The Köppen climate classification is an empirical system for dividing the world’s climates into categories defined mainly by long-term monthly temperature and precipitation. Developed by the German-Russian climatologist Wladimir Köppen and subsequently revised by other researchers, it distinguishes five principal groups: tropical, dry, temperate, continental, and polar. Additional letters describe rainfall seasonality, summer warmth, winter severity, or the intensity of aridity. Familiar codes such as Af, BWh, Csa, and Dfb therefore summarize different combinations of thermal and moisture conditions.
The system was designed to approximate the climatic boundaries associated with major vegetation formations rather than to classify atmospheric circulation directly. Its enduring importance comes from the combination of ecological relevance, limited data requirements, and a compact notation suitable for global maps. It is widely used in geography, meteorology, ecology, and environmental research. However, “Köppen classification” does not denote a single unchanging algorithm: published versions differ in several thresholds and implementation details, and a location’s assigned class also depends on the observation period and spatial resolution.
Historical development and botanical foundations
Köppen’s classification emerged from nineteenth-century efforts to connect plant geography with measurable atmospheric conditions. Earlier descriptions commonly divided the world into broad latitudinal belts, but latitude alone could not explain deserts beside humid regions, mild maritime winters, or cold conditions on tropical mountains. Botanical geography offered a different organizing principle: vegetation reflected the combined effects of heat, moisture, and seasonality. Köppen sought numerical climatic limits that would reproduce these broad biological divisions.
An important influence was the work of the botanist August Grisebach, whose 1872 study of world vegetation organized plant formations in relation to environmental conditions. Köppen’s own background combined natural science with meteorological observation. His early climatic regionalization appeared in 1884, followed by substantial revisions, notably in 1900 and 1918. These were stages in a continuing effort to turn broad vegetation-associated zones into a practical classification based on instrumental records.
The botanical foundations remain visible in the system’s thresholds. A coldest-month temperature of separates tropical climates from cooler groups. A warmest-month temperature near distinguishes climates capable of supporting extensive tree growth from polar climates. These limits are useful approximations, not universal physiological boundaries. Tree survival also depends on species, soil conditions, snow cover, exposure, and the length and intensity of the growing season.
Köppen and Alfred Wegener applied climatic and geological reasoning in their 1924 book Die Klimate der geologischen Vorzeit, concerning climates of the geological past. Köppen later contributed his mature classification to the Handbuch der Klimatologie, edited with Rudolf Geiger. Geiger continued revising and disseminating the system after Köppen’s death in 1940, giving rise to the widely used name Köppen–Geiger classification.
Twentieth- and twenty-first-century researchers converted the classification into increasingly standardized digital procedures. Important global mapping studies include those by Kottek and colleagues in 2006, Peel, Finlayson, and McMahon in 2007, and Beck and colleagues in 2018. These studies differ in their datasets, spatial resolution, baseline periods, and some classification conventions. Consequently, their maps should be understood as distinct implementations of a shared framework rather than interchangeable reproductions of one original map. The system’s history is therefore both conceptual and computational: its organizing ideas have remained recognizable while its observational foundations and mapping methods have changed considerably.
Classification logic, symbols, and threshold conventions
The classification uses twelve long-term monthly mean temperatures and twelve long-term monthly precipitation totals for each location. These are commonly calculated over a climatological normal period of about thirty years. Temperature values are generally expressed in degrees Celsius and precipitation in millimetres. From these monthly values, an implementation derives annual mean temperature, annual precipitation, the temperatures of the warmest and coldest months, and several measures of rainfall seasonality.
The five principal groups are:
| Group | Broad description | Main distinguishing feature |
|---|---|---|
| A | Tropical | Every month has a mean temperature of at least |
| B | Dry | Annual precipitation falls below a temperature- and seasonality-dependent threshold |
| C | Temperate | Winters are relatively mild, with at least one sufficiently warm month |
| D | Continental or snow | Winters are cold, but at least one month is sufficiently warm |
| E | Polar | The warmest month remains below |
The dry-climate test is normally applied before assigning the non-arid thermal groups. Thus a hot desert is BWh, not an A climate, even when every month exceeds . This precedence is essential: the B group recognizes water limitation, whereas the other principal groups primarily organize thermal conditions.
The second letter has different meanings in different groups. In A climates, f, m, and w or s distinguish rainforest, monsoon, and savanna rainfall regimes. In B climates, W means desert and S means steppe. In C and D climates, s, w, and f indicate dry summers, dry winters, and the absence of either qualifying dry season. For E climates, T and F distinguish tundra from ice-cap conditions. Third letters further describe heat or cold: examples include a for a hot summer, b for a warm summer, and h or k for hot or cold dry climates.
A consequential variation concerns the C–D boundary. Some implementations use a coldest-month mean of , whereas others use . A locality with a coldest month averaging can therefore be D under the first convention and C under the second, without any disagreement about its observations. The descriptions below use the boundary.
Precise treatment of equality at thresholds also varies. A commonly used convention assigns C climates a coldest month above but below , D climates a coldest month at or below , and both a warmest month of at least . Published comparisons must specify such details, particularly where rounded monthly data place a location directly on a boundary.
Tropical climates: rainforest, monsoon, and savanna
Group A includes non-arid climates in which every month has a mean temperature of at least . Temperature seasonality is often small near the equator, although daily temperature ranges can be substantial. Because sustained winter cold is absent, rainfall amount and seasonal distribution provide the main subdivisions. The group encompasses continuously wet equatorial regions, strongly seasonal monsoon environments, and tropical climates with pronounced dry seasons.
An Af, or tropical rainforest, climate has at least mm of precipitation in its driest month. This criterion excludes a climatologically dry month under the classification’s rules, although it does not imply rain every day or identical rainfall throughout the year. Equatorial portions of the Amazon Basin, the Congo Basin, and the Malay Archipelago contain extensive Af regions. Singapore provides a familiar urban example. Persistent moisture convergence and frequent convection help sustain rainfall, but local topography and exposure still produce important variation.
An Am, or tropical monsoon, climate has a driest month below mm but receives enough annual precipitation to satisfy:
where is driest-month precipitation and is annual precipitation, both in millimetres. The annual-total term allows a short dry season where the wet season is exceptionally rainy. For example, if annual precipitation is mm, the threshold is mm. A driest month of mm would then qualify as Am rather than savanna, provided the other requirements are met. Parts of coastal India and tropical coastal regions elsewhere exhibit this combination.
Tropical climates that fail both the Af and Am tests are generally classified as Aw, with a winter dry season, or As, with a summer dry season. Aw is much more widespread. Its distribution includes extensive areas of tropical Africa, central Brazil, and northern Australia. Seasonal movement of tropical rainfall belts and changes in monsoon circulation produce a wet warm season followed by months of reduced rainfall. Some implementations simplify tropical savanna classification and do not retain As as a separate mapped class.
Names such as “rainforest” and “savanna” describe climatic affinities, not guaranteed vegetation cover. Fire, grazing, soil properties, drainage, and land use influence whether a landscape supports closed forest, woodland, or grassland. Seasonal flooding can maintain grasslands within otherwise humid regions, while deep-rooted forests can survive a marked dry season. The Köppen code therefore identifies a climatic setting rather than directly mapping an ecosystem or land-cover type.
Dry climates: the aridity threshold and its subdivisions
Group B differs from the other main groups because its first-order distinction is moisture deficiency. A fixed annual rainfall limit would perform poorly across climates: mm of annual precipitation has different implications in a cold interior basin and a hot subtropical lowland. Köppen therefore used annual mean temperature and rainfall seasonality as simple proxies for the atmospheric demand for water.
In a widely used modern formulation, the annual precipitation threshold , in millimetres, is:
Here is annual mean temperature in degrees Celsius. Summer is conventionally April–September in the Northern Hemisphere and October–March in the Southern Hemisphere, with winter comprising the opposite six months. The larger allowance for summer rainfall reflects its coincidence with generally greater evaporative demand.
A climate belongs to B when annual precipitation is below . It is BW, or desert, if precipitation is below half that threshold; otherwise it is BS, or steppe. The third letter is h when annual mean temperature is at least , and k when it is lower. These rules yield BWh, BWk, BSh, and BSk. “Cold desert” thus refers to annual mean temperature under the specified convention, not necessarily to cool summers.
For a location with an annual mean temperature of and no strongly dominant rainfall half-year, the threshold is mm. Annual precipitation below mm gives BW, while a total from mm up to but not including mm gives BS. If of the same location’s rainfall arrives in summer, the threshold rises to mm. The same annual rainfall can consequently receive a different classification when its seasonal distribution changes.
BWh occurs across much of the Sahara and the Arabian Peninsula. BWk characterizes many continental and elevated deserts, including parts of Central Asia. BSh and BSk mark semiarid belts such as portions of the Sahel and the Eurasian steppe. Their formation reflects several mechanisms: descending subtropical air, continental distance from moisture sources, cold coastal waters, and rain shadows behind mountain ranges.
The aridity equation is empirical rather than a complete physical water balance. It does not explicitly calculate radiation, wind, soil-water storage, or potential evapotranspiration. Nor does annual precipitation alone measure water availability from rivers, snowmelt, or irrigation. A B climate is a long-term climatic category; it is not synonymous with a temporary drought, which can also occur in normally humid regions.
Temperate climates: maritime, Mediterranean, and subtropical patterns
Under the boundary convention, group C comprises non-arid climates with a coldest-month mean above but below , and a warmest-month mean of at least . The group is often called “temperate,” although it includes many climates commonly described as subtropical. Its wide geographic range reflects mild oceanic winters, warm eastern continental margins, Mediterranean rainfall regimes, and cool tropical highlands.
The second letter describes rainfall seasonality. A dry summer, designated s, generally requires the driest summer month to receive less than mm and less than one-third of the precipitation of the wettest winter month. A dry winter, designated w, requires the driest winter month to receive less than one-tenth of the precipitation of the wettest summer month. f indicates that neither dry-season test is satisfied. It does not mean that every season receives equal rainfall. Unusual rainfall distributions can satisfy competing seasonal tests, so classification software must specify precedence.
The third letter distinguishes summer temperatures. a indicates a warmest month of at least . b indicates a warmest month below , with at least four months averaging at least . c indicates only one to three months at or above , again without a hot summer. These combinations distinguish climates with very different growing seasons even where their winters are similarly mild.
Csa and Csb are usually called Mediterranean climates. Their wet winters and dry summers occur around the Mediterranean Sea, in coastal California, central Chile, the southwestern Cape, and parts of southern Australia. Summer dryness is associated with subtropical high pressure, whereas winter rainfall accompanies the seasonal return of midlatitude storms. Hotter inland settings tend toward Csa, while cooler coastal settings often favor Csb. The classification describes the observed thermal and rainfall pattern, not the circulation mechanism itself.
Cfa, commonly called humid subtropical, occurs in parts of the southeastern United States, eastern China, southern Japan, and southeastern South America. Cwa also has hot summers but a qualifying winter dry season, prominent in parts of monsoon-influenced Asia. Elevated tropical and subtropical regions can support Cwb or Cwc climates: altitude lowers temperature sufficiently to exclude group A despite a low latitude.
Cfb, usually called oceanic or marine west coast, is widespread in western Europe and New Zealand, while Cfc describes cooler maritime climates with short summers. Oceanic thermal moderation limits winter cold and summer heat. These labels nevertheless encompass substantial differences in cloudiness, wind, total rainfall, and frost frequency that are not separately encoded.
Continental climates and the role of winter severity
Group D includes non-arid climates with a coldest-month mean at or below and a warmest-month mean of at least , using the convention adopted here. These climates are concentrated in the Northern Hemisphere, especially across North America and Eurasia. The Southern Hemisphere has comparatively little land at the latitudes where large seasonal temperature contrasts develop, so extensive lowland D climates are largely absent there.
The group’s strong annual temperature range results mainly from continentality and high-latitude seasonality. Large landmasses cool rapidly during long winter nights, while their interiors lack the moderating influence of nearby oceans. Snow cover can reinforce winter cooling by reflecting sunlight and insulating the ground from atmospheric heat exchange. Summer solar input can nevertheless support warm or hot conditions, separating D climates from polar environments with persistently cool summers.
The precipitation letters f, s, and w use the same broad seasonal tests as in group C. Df climates do not meet the specified dry-summer or dry-winter criteria. Dw climates have strongly dry winters, especially in eastern Asia, where winter continental high pressure contrasts with summer moisture transport. Ds climates have qualifying dry summers and occur more locally, including elevated terrain in some continental regions adjacent to summer-dry climatic regimes.
The summer letters a and b distinguish hot-summer and warm-summer climates. Dfa and Dfb are frequently called humid continental climates, though the word “humid” should not be interpreted as a direct measurement of year-round atmospheric humidity. Dfa occurs in parts of the central and eastern United States and southeastern Europe; Dfb is extensive in southern Canada, northeastern Europe, and western Russia. Their ecological and agricultural differences partly reflect growing-season length, winter cold, and moisture timing.
Climates with only one to three months averaging at least are generally assigned c, unless their coldest month falls below approximately , in which case d identifies an exceptionally severe winter. Dfc and Dwc occupy large subarctic regions, while Dfd and Dwd occur in especially cold continental interiors, notably parts of northeastern Siberia. Minor conventions at the boundary differ among implementations.
The terms “continental,” “snow,” and “subarctic” emphasize different characteristics and are not exact synonyms. Snowfall is not itself an input to the classification: precipitation totals ordinarily combine water equivalents of rain and snow. Nor does a D code specify snow depth, soil freezing, or the presence of permafrost. Those depend on additional controls, including insulation by snow, vegetation, ground materials, and the duration of cold conditions.
Polar climates and the treatment of mountains
Group E consists of climates whose warmest month averages below . Unlike D climates, they lack even a brief period of sufficiently high monthly mean temperature to cross the system’s approximate tree-growth threshold. The decisive feature is therefore summer cold, not merely a severe winter. Some maritime polar locations have less extreme winter temperatures than continental interiors classified in group D.
The two principal subdivisions are ET, tundra climate, and EF, ice-cap climate. ET has a warmest-month mean of at least but below . EF has every monthly mean below . These limits refer to monthly averages, not to the absolute occurrence of thawing or freezing. An ET location can experience frost during every month, while brief above-freezing episodes can occur at an EF site without raising a monthly average above zero.
ET climates occur along many Arctic coasts and islands, on portions of the Antarctic Peninsula and nearby islands, and in high mountains. Their vegetation may include mosses, lichens, sedges, grasses, and low shrubs where soils and moisture allow growth. EF climates characterize the interiors of the great ice sheets of Antarctica and Greenland. However, the climate classification does not directly measure ice thickness, glacier movement, or surface mass balance.
Low temperatures reduce atmospheric moisture capacity, so some polar regions receive little precipitation. Much of interior Antarctica is extremely dry in terms of annual precipitation, yet it is normally classified EF rather than as an ordinary hot or cold desert. Snow can accumulate over long periods because melting and other losses are limited. The climatic distinction between an ice-cap environment and a B desert illustrates why low precipitation alone does not define the major groups.
High elevations create an important interpretive issue. The basic system has no separate causal category for mountain climates: a high station is classified from its monthly temperature and precipitation in the same way as any other station. Tropical mountains can therefore contain C, ET, and EF conditions in successive elevation belts. Parts of the Tibetan Plateau contain cold, dry, and tundra-classified environments over relatively short distances.
Some modified maps add H for highland climate. This is a supplementary convention, not one of the five principal Köppen groups, and it has no universally accepted set of thresholds. It can conveniently emphasize altitude, but it also combines environments with very different rainfall seasons, thermal ranges, and exposure. Orographic precipitation, slope orientation, valley inversions, and wind redistribution of snow create mountain-scale variation that a single highland label—or a coarse-resolution Köppen map—cannot fully represent.
Constructing global maps from observations and gridded data
A Köppen map is the output of both classification rules and a climate dataset. Early maps relied on relatively sparse station observations and expert interpretation. Modern maps usually apply the rules to gridded monthly temperature and precipitation fields within a geographic information system. Each grid cell receives a class, producing a categorical spatial dataset rather than a direct observation of a sharply bounded natural region.
The first major methodological choice is the reference period. A map based on 1961–1990 conditions can differ from one based on 1991–2020 because of climatic change, natural variability, and changes in observational coverage. The period’s length also matters: short records are more sensitive to unusual decades. Classification is normally applied to the twelve climatological monthly values, rather than by classifying each individual year and selecting the most frequent annual result. These procedures need not produce identical answers.
A second choice concerns spatial estimation. Weather stations are unevenly distributed, with sparse coverage in deserts, polar regions, and many mountains. Spatial interpolation estimates conditions between stations using distance, elevation, and other predictors. Temperature generally responds strongly to elevation, whereas precipitation depends more irregularly on relief, prevailing winds, and exposure. Combining observations with a digital elevation model can improve spatial detail, but small grid cells do not guarantee equally detailed underlying evidence.
The widely used 2007 map by Peel, Finlayson, and McMahon employed a grid, with a north–south spacing of roughly 11 km. Beck and colleagues’ 2018 work presented global maps at approximately 1-km resolution, including a 1980–2016 historical baseline and future projections. Such products made local climatic contrasts more visible, particularly in complex terrain. Nevertheless, the apparent precision of a boundary on a high-resolution map should not be confused with certainty about its exact position.
Classification uncertainty is greatest near thresholds. A small change in estimated coldest-month temperature can switch a cell from Cfb to Dfb under the convention. A modest precipitation difference can move a semiarid cell across the BS–BW boundary. Some mapping studies therefore compare multiple input datasets or provide confidence measures based on agreement among alternative classifications.
Presentation introduces further decisions. Color schemes, map projection, and cartographic generalization influence how patterns appear. A categorical raster can preserve isolated mountain cells that disappear on a generalized world map. Conversely, coarse grids can blend coastlines, islands, and mountain belts. Reproducible mapping consequently requires documentation of the climate period, input fields, grid resolution, threshold conventions, missing-data treatment, and classification algorithm.
Applications in ecology, environmental analysis, and comparison
The classification’s principal practical strength is that it converts many climatic measurements into a small number of recognizable categories. Researchers can compare areas sharing broad thermal and rainfall regimes without listing twenty-four monthly values for every site. This compression supports regional description, sampling design, and global synthesis, especially where the objective is to distinguish broad environmental settings rather than explain detailed physical processes.
In ecology, Köppen classes provide a useful background for studying species distributions, vegetation structure, and biodiversity. The link is strongest at broad scales because the system was designed around vegetation-associated climatic limits. Researchers may compare forest traits across Af, Am, and Aw regions or examine differences between Mediterranean and humid temperate environments. Nevertheless, climatic similarity does not establish ecological equivalence: evolutionary history, soils, disturbance, and dispersal barriers can produce different communities within the same class.
Agricultural analysis uses the categories to describe broad growing environments. Csa suggests a combination of winter rainfall and summer dryness, whereas Cwa indicates a strong warm-season rainfall concentration. These distinctions help organize comparisons of crop calendars and water demand. A Köppen code alone, however, does not provide frost dates, accumulated growing-degree days, soil-water capacity, or crop-specific heat limits. Such variables often matter more directly for an individual crop than the climate-group name.
In hydrology, climate classes can help stratify catchments for comparative studies. Rainfall seasonality and temperature affect evapotranspiration, snow storage, streamflow timing, and runoff generation. Two drainage basins with the same class can still behave differently because of geology, basin size, vegetation, and human regulation. Classification is therefore often one explanatory variable among many, rather than a substitute for a measured water budget or a process-based hydrological model.
The system also appears in building research, energy studies, and analyses of urban environments. Researchers compare heating and cooling requirements or building performance across broad climate types. However, building standards frequently use their own climatic zones, based on degree days, moisture criteria, or regulatory requirements. Those zones are not automatically equivalent to Köppen classes. Local urban heat islands can additionally make a city station warmer than its surrounding countryside, affecting threshold-based classifications near a boundary.
Its categorical form is particularly convenient for statistical grouping, but that convenience entails information loss. Two sites on opposite sides of a threshold may be more climatically similar than two distant sites within one class. Comparisons based on the classification therefore benefit from retaining the underlying continuous temperature and precipitation variables. The letters communicate an environmental pattern; they do not establish a causal explanation for every ecological, economic, or social difference between regions.
Climate change, boundary shifts, and interpretive limits
Köppen classifications provide an intuitive way to describe climate change because shifts in temperature and precipitation can move locations between familiar categories. Warming may convert an ET climate into a D or C climate when the warmest month crosses , or move a D climate into C as winters become milder. Increasing summer temperatures can change a b summer designation to a when the warmest month reaches .
Dry-climate transitions can occur through temperature as well as rainfall. In the standard aridity equation, a increase in annual mean temperature raises the precipitation threshold by mm, assuming the seasonal rainfall category remains unchanged. A site can consequently move into group B even without a decline in annual precipitation. This is a property of the empirical classification and should not be mistaken for a complete estimate of changing evaporative demand.
Future maps commonly apply the rules to monthly outputs from general circulation models or Earth system models. Because global model grids are relatively coarse, researchers may use downscaling and bias adjustment before classification. Different emissions scenarios, models, and processing methods produce different boundary shifts. A climate-model ensemble can show where projected transitions are consistent across simulations and where they remain sensitive to uncertain temperature or precipitation changes.
A mapped class transition does not imply instantaneous ecosystem replacement. Long-lived trees may persist after climatic averages cross a threshold, while migration, seed establishment, soil development, and disturbance operate at different rates. Conversely, ecological damage can occur without a class change if heat extremes, drought duration, fire weather, or pest pressures intensify within the same broad category. Climate-zone movement and vegetation movement are related but distinct phenomena.
The system also omits many important dimensions of climate. Monthly means conceal heatwaves, cold extremes, rainfall intensity, and the sequence of wet and dry days. Annual or monthly precipitation does not directly express flood risk, infiltration, or plant-available moisture. Variability associated with the El Niño–Southern Oscillation can strongly affect a region even when its long-term Köppen code remains unchanged. Wind, solar radiation, and atmospheric moisture are not explicit classification inputs.
Alternative systems emphasize different purposes. The Thornthwaite classification gives a more explicit role to evapotranspiration and moisture balance, while the Trewartha classification modifies thermal divisions to improve some regional and vegetation relationships. Köppen–Geiger remains especially useful for compact global description, but its boundaries are operational thresholds within continuous climatic fields. A shift from one code to another identifies a change relative to those thresholds, not a universal physical discontinuity or a complete measure of environmental transformation.