitation efficiency describes how effectively supplied moisture or cloud condensate becomes precipitation. Its numerical value depends on the definition: surface precipitation divided by condensation is not the same ratio as precipitation divided by moisture inflow. Evaporation and sublimation during descent reduce surface precipitation relative to production aloft.\n\nOver the whole planet, horizontal transport cancels when integrated globally. Over sufficiently long periods, changes in atmospheric storage are small relative to annual fluxes, so total evaporation approximately equals total precipitation. This does not imply regional equality: the oceans evaporate more water than they receive as precipitation, while land collectively receives more precipitation than it loses through evapotranspiration. Atmospheric transport carries a net amount of order \\(40{,}000\\) cubic kilometres annually from ocean to land.\n\nA catchment budget tracks a different reservoir: water stored on and below the land surface. In its simplest form, \\(P=ET+R+\\Delta S\\), with all quantities expressed over the same area and interval; \\(R\\) is net water export and \\(\\Delta S\\) is storage change in soil, groundwater, snow, lakes, and other reservoirs. Groundwater crossing catchment boundaries and human water transfers require additional accounting. River discharge therefore need not match precipitation minus evapotranspiration over a particular month or year, because stored water can accumulate or be released."
    },
    {
      "heading": "Liquid, frozen, and freezing forms",
      "text": "Rain and drizzle consist of liquid water drops, distinguished by size, concentration, and falling characteristics rather than by precipitation intensity alone. Drizzle comprises numerous, closely spaced drops, usually less than 0.5 millimetres in diameter, whose slow descent can make them appear to float with air currents. It commonly falls from low stratus clouds. Raindrops are generally larger than 0.5 millimetres; smaller drops are also classified as rain when they are comparatively widely separated. Thus, light rain is not necessarily drizzle: “light” describes intensity, whereas “drizzle” identifies a particle population. At approximately sea-level air density, a 0.1-millimetre drop falls at about 0.3 metres per second, compared with roughly 4 metres per second for a 1-millimetre raindrop. Larger raindrops become flattened rather than teardrop-shaped.\n\nSnow consists of ice crystals falling individually or joined into aggregates. Individual crystals may form plates, columns, needles, or branched dendrites, with their shapes depending on temperature and water-vapour conditions during growth. Aggregates form when crystals collide and adhere; the large flakes often seen near the melting point can contain many crystals. Snow grains are much smaller, opaque white particles, generally less than 1 millimetre across, that resemble tiny grains rather than recognizable branched crystals. They commonly fall from stratus or fog and do not bounce or shatter on hard surfaces. Diamond dust, also called ice-crystal precipitation, consists of minute crystals that descend slowly, sometimes from an apparently clear sky in very cold air. Their reflective faces can make them visible as sparkling points in sunlight.\n\nGraupel, or snow pellets, consists of white, opaque particles formed when supercooled cloud droplets freeze onto snow crystals. Typically about 2–5 millimetres across, these particles are porous and relatively soft: they can bounce on impact but are readily crushed. Hail instead consists of harder balls or irregular lumps of ice, conventionally at least 5 millimetres in diameter, often with alternating clear and opaque layers. It develops chiefly in convective clouds with strong updrafts that support growing particles while they collect supercooled liquid water. Graupel and hail therefore differ in structure and growth environment, not merely in size; small hail should not be confused with the generally smaller ice pellets produced by refreezing during descent.\n\nThe vertical temperature profile determines how particles change phase before reaching the ground. Snow can survive through an entirely subfreezing column, and sometimes through a shallow above-freezing layer if melting remains incomplete. If snow falls through a sufficiently deep warm layer, it melts into rain. Where a subfreezing layer lies beneath that warm layer, two different outcomes are possible. A sufficiently cold and deep lower layer allows the drops to freeze before reaching the surface, producing hard, usually transparent or translucent ice pellets. A shallower or less intensely cold layer may leave drops liquid but supercooled—below \\(0\\,^\\circ\\mathrm{C}\\)—so that they freeze on contact with surfaces cold enough to sustain freezing. This is freezing rain, which can produce a smooth glaze. Freezing drizzle involves smaller supercooled drops and can also develop without an overlying melting layer.\n\nMixed precipitation occurs when different particle types reach the surface together, for example rain and partially melted snow near a rain–snow boundary. Small changes in melting-layer depth, humidity, or surface elevation can alter the mixture over short distances. Terminology is not uniform: in United States and much Canadian meteorological usage, sleet means ice pellets, whereas in British usage it generally means a mixture of rain and snow. Explicit terms such as “ice pellets” or “rain and snow mixed” avoid this ambiguity."
    },
    {
      "heading": "Cloud microphysics and particle growth",
      "text": "Cloud particles begin growing when water vapour condenses or forms ice, but their formation depends on more than reaching saturation. Saturation occurs when the vapour pressure equals the equilibrium saturation vapour pressure at the prevailing temperature; supersaturation is the excess above that value, usually expressed as a percentage. Most liquid cloud droplets form on cloud condensation nuclei, such as sea-salt particles and soluble components of atmospheric aerosols. Dissolved material lowers the equilibrium vapour pressure over a droplet, whereas surface curvature raises it. Together, these effects determine the supersaturation at which an aerosol particle activates into a growing cloud droplet. In many clouds, supersaturations relative to liquid water are only a few tenths of a percent. Typical cloud droplets have diameters of roughly 10–20 micrometres, far smaller than millimetre-sized raindrops.\n\nGrowth by condensation alone is too slow to explain the rapid production of large raindrops. Under approximately steady conditions, a droplet’s squared radius increases roughly in proportion to time: progressively larger droplets therefore gain radius more slowly. A drop 1 millimetre in diameter contains as much water as about one million droplets 10 micrometres in diameter. In liquid clouds, collision–coalescence bridges much of this size gap. Larger droplets generally fall faster than smaller ones and can overtake and collect them. Collection efficiency depends both on whether droplets collide—small droplets may follow airflow around a larger drop—and on whether colliding droplets merge rather than separate. Turbulence and variations in droplet size alter encounter rates. Once some drops become efficient collectors, their growth accelerates, broadening the drop-size distribution. Collisions can also fragment drops, while large drops may undergo aerodynamic breakup; these processes prevent unrestricted growth and replenish smaller sizes.\n\nIce formation introduces additional pathways. Ice-nucleating particles, including some mineral-dust and biological particles, enable freezing or ice formation at temperatures warmer than those required for homogeneous freezing. Their effectiveness depends on composition, temperature, and how they interact with water. Liquid droplets can remain supercooled well below \\(0\\,^\\circ\\mathrm{C}\\); homogeneous freezing becomes important near \\(-38\\,^\\circ\\mathrm{C}\\), with the precise threshold depending on droplet size and conditions. Mixed-phase clouds consequently contain both ice crystals and supercooled droplets. Vapour deposition grows ice crystals into plates, columns, dendrites, and other habits whose development depends on temperature and supersaturation.\n\nThe Wegener–Bergeron–Findeisen process, associated with Alfred Wegener’s 1911 work, Tor Bergeron’s 1935 publication, and Walter Findeisen’s 1938 study, exploits the lower saturation vapour pressure over ice than over liquid water at the same subfreezing temperature. At \\(-10\\,^\\circ\\mathrm{C}\\), the respective values are approximately 260 and 286 pascals. Air can therefore be supersaturated relative to ice but subsaturated relative to liquid water: ice grows by vapour deposition while liquid droplets evaporate, supplying additional vapour. This is not a direct transfer through contact between particles. Nor must droplets always evaporate in a mixed-phase cloud; sufficiently strong cooling can maintain supersaturation relative to both phases, allowing liquid and ice to grow simultaneously.\n\nIce particles also enlarge through aggregation, in which crystals collide and stick together, and riming, in which collected supercooled droplets freeze onto their surfaces. Aggregation produces snowflakes, whereas sustained riming produces increasingly compact particles such as graupel. Hail grows as embryos collect supercooled water in strong updrafts that prolong their residence in the growth region. If latent heat cannot be removed rapidly enough to freeze all collected water immediately, a liquid surface layer develops, producing wet growth. During descent, melting can convert ice particles into raindrops; evaporation shrinks liquid drops, and sublimation removes mass from ice directly into vapour. Thus, the particle population reaching the ground reflects both growth inside clouds and modification below them."
    },
    {
      "heading": "Atmospheric lifting and precipitation-producing systems",
      "text": "Rising air encounters lower atmospheric pressure, expands, and cools through an adiabatic process. Unsaturated air cools at approximately 9.8 °C per kilometre of ascent; after saturation, the release of latent heat during condensation reduces the cooling rate, commonly to about 4–7 °C per kilometre in the lower troposphere, depending on temperature and moisture content. Ascent can therefore produce cloud, but does not guarantee precipitation at the surface: particles must grow sufficiently, and falling precipitation may evaporate in dry air below. Moisture supply, the depth and persistence of ascent, and atmospheric stability jointly govern precipitation development. Stable air resists vertical displacement, whereas sufficiently buoyant air can accelerate upward and sustain deep convection.\n\nConvective precipitation develops in localized updrafts, ranging from isolated showers to thunderstorms and organized mesoscale convective systems extending across hundreds of kilometres. Surface heating, low-level convergence, fronts, or terrain can initiate ascent; precipitation-cooled outflows can subsequently lift adjacent warm, moist air and generate new cells. Individual convective cells commonly produce strong spatial contrasts in rainfall over short distances. In organized systems, intense convective cores frequently coexist with broader stratiform regions, where precipitation falls through comparatively weak, widespread vertical motion. Ice particles formed in convective updrafts can be transported into these regions and continue growing or aggregating. “Convective” and “stratiform” therefore describe precipitation structures and processes, not mutually exclusive categories of storms.\n\nIn extratropical cyclones, large-scale circulation and frontal zones organize ascent over areas much broader than individual convective cells. Along a warm front, relatively warm, moist air rises over a colder air mass, often producing an extensive cloud and precipitation shield ahead of the surface front. A cold front can force a narrower band of stronger ascent, although its precipitation may be convective, stratiform, or mixed. A warm conveyor belt—a coherent stream of moist air ascending through the cyclone—helps supply widespread precipitation, including bands wrapping around the low-pressure centre. Upper-level disturbances and atmospheric convergence also contribute to ascent; frontal precipitation cannot be explained solely as one air mass mechanically pushing beneath another.\n\nMountains force cross-barrier airflow upward, promoting cooling and precipitation on windward slopes. Enhancement depends on moisture availability, wind speed and direction, atmospheric stability, and the height and width of the barrier. Deep, moist flow can carry cloud and precipitation across a crest, while stable low-level air may be blocked and diverted around the mountains. Existing precipitation falling through shallow mountain cloud can also grow by collecting cloud droplets, a process termed the seeder–feeder mechanism. On the leeward side, descending air warms and its relative humidity decreases, favouring evaporation and a rain shadow. The Olympic Mountains of Washington illustrate strong windward enhancement and much drier conditions to the northeast, but local rainfall patterns vary with storm airflow.\n\nTropical cyclones draw moisture from warm oceans and from moist air converging into their circulation. Strong ascent concentrates heavy precipitation in the eyewall and spiral rainbands, while broader cloud regions also produce stratiform rain. Rainfall distribution depends on storm motion, vertical wind shear, moisture, and terrain; slow-moving systems can sustain large accumulations without exceptionally strong surface winds. Atmospheric rivers, by contrast, are elongated corridors of concentrated water-vapour transport, usually associated with extratropical weather systems. They are not themselves continuous rainbands: substantial precipitation develops where their moisture encounters frontal ascent or terrain. Landfalling atmospheric rivers are important sources of heavy rain and mountain snow along western North America.\n\nLake-effect precipitation occurs when cold air crosses a relatively warm, unfrozen lake, acquiring heat and water vapour that destabilize the lower atmosphere. Convergence and shoreline geometry can organize the resulting convection into narrow bands, often producing highly localized snowfall downwind of the North American Great Lakes. Band position and intensity depend on wind direction, the distance travelled over open water, inversion height, and wind shear. Downwind hills can further enhance snowfall, whereas extensive ice cover generally reduces the lake’s heat and moisture supply."
    },
    {
      "heading": "Global distribution and seasonal regimes",
      "text": "Broad precipitation belts reflect the global circulation of the atmosphere, modified by continents and oceans. Near the equator, converging trade winds and ascending moist air support frequent convection in the intertropical convergence zone (ITCZ). Its seasonal migration produces shifting rainfall belts rather than a permanently wet equatorial line; some equatorial locations experience two rainfall maxima as the convergence zone passes nearby twice each year. In the subtropics, generally around 20–35° latitude, descending air associated with the Hadley circulation warms and lowers relative humidity, suppressing widespread cloud development. Many major deserts occupy these latitudes. Farther poleward, midlatitude storm tracks carry cyclones and fronts, producing broad precipitation zones whose positions and strength vary seasonally. Polar regions generally receive little precipitation because cold air contains little water vapour and moisture transport is limited, although exposed coastal mountains can receive substantial snowfall.\n\nMonsoonal regimes feature pronounced seasonal changes in circulation and rainfall. During the South Asian summer monsoon, moist airflow from the Indian Ocean supplies widespread rain, while winter circulation brings much drier conditions to much of the subcontinent. The contrast is not universal: southeastern India receives an important share of its annual rainfall during the October–December northeast monsoon, when air acquires moisture over the Bay of Bengal. Mediterranean climates have the opposite seasonal emphasis. Summer subtropical subsidence suppresses rainfall, whereas the winterward expansion of midlatitude storm activity brings rain to regions such as the Mediterranean Basin, central Chile, southwestern Australia, and coastal California. Thus, a winter-rainfall climate and a summer-monsoon climate can have similar annual totals but very different monthly distributions.\n\nMaritime precipitation depends on exposure to prevailing winds, not simply proximity to the sea. Western coasts in the midlatitudes often receive frequent frontal precipitation from oceanic westerlies; exposed uplands of western Britain and New Zealand’s South Island are examples. Oceanic moisture supply and relatively weak seasonal temperature contrasts can support precipitation throughout the year, although winter maxima are common. Continental interiors are often drier because air masses lose moisture along their paths and encounter mountain barriers. In many temperate interiors, summer heating and increased atmospheric moisture favour a warm-season precipitation maximum, while cold winters bring smaller liquid-water-equivalent totals. Nevertheless, continentality does not prescribe a single regime: seasonal wind direction, storm tracks, and moisture recycling through evapotranspiration also matter.\n\nElevation enhances precipitation where moist airflow is forced upward, but precipitation does not increase indefinitely with height. Moisture depletion, airflow blocking, and the altitude of cloud layers can produce maxima below mountain summits. Along the southern Himalayas, summer monsoon flow produces heavy precipitation on exposed slopes, while areas beyond the main barriers, including Ladakh and parts of the Tibetan Plateau, are much drier. The Andes likewise create contrasting regimes: tropical easterly moisture supports abundant precipitation on their Amazon-facing slopes, whereas farther south, Pacific westerlies wet western Patagonia and leave eastern Patagonia in a rain shadow.\n\nThe Atacama region illustrates the combined effects of circulation, mountains, and ocean conditions. Subtropical subsidence inhibits deep cloud development, the Andes restrict moisture arriving from the east, and the cold Humboldt Current and coastal upwelling help maintain a stable lower atmosphere. Parts of the hyperarid interior receive average annual precipitation of less than 5 millimetres. Coastal low cloud and fog can nevertheless supply water locally through interception, without producing comparable rainfall totals.\n\nLow precipitation alone does not determine aridity, which depends on water supply relative to atmospheric evaporative demand. A common climatic index is \\(P/PET\\), the ratio of annual precipitation to potential evapotranspiration. A cool region and a hot region receiving the same annual precipitation can therefore have very different moisture balances. Seasonal distribution further modifies water availability: precipitation concentrated in a short wet season may be followed by months of strong evaporative demand, whereas an equal annual amount distributed more evenly produces shorter dry intervals."
    },
    {
      "heading": "Variability, extremes, and statistical description",
      "text": "Precipitation is intermittent: a location’s record combines intervals without precipitation with events whose intensities vary substantially. Consequently, the same monthly total can result from frequent light rain or a few downpours. Wet- and dry-spell statistics describe consecutive days above or below a specified threshold; a common convention classifies a wet day as one receiving at least 1 mm. Spell lengths depend on persistent circulation patterns as well as individual storms, so daily occurrences are not generally independent. Diurnal cycles also vary geographically: solar heating often favors afternoon convection over land, whereas nocturnal rainfall maxima occur in some continental regions through organized storms and low-level moisture transport. Spatially, intense convective cores may occupy only a few kilometres, while regional precipitation systems extend hundreds of kilometres. Averaging over larger areas smooths localized peaks; a point maximum is therefore not interchangeable with a drainage-basin average.\n\nInterannual variability is strongly influenced by the El Niño–Southern Oscillation (ENSO), which changes tropical Pacific convection and atmospheric circulation beyond the Pacific. During El Niño, rainfall commonly decreases over Indonesia and increases in parts of the central and eastern equatorial Pacific; southern United States winters also tend to be wetter. These are seasonal tendencies, not outcomes guaranteed for every event. Responses depend on the location and strength of ocean warming, the season, and other circulation patterns. The North Atlantic Oscillation, for example, alters Atlantic storm tracks and winter precipitation across Europe, with its positive phase generally favoring wetter conditions in northern Europe and drier conditions around the Mediterranean. Multidecadal fluctuations can produce apparent trends over short records, making the chosen observation period important when separating variability from longer-term change.\n\n“Heavy precipitation” requires an explicit threshold, accumulation interval, and spatial support. An absolute criterion might identify days exceeding 50 mm, whereas a relative criterion might identify daily totals above the local 95th or 99th percentile of wet-day precipitation during a fixed reference period. The percentile approach accommodates regional differences but does not imply equal water amounts or impacts across climates. Common statistical indices include the annual maximum one-day total, annual maximum consecutive five-day total, and precipitation accumulated on days exceeding a reference percentile. A storm delivering 60 mm in one hour has a much greater average intensity than one delivering the same amount in 24 hours. Conversely, a season with exceptionally high precipitation need not contain a record-breaking hourly event.\n\nIntensity–duration–frequency relationships describe how precipitation intensity varies with accumulation duration and estimated rarity at a specified location. They are commonly fitted using annual maxima or suitably separated events exceeding a high threshold. For annual maxima, the return period \\(T\\), in years, is related to annual exceedance probability \\(p\\) by \\(T=1/p\\). Thus, a “100-year” precipitation amount has a 1% probability of being exceeded in any given year under the fitted model—not a scheduled recurrence once per century. Assuming independent years and an unchanged probability, its chance of being exceeded at least once in 100 years is\n\n\\[1-(1-0.01)^{100}\\approx 0.634.\\]\n\nIt can also be exceeded in consecutive years. Estimates depend on record length, distribution choice, event independence, and whether the underlying statistics remain stationary; extrapolating far beyond the observed record introduces substantial uncertainty.\n\nMeteorological drought is a sustained precipitation deficit relative to conditions expected for a location, season, and duration. Agricultural drought concerns inadequate soil moisture for vegetation or crops, while hydrological drought concerns reduced streamflow, reservoir storage, or groundwater; these can develop and recover at different times. The Standardized Precipitation Index, introduced by Thomas McKee, Nolan Doesken, and John Kleist in 1993, expresses accumulated precipitation relative to a fitted local probability distribution. Three-month and twelve-month values describe different timescales of deficit. A dry month during the usual wet season may represent a larger anomaly than the same total during the dry season. Reference-period selection affects the estimated anomaly, and precipitation-only indices do not directly represent evaporative demand, soil properties, or water management."
    },
    {
      "heading": "Ground measurement and observational records",
      "text": "Systematic precipitation measurement requires a collector of known area, consistent observation times, and a record of instrument exposure. In Korea, the cheugugi rain gauge was introduced in 1441 during King Sejong’s reign; standardized gauges and reporting arrangements followed in 1442, supporting administrative assessment of rainfall and agricultural conditions. During the nineteenth century, national meteorological services and volunteer networks expanded the geographical coverage and regularity of observations. The United States Cooperative Observer Program, established in 1890, became an important source of daily precipitation records. Standardization made observations more comparable, but historical series still reflect changes in instruments, locations, and observing practices.\n\nA manual storage gauge collects precipitation through a horizontal opening into a container. The collected volume divided by the opening area gives precipitation depth; a graduated measuring tube allows small amounts to be read precisely. Storage gauges provide an accumulation between visits rather than its detailed timing, and can lose water through evaporation, leakage, or overflow. A tipping-bucket gauge directs water into a small pivoting bucket assembly: each tip represents a nominal increment, commonly 0.2 millimetres in some designs. Recording tip times resolves rainfall intermittency, but light rainfall is temporally quantized, and intense rainfall can be underestimated while the mechanism tips. A weighing gauge records changes in collected mass, allowing liquid and solid precipitation to be measured as water equivalent; wind-induced vibration, temperature sensitivity, and servicing complicate the record. A disdrometer measures particle characteristics, such as diameter and fall speed, using impact or optical detection. It estimates size distributions and precipitation rate, but samples a small volume and is sensitive to particle shape, wind, and overlapping particles.\n\nObservations of snow distinguish fresh snowfall, total snow depth, and snow-water equivalent: the depth of liquid water obtained by melting a snow sample. Fresh snowfall is measured on a level snow board cleared at specified intervals, whereas snowpack depth is measured on undisturbed ground, preferably at several representative points. A core of known cross-sectional area can be melted or weighed to determine snow-water equivalent. No fixed snow-to-water ratio is adequate: crystal form, riming, melting, compaction, and wind packing all alter snow density. Measurement frequency also matters because settling between readings reduces the apparent accumulation. Heated gauges melt frozen precipitation for automated measurement, but heating can increase evaporative losses.\n\nWind undercatch is a major systematic error, especially for snow. Air accelerates and is deflected around a gauge opening, carrying some falling particles away from the collector. Wind shields reduce this effect without eliminating it; correction depends on wind speed, precipitation type, and gauge configuration. Water retained on the funnel and container produces wetting losses, particularly important for small events. Gauges require level rims and exposure that avoids both obstruction by buildings or trees and excessive turbulence. A “trace” denotes observed precipitation below the measurable threshold, not necessarily zero. Quality control checks impossible values, accumulation across missed observations, clock errors, and disagreement with nearby stations, while retaining plausible localized extremes.\n\nA gauge is a point sample, whereas a drainage basin requires an areal estimate. Arithmetic averaging, Thiessen polygons, and interpolation methods weight stations differently; elevation-aware approaches can represent systematic terrain gradients when sufficient observations exist. Network density determines how well short-lived, localized showers are sampled. Mountains combine steep precipitation gradients with sparse access, polar regions combine sparse stations with severe blowing-snow errors, and oceans have very few long-term fixed gauges. Shipboard measurements face airflow distortion and moving sampling locations.\n\nLong records require metadata documenting relocations, instrument replacements, exposure changes, and observation schedules. These changes can introduce artificial shifts into a time series or alter which calendar day receives an accumulation. Record homogenization compares a station with suitable neighbours to identify discontinuities, but precipitation’s spatial variability makes corrections less straightforward than for temperature. Raw observations, quality flags, metadata, and adjusted series therefore serve distinct purposes and should remain distinguishable."
    },
    {
      "heading": "Radar, satellite, and combined precipitation estimates",
      "text": "Weather radar measures electromagnetic energy scattered back by hydrometeors within a sampled atmospheric volume, rather than collecting water at the surface. For spherical liquid drops small relative to the radar wavelength, the reflectivity factor is \\(Z=\\sum_i N_iD_i^6\\), where \\(N_i\\) is the number concentration in a drop-size class and \\(D_i\\) its diameter. Consequently, a few large drops can dominate the signal. Reflectivity is commonly expressed logarithmically in dBZ. Converting it to rainfall rate requires assumptions about drop-size distributions and fall speeds, often summarized by \\(Z=aR^b\\). The widely used relation \\(Z=200R^{1.6}\\), with conventional units of millimetres to the sixth power per cubic metre for \\(Z\\) and millimetres per hour for \\(R\\), is an empirical example, not a universal conversion. Drizzle, thunderstorms, and hail-containing storms can produce different relationships.\n\nDual-polarization radars transmit and receive horizontally and vertically polarized signals. Differential reflectivity indicates particle shape and size characteristics; the correlation coefficient helps distinguish uniform rain from mixed hydrometeors or nonmeteorological echoes; and specific differential phase supports rainfall estimation, especially in heavy rain. These measurements improve precipitation classification but do not eliminate sampling errors. Attenuation weakens the beam along its path, particularly at shorter wavelengths, while mountains and structures can block it. With increasing range, beam height and width increase, so radar may overshoot shallow precipitation or sample conditions far above the surface. Melting snow produces an enhanced-reflectivity “bright band”: wet aggregates scatter strongly and can yield excessive rainfall estimates if interpreted as ordinary raindrops. Evaporation, melting, and wind-driven displacement below the sampled volume further separate radar estimates from surface accumulation.\n\nSatellite remote sensing provides coverage over oceans and regions with sparse ground networks. Passive microwave instruments measure brightness temperatures: liquid hydrometeors modify microwave emission, while ice particles scatter radiation, particularly at higher frequencies. Retrieval algorithms infer precipitation from these signals using physical calculations, databases, and statistical relationships. Infrared instruments instead observe cloud-top radiances; cold, high clouds are associated statistically with rainfall, but non-raining cirrus and warm-topped raining clouds weaken that association. Geostationary infrared observations offer frequent sampling, whereas low-Earth-orbit microwave observations provide intermittent, more physically direct constraints. Spaceborne precipitation radars measure vertical echo profiles along narrower swaths. The Tropical Rainfall Measuring Mission (TRMM), launched in 1997, carried the first spaceborne precipitation radar. The Global Precipitation Measurement (GPM) Core Observatory, launched in 2014, combines a dual-frequency precipitation radar with a microwave imager and provides a reference for a wider satellite constellation.\n\nCombined products exploit complementary observations rather than treating any instrument as ground truth. Gauge adjustment can correct radar or satellite accumulation biases, although gauges themselves have wind-related losses and represent points rather than entire grid cells. Merging methods weight observations by estimated error, interpolate between available measurements, and sometimes use atmospheric-model information. GPM’s Integrated Multi-satellitE Retrievals for GPM (IMERG), for example, combines microwave estimates with infrared information and temporal interpolation to provide half-hourly fields on a \\(0.1^\\circ\\) grid. Its Final product incorporates monthly gauge analysis. Such grid values are retrieved area estimates: grid spacing does not imply that every cell has been directly observed at every output time.\n\nUncertainty varies with surface conditions, precipitation type, and sampling. Snow-covered ground complicates microwave retrievals because its scattering signature can resemble atmospheric ice. Mountains combine radar blockage with sharp precipitation gradients and shallow orographic clouds. Light rain and snowfall may fall below instrument sensitivity, while rapidly developing storms can intensify or dissipate between satellite overpasses, defeating interpolation. Near-real-time products prioritize delivery speed and use the observations then available; delayed research products can incorporate additional gauges, revised calibration, and reprocessing. Comparisons therefore need to identify the product version, processing stream, accumulation interval, and spatial scale."
    },
    {
      "heading": "Forecasting and numerical representation",
      "text": "Quantitative precipitation forecasting (QPF) predicts accumulated liquid-water-equivalent precipitation over a specified area and interval, together with its location and timing. A forecast of 25 millimetres in six hours is different from the same amount over 24 hours, particularly for flood prediction. Forecast products range from hourly grids a few kilometres across to basin-average totals and regional outlooks several days ahead. Precipitation type requires additional information: identical water-equivalent amounts may reach the ground as rain, snow, ice pellets, or freezing rain, depending on the atmospheric temperature profile and particle evolution during descent.\n\nNumerical weather prediction models calculate atmospheric motion, heat transfer, and moisture transport on a three-dimensional grid. Their microphysics schemes represent transfers among water vapour, cloud droplets, rain, and various frozen particles through condensation, evaporation, freezing, melting, and collection. Some schemes predict only the mass of each particle category; others also predict particle number concentrations, allowing size distributions to vary more independently. Latent heating from these processes feeds back on atmospheric motion. In coarser models, individual convective updrafts are unresolved, so convection schemes estimate their collective transport of heat and moisture and their precipitation production. Convection-permitting models, commonly using grid spacings of approximately 1–4 kilometres, represent larger convective circulations explicitly, but still parameterize microphysics and unresolved turbulence. They do not resolve every cloud or turbulent eddy. The United States’ High-Resolution Rapid Refresh system, operational since 2014, provides a practical example: it uses a 3-kilometre grid and updates forecasts hourly.\n\nRadar-based nowcasting extrapolates observed precipitation patterns, often by estimating their motion between successive radar images. Storm-tracking methods follow individual cells, while image-based methods estimate a spatially varying motion field. Such forecasts are especially useful over the next several tens of minutes to roughly two hours when existing storms retain their structure. Their skill declines when cells initiate, intensify, merge, or dissipate; extrapolation alone cannot reliably predict precipitation that has not yet formed. Data assimilation combines observations with an earlier model forecast to estimate the atmospheric initial state. Radar reflectivity and Doppler velocity, satellite observations, and conventional measurements can constrain precipitation structures, winds, and moisture. With increasing lead time, model-predicted storm development becomes more important than simply extrapolating current echoes.\n\nEnsemble forecasting samples forecast uncertainty through multiple simulations with perturbed initial conditions, different model formulations, or stochastic representations of unresolved processes. The proportion of members exceeding a threshold can estimate its probability, although ensemble frequencies often require statistical calibration. A 40 percent probability of measurable precipitation at a location during a stated period describes the chance of exceeding the specified measurable threshold; it does not mean rain will last for 40 percent of that period. Probabilities of exceeding 10, 25, or 50 millimetres communicate accumulation uncertainty more directly than a single deterministic total.\n\nVerification separates several aspects of performance. Occurrence scores count hits, misses, false alarms, and correct non-events relative to a threshold; accumulation errors assess predicted amounts. The Brier score measures the mean squared difference between a forecast probability and the corresponding binary outcome. Reliability asks whether events forecast with, for example, 30 percent probability occur about 30 percent of the time. Spatial verification is essential because an intense storm displaced by 20 kilometres may produce both a miss and a false alarm in gridpoint scoring despite having realistic structure. Neighbourhood methods, including the fractions skill score introduced in 2008, compare forecast and observed fractions exceeding a threshold within surrounding areas. Scores therefore depend on spatial scale, accumulation interval, threshold, and observational uncertainty.\n\nPersistent errors arise from uncertain initial moisture and winds, imperfect terrain representation, and microphysical assumptions governing particle growth and evaporation. Convective initiation is particularly sensitive to shallow boundaries, soil conditions, and small differences in inhibition. Near the rain–snow transition, small errors in the vertical temperature and humidity profiles can change melting, evaporative cooling, and refreezing sufficiently to alter the predicted surface precipitation type."
    },
    {
      "heading": "Hydrological, ecological, and societal effects",
      "text": "Precipitation reaching land is partitioned among vegetation interception, soil storage, surface runoff, and drainage toward groundwater. Leaves and branches retain some water, which may evaporate without reaching the soil; throughfall and stemflow deliver the remainder unevenly beneath vegetation. Infiltration transfers water across the soil surface, while deeper percolation can recharge an aquifer. Runoff develops when rainfall intensity exceeds the soil’s infiltration capacity, or when the soil becomes saturated and cannot accommodate additional water. Consequently, identical rainfall totals can produce different responses: dry, permeable soil may absorb much of a storm, whereas saturated ground, compacted fields, and paved surfaces can generate rapid runoff. Dryness does not always promote absorption, because surface crusts and water-repellent soils can restrict infiltration. At catchment scale, precipitation is balanced by evapotranspiration, water export, and changes in storage, with groundwater transfers also contributing where subsurface and topographic boundaries differ.\n\nRainfall can raise river discharge within minutes in small, steep catchments, while water moving through soils and aquifers sustains flow between storms. Mountain snow stores winter precipitation and releases it during spring and summer melt, separating the timing of precipitation from downstream water availability. In the western United States, snowmelt supplies irrigation systems and reservoirs during the comparatively dry growing season. Accumulated snowfall also adds mass to glaciers; only the portion surviving subsequent melt and other losses contributes to longer-term storage. Agriculture depends on both the amount and availability of water: 1 millimetre of rainfall over a hectare represents 10 cubic metres, but not all of it remains accessible to roots. Irrigation supplements rainfall deficits, while hydropower production depends on inflow, reservoir storage, operating constraints, and the vertical drop through turbines.\n\nPrecipitation influences vegetation structure, primary production, and habitat distribution through soil-water availability and the duration of flooding. In seasonally dry grasslands, rainfall pulses stimulate plant growth and microbial activity, but small events may wet only the surface and evaporate before reaching deeper roots. Rain during germination or flowering can have different consequences from the same amount falling outside the growing season. Excess water can also limit growth by reducing oxygen availability in saturated soils. For wetlands, the depth, timing, and duration of inundation—the hydroperiod—help determine which plants establish and which habitats remain available to fish, amphibians, and waterbirds. Floodplain ecosystems therefore depend not simply on annual rainfall totals but on how rainfall generates river floods and subsequent drawdown.\n\nRaindrop impacts detach exposed soil particles, and overland flow transports them into channels, coupling precipitation to erosion, sediment delivery, and reservoir siltation. Vegetation cover, slope, soil properties, and land management modify these responses. Short, intense storms can produce flash floods, especially in steep valleys and urban catchments, while prolonged rainfall can generate widespread river flooding; sustained wet conditions contributed to the 1993 Mississippi–Missouri floods. Rainfall also triggers landslides by increasing pore-water pressure, thereby reducing effective stress and frictional resistance along potential failure surfaces. These rainfall-driven hazards differ from storm surge, which is an abnormal coastal rise in seawater driven principally by storm winds and atmospheric pressure. Coastal flooding can nevertheless become more severe when surge obstructs the discharge of rainfall-fed rivers and drainage systems.\n\nSolid and freezing precipitation create additional hazards. Hail damages crops, roofs, vehicles, and aircraft; freezing rain deposits ice that loads trees and power lines and reduces traction; heavy snow obstructs transport and can impose substantial structural loads. Snow load depends on water equivalent and density rather than depth alone. Drainage engineers use precipitation intensity, duration, and exceedance probability to size culverts, sewers, and detention facilities, translating rainfall into runoff estimates that account for catchment conditions. Transport operators combine precipitation forecasts with pavement temperatures and observations to organize de-icing and snow clearance. Water-resource planners use seasonal precipitation, snow-water storage, and inflow estimates to allocate supplies and manage the competing reservoir requirements of flood storage, irrigation, hydropower, and environmental flows."
    },
    {
      "heading": "Precipitation in a changing climate",
      "text": "Warming increases the atmosphere’s capacity to hold water vapour, but does not produce a uniform increase in rainfall. The Clausius–Clapeyron relation describes how saturation vapour pressure changes with temperature: near typical surface temperatures, it increases by approximately 7% per degree Celsius. If relative humidity remains approximately constant, atmospheric moisture increases at a similar rate. This thermodynamic relationship is not a prediction that annual precipitation at every location will increase by 7% per degree. Rainfall also requires moisture convergence, ascent, and cloud processes; changes in atmospheric circulation can therefore produce declining precipitation even where the air contains more water vapour.\n\nGlobal-mean precipitation is constrained by the atmospheric energy budget. Condensation releases latent heat, and, in the long-term global average, this heating must approximately balance atmospheric radiative cooling after accounting for sensible heat transferred from the surface. Radiative cooling generally increases less rapidly with warming than atmospheric moisture does. Consequently, climate models typically project global-mean precipitation increases of roughly 1–3% per degree of global warming, substantially below the moisture increase. The response also depends on the forcing: increased greenhouse-gas concentrations can initially reduce atmospheric radiative cooling and precipitation before surface warming produces a compensating increase. Local precipitation is less directly constrained by this global balance because atmospheric transport redistributes both moisture and energy.\n\nThe Intergovernmental Panel on Climate Change concluded in its 2021 Sixth Assessment Report that heavy precipitation had increased in frequency and intensity since the 1950s over most land areas with observations sufficient for trend analysis, and that human-induced climate change was likely the main driver. The assessment projected that extreme daily precipitation would intensify by about 7% per degree of global warming at the global scale, although regional responses differ. Short-duration convective rainfall can show different sensitivities because storm organization, instability, and moisture supply also change. Mean precipitation has a more geographically varied response: projections generally indicate increases at high latitudes and decreases in some subtropical regions, including the Mediterranean. Shifts in storm tracks, monsoon circulation, and wet-season timing can alter seasonal totals independently of changes in the heaviest events. A region can consequently experience fewer wet days but more intense rainfall when storms occur.\n\nWarming also changes precipitation phase. Near the rain–snow transition, higher temperatures generally increase the fraction falling as rain and raise the elevation at which snowfall predominates. Reduced snowfall and earlier melt can diminish seasonal snow accumulation, particularly at lower elevations and in relatively mild mountain climates. In colder environments, however, increased moisture and precipitation can still increase snowfall while temperatures remain sufficiently low. Rain-on-snow conditions likewise have a nonuniform response: they can become more frequent where rain increasingly falls onto an existing snowpack, but less frequent at lower elevations where persistent snow cover disappears. Their hydrological consequences depend on snow temperature, liquid-water storage, frozen-ground conditions, and the energy available for melting, rather than rainfall amount alone.\n\nObserved trends, attribution, and projections answer different questions. A trend describes a measured change over a specified period; detection assesses whether a change can be distinguished from internal variability, while attribution evaluates the contributions of human and natural influences. Event-attribution studies compare the probability or intensity of a defined event in the present climate with a counterfactual climate without human influence; they do not imply that warming was its sole cause. Model projections describe conditional futures under specified emissions or warming levels. Confidence is limited by short or inhomogeneous records, sparse mountain and ocean observations, and variability associated with phenomena such as the El Niño–Southern Oscillation. Model resolution and representations of convection and terrain introduce additional uncertainty, particularly for localized downpours and elevation-dependent rain–snow transitions."
    },
    {
      "heading": "Aerosols, precipitation chemistry, and weather modification",
      "text": "Atmospheric aerosols influence precipitation by supplying cloud condensation nuclei and ice-nucleating particles. At a given liquid-water content, more activated condensation nuclei generally produce more numerous, smaller droplets, potentially delaying collision–coalescence and drizzle formation. Sean Twomey’s 1959 work helped establish the relationship between aerosol populations, updraft velocity, and cloud droplet numbers. Some particles, including mineral dust and certain biological materials, promote ice formation in supercooled clouds; their effectiveness depends on temperature, composition, and particle surface properties. These microphysical effects do not imply a universal change in surface rainfall. Cloud depth, humidity, entrainment, updraft strength, and the availability of liquid water determine whether altered particle populations suppress, redistribute, or sometimes enhance precipitation.\n\nPrecipitation also removes atmospheric substances through in-cloud scavenging, when gases and particles enter growing cloud droplets or ice particles, and below-cloud scavenging, when falling hydrometeors collect material beneath cloud base. Dissolved substances and captured particles then reach the surface through wet deposition, distinct from dry deposition directly onto surfaces. Sulfur dioxide and nitrogen oxides undergo oxidation to produce sulfate and nitrate, contributing to acid deposition. Unpolluted rain equilibrated with atmospheric carbon dioxide has an approximate pH of 5.6, but this is a chemical reference rather than a universal natural baseline: dust, sea salt, organic acids, and ammonia modify precipitation chemistry. Wet deposition transfers nitrogen to terrestrial and aquatic ecosystems, where it can supply nutrients or contribute to nutrient enrichment. Deposition depends on both concentration and precipitation amount; low concentrations during heavy rainfall can still yield substantial chemical fluxes.\n\nChanges at the land surface can also affect precipitation indirectly. Urbanization alters sensible heat release, aerodynamic roughness, and aerosol emissions; urban heating and convergence can influence thunderstorm initiation or rainfall downwind under suitable conditions. Irrigation increases evapotranspiration and often cools the surface, changing both atmospheric moisture and boundary-layer development. Deforestation modifies moisture recycling, roughness, and surface energy partitioning. These influences interact with background winds, atmospheric stability, and regional moisture transport. An observed urban–rural or irrigated–nonirrigated rainfall contrast therefore does not, by itself, isolate a causal effect.\n\nCloud seeding deliberately introduces particles into suitable existing clouds. Glaciogenic seeding targets supercooled liquid water, commonly using silver iodide as an ice-nucleating material or dry ice to generate strong local cooling. Vincent Schaefer demonstrated dry-ice seeding of a supercooled cloud in 1946; Bernard Vonnegut identified silver iodide’s ice-nucleating properties in 1947. The intended sequence is ice initiation, crystal growth using available water, and fallout, but excessive ice competition or evaporation below cloud can limit surface accumulation. Hygroscopic seeding instead introduces soluble particles, often salts, to broaden droplet size distributions and encourage collision–coalescence in liquid clouds. Neither method creates the moisture supply or atmospheric lifting required for clouds to exist.\n\nEvaluating seeding requires separating its effect from large natural variations in cloud development. Randomized allocation of suitable cases to seeded and unseeded treatments, predefined eligibility criteria and outcome measures, and independent observations reduce selection bias. Target–control comparisons require evidence that the areas respond comparably to weather variations. Statistical estimates should report uncertainty, while aircraft measurements, radar, and ground instruments test whether the proposed physical sequence actually occurred. A seeded plume detected aloft is not equivalent to a measurable increase in catchment water supply.\n\nThe 2017 Seeded and Natural Orographic Wintertime Clouds: The Idaho Experiment (SNOWIE) combined aircraft, radar, and surface observations in winter mountain clouds. Analyses published in 2020 quantified snowfall associated with identifiable seeding signatures during selected cases, providing direct evidence that seeding can generate additional snowfall under particular supercooled orographic conditions. Such event-level findings do not establish a fixed seasonal percentage increase, effectiveness in other cloud regimes, or reliable drought relief; those claims require separate, appropriately controlled evidence."
    }
  ],
  "links": [
    "water",
    "cloud",
    "atmosphere-of-earth",
    "snow",
    "water-cycle",
    "climate",
    "climate-change",
    "evaporation",
    "evapotranspiration",
    "drainage-basin",
    "temperature",
    "condensation",
    "pressure",
    "adiabatic-process",
    "latent-heat",
    "humidity",
    "wind",
    "himalayas",
    "tibetan-plateau",
    "el-nino-southern-oscillation",
    "time-series",
    "radar",
    "remote-sensing",
    "measurement-uncertainty",
    "surface-runoff",
    "groundwater",
    "aquifer",
    "river-discharge",
    "glacier",
    "irrigation",
    "primary-production",
    "habitat",
    "wetland",
    "erosion",
    "storm-surge",
    "earth-energy-budget",
    "climate-model",
    "greenhouse-gas",
    "intergovernmental-panel-on-climate-change",
    "ph",
    "ecosystem",
    "urbanization",
    "deforestation"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}