Thunderstorms are among the most striking and dramatic weather phenomena. On a global scale, thousands of thunderstorms occur each day. According to estimates, as many as 40,000 thunderstorms can form daily, and at any given moment, nearly 1,800 thunderstorm events may be happening simultaneously around the world. Their occurrence and intensity vary depending on the season and geographic location. The highest thunderstorm activity is typically observed during the warmer months when the atmosphere has ample energy available.

Climatological mean annual lightning stroke density (2010–2020); Earth System Science Data

In the Northern Hemisphere, we are now entering the second half of summer, which brings a gradual decline in thunderstorm activity. Days are getting shorter, solar energy is decreasing, and with it, the likelihood of strong thunderstorms is also diminishing. While May, June, and July are statistically the most active months, August and September tend to be calmer. Nevertheless, very strong and dangerous storms can still occur during this period.

2024 Severe weather report for the USA; NOAA

Given their diversity, intensity, and impacts, thunderstorms deserve closer attention. In this article, we will focus primarily on on the conditions that lead to their formation and the factors that determine their strength.

What is a thunderstorm?

A thunderstorm, or convective storm, is a combination of optical phenomena (such as rain, hail, shelf cloud, wall cloud), electrical phenomena (lightning), and acoustic phenomena (thunder) that accompany the development of cumulonimbus cloud formations.

An observer at a meteorological station records a thunderstorm when lightning is seen and thunder is subsequently heard. The occurrence of these electrical and acoustic phenomena is the decisive criterion for officially recording a thunderstorm in the station log and marking it on a synoptic weather map. Precipitation or dark clouds alone, without accompanying lightning and thunder, are not sufficient for a thunderstorm to be recorded.

Synoptic chart of Europe issued by DWD for 5 August 2025, 18 UTC; DWD

Thunderstorm Project

Much of what we know today about thunderstorms comes from the Thunderstorm Project, carried out in the 1940s by the U.S. Weather Bureau. It was an exceptionally ambitious field research effort in which scientists used a network of surface measurements, radar observations, and aircraft flights through storm clouds to study thunderstorms in great detail. The project focused on thunderstorm activity over Ohio and Florida.

The Thunderstorm Project in Ohio, 1947; NOAA

One of the most important outcomes of this project was the development of the concept of the thunderstorm cell (single cell) and the identification of key structural components of thunderstorms: updrafts, downdrafts, inflow, outflow, boundary interfaces, and others.

Thunderstorm cell concept

The basic "building blocks" of every thunderstorm are thunderstorm cells. During its life cycle, a thunderstorm cell goes through three developmental stages: the cumulus stage, the mature stage, and the dissipating stage.

Life cycle of a thunderstorm cell; NOAA

The cumulus stage

In the growth stage, a thunderstorm cell is formed by an updraft of air rising from the surface. At first, it rises as an invisible “bubble.” When condensation or desublimation of water vapor occurs, a visible cloud begins to form, composed of cloud droplets or ice crystals.

Cumulus humilis; WMO

If the updraft weakens, the cloud stops growing and only small cumulus clouds (cumulus humilis) form, typical of calm, warm summer weather.
However, if the updraft is strong enough, the cloud continues to grow. Microphysical processes inside the cloud (e.g. collision and coalescence of particles) lead to the growth of droplets and crystals.
Once the particles become large enough that their weight overcomes the strength of the updraft, they begin to fall and form a downdraft.

Mature stage of a thunderstorm cell

During the mature stage, the cumulonimbus cloud reaches its full development, with both updrafts and downdrafts occurring simultaneously. This is the most intense phase of the thunderstorm cell, marked by strong precipitation and frequent lightning. The updraft can rise all the way to the tropopause, or even into the lower stratosphere, where it encounters warmer air that halts its ascent and forces it to spread horizontally. This process forms the characteristic anvil shape of the cloud, which can extend more than 100 km ahead of the storm core.

Cumulonimbus cloud with lightning and tilted updraft. Captured in Lošinj, Croatia. Photographer: Sandro Puncet (Croatia); WMO.

Above the anvil, vertical protrusions known as overshooting tops (also called "warm domes") often appear. These indicate a very strong updraft penetrating into the lower stratosphere. The updraft continues to be fueled by inflow, which supplies warm, moist air from below into the storm.

An idealized supercell; NOAA

The downdraft forms as a result of falling precipitation particles. The evaporation or sublimation of these particles cools the descending air, which then spreads out upon reaching the surface and creates what is known as a cold air outflow, a diverging current of cool air near the ground.

A microburst is a localized column of rapidly descending air within a thunderstorm, capable of causing severe surface damage and posing serious risks to life; NOAA

If the boundary between the cold outflow and the surrounding warm air is well-defined, it is referred to as a gust front. This boundary is often accompanied by gusty winds and can be visually identified by the presence of arcus clouds, such as shelf clouds.

Diagram of updrafts and downdrafts in a thunderstorm, illustrating the gust front in a vertical cross-section and on a radar reflectivity image (small inset); NOAA

In the next stage of storm evolution, the leading edge of the outflow may extend as far as 200 km from the original location of the thunderstorm cell and can serve as a triggering mechanism for the formation of new thunderstorm cells.

Shelf cloud. Photographer: Danijel Palčić. Taken in Pag, Croatia; WMO

Dissipating stage of a thunderstorm cell

The final stage in the life cycle of a thunderstorm cell is the dissipating stage. At this point, the cold air outflow cuts off the supply of warm, moist air from the surface, effectively halting the updraft. This marks the end of the thunderstorm cell's life cycle.

Mammatus clouds belong to the late mature stage or early dissipation stage of a storm, when strong updrafts begin to weaken and the cloud starts to break apart; Met Office

During this phase, light precipitation typically dominates, with little or no significant thunderstorm activity. It is estimated that only about 20% of the condensed water vapor formed in the updraft reaches the ground. The rest either evaporates in the downdraft or remains as residual cloudiness, such as thinly scattered cirrus clouds, which gradually mix with the surrounding atmosphere and completely dissipate.

The updraft velocities measured during the Thunderstorm Project most frequently ranged between 5 and 10 m/s (18 and 36 km/h; 11 and 22 mph), occasionally exceeding 15 m/s (54 km/h; 34 mph). Downdrafts showed approximately the same velocity values, but were generally weaker than the updrafts. The vertical extent of the convective layer reached approximately 10 km. The lifespan of a single thunderstorm cell ranged from 30 to 40 minutes.

Essential ingredients for thunderstorm formation

What determines whether, on a warm and sunny day, only fair-weather cumulus clouds appear in the sky, or whether a towering thunderstorm cloud will develop?

Stability/instability in the atmosphere and vertical extent of cloudiness: cumulus humilis, cumulus congestus, and cumulonimbus; Cengage 2012, atoc.colorado.edu

For thunderstorm clouds to form, three key conditions must be met:

• Conditionally unstable atmosphere

• Moist air in the lower and middle troposphere:
  Moist air serves as the fuel for thunderstorms.When water vapor in the cloud condenses or desublimates, latent heat is released, which provides the storm with additional energy.

• Lift (also known as the triggering mechanism):
  A triggering impulse is essential to initiate the upward movement of air to a level where it becomes warmer than the surrounding atmosphere and begins to rise autonomously, forming thunderstorm clouds. This impulse may originate from a frontal boundary, low-level convergence, differential surface heating, a dryline, a mountain slope, or even a small hill. Occasionally, an outflow of cold air from a nearby rain shower or thunderstorm may also initiate new convection. Accurately forecasting where such lift will occur remains one of the most challenging aspects of thunderstorm prediction.

What does it mean when the atmosphere is conditionally unstable?

Let’s start with the basics. A stable atmosphere resists upward motion, in such conditions, air rises only if some external force pushes it upward. An unstable atmosphere, on the other hand, encourages vertical motion. Whether the atmosphere supports or suppresses upward movement depends on its temperature stratification, specifically on how rapidly the temperature decreases with height.

Illustration demonstrating air parcel stability and instability. The bowl represents the atmospheric state, while the red ball symbolizes an air parcel that receives energy to initiate movement; NOAA

Atmospheric stability is most commonly assessed using the air parcel method.

Imagine a bubble of air. As it is lifted, it begins to cool due to expansion. This happens because the pressure decreases with altitude in the atmosphere, causing the parcel to expand and lose heat.

Several physical assumptions apply to our air parcel. For instance, we assume that it cools adiabatically, meaning without exchanging heat with its surroundings.

If the air is dry, the parcel cools at a rate of approximately 10 °C per kilometer. However, if it contains enough water vapor, condensation might begin as it cools. This process releases latent heat, which slows the rate of cooling, typically reducing it to between 4 and 7 °C per kilometer.

And this is where conditional instability comes into play. It occurs when the temperature decreases with height in such a way that:

A dry air parcel is cooler than the surrounding air as it rises and therefore tends to sink back to its original position. In this case, the environment is stable for the dry parcel.

A saturated parcel is warmer than the surrounding air as it rises and continues to rise due to buoyancy. In this case, the environment is unstable for the saturated parcel.

Example of conditionally unstable temperature stratification on the Skew-T log-P diagram (Windy.com): The red curve shows the environmental temperature profile, T is the measured air temperature at a level, Tp is the air parcel temperature. The blue bubble represents an air parcel with arrows showing buoyant force direction; Windy.com

Our air parcel could rise from the ground in a conditionally unstable atmospheric profile as follows:

• At the beginning, the temperature of our air parcel is the same as the surrounding air (environment). To start rising, the air parcel needs a lifting mechanism.

• As it ascends, it cools at the dry adiabatic lapse rate (approximately 10 °C per kilometer, as shown on a thermodynamic diagram following the dry adiabat).

• As the air parcel cools, its humidity increases until it reaches 100%, causing condensation to begin and a visible cloud to form. The height where this saturation occurs is called the lifting condensation level (LCL).

• During further forced ascent, it cools at the moist adiabatic lapse rate (as shown on a thermodynamic diagram following the moist/wet adiabat).

• At the level of free convection (LFC), its temperature equals that of the surrounding air.

• Above this level, the air parcel becomes warmer than its surroundings and begins to rise freely due to buoyancy.

• The parcel will rise freely until its temperature matches that of the surrounding air once again (the level of neutral buoyancy), at which point its ascent will begin to slow. This occurs no later than the tropopause, where temperature starts to increase sharply with altitude.

During intense thunderstorms, an air parcel can ascend all the way through the troposphere, sometimes reaching the tropopause and even slightly entering the lower stratosphere.

Significant levels on the thermodynamic diagram: LCL, CCL, LFC; NOAA

The amount of energy available to our air parcel for rising due to buoyant force can be expressed by a quantity called CAPE, which stands for Convective Available Potential Energy.
It is one of the key parameters in meteorology, describing the degree of atmospheric instability, and therefore the potential for the development of convective phenomena such as thunderstorms, showers, or hazardous accompanying weather events.

One might ask whether an absolutely unstable atmosphere is also favorable for thunderstorm development. That is, an atmosphere that is unstable for both dry and saturated air parcels. The answer is no. This type of stratification leads to strong turbulent mixing, which quickly restores thermal equilibrium within the affected layer of the atmosphere. Therefore, absolute instability occurs only rarely, typically in a shallow layer near the surface. This can happen, for example, during intense solar heating or when very cold air moves over a warm body of water.

The role of wind shear in the dynamics of convective storms

Conditional instability, moisture, and a lifting mechanism influence the development and intensity of the upward air motion. Another important parameter that plays a significant role in the dynamics of convective storms is wind, particularly its direction and speed at different altitudes above the ground. The change in wind speed and direction with height is known as wind shear. Although wind shear is not essential for thunderstorm initiation itself, it has a crucial impact on storm structure.

The strength of the shear causes a vertical tilt of the updraft and downdraft.

In weak shear conditions, the downdraft often cuts off the updraft from the supply of moist air in the lower troposphere, which leads to the rapid dissipation of the storm.

On the other hand, strong wind shear tends to separate and tilt these flows, allowing for the development of a more intense and longer-lasting storm, with a greater likelihood of producing hazardous accompanying phenomena.

A storm with little wind shear has a vertical updraft quickly weakened by rain, while strong wind shear causes a tilted updraft that keeps rain away from it; BOM

When everything falls into place…

When atmospheric instability, a lifting mechanism, and wind shear come together in the right combination, very powerful thunderstorms can develop.

These storms may be accompanied by a wide range of hazardous phenomena, such as strong gusty winds, downbursts (microbursts), derechos, heavy rain, flash flooding, hail, lightning, and even tornadoes. Such events can cause extensive damage to property, disrupt infrastructure, and pose a serious threat to human life.

Tornado; WMO

How to observe thunderstorms on Windy.com

Windy.com provides a range of interactive layers and tools that allow users to monitor ongoing thunderstorms and analyze the conditions that may lead to their development.

For direct observation of thunderstorms, the following layers are especially useful:

• Radar – shows current precipitation. Intense storms appear as well-defined cores with high reflectivity.

• Satellite – displays cloud cover.
  On VIS (visible spectrum) images, storm structure such as overshooting tops can be clearly seen during daylight hours.
  On NIR (near-infrared) images, vertically extensive thunderstorm clouds have very cold tops that are represented by a color scale on the NIR product, making them easy to identify.

Satellite image capturing severe thunderstorms over the Northern Plains and Canadian Prairies, 7 August 2025; Windy.com

To analyze thunderstorm potential, these tools and forecast layers are helpful:

• Radiosondes (Sounding) – helps evaluate the temperature profile and atmospheric instability.

• CAPE – the higher the value, the more energy is available for convection. Values above 1000 J/kg suggest a potential for strong thunderstorms.

• Rain and thunderstorm – a forecast layer showing modeled precipitation and thunderstorm activity.

• Wind at altitude – wind direction and speed at various pressure levels (e.g., 850 hPa and 500 hPa) can reveal the presence of wind shear, which plays a key role in storm organization (e.g., multicell or supercell development).

• Pressure fields and fronts – thunderstorms often form in low-pressure zones, along frontal boundaries, or in convergence areas, which are clearly visible on the pressure map.

The combined use of these data provides both a real-time overview of storm activity and valuable forecasts of where new thunderstorms are likely to develop within the coming hours.

Image of radar reflectivity capturing severe thunderstorms over the Northern Plains and Canadian Prairies, 7 August 2025; Windy.com

September is the first meteorological month of autumn in the Northern Hemisphere. Days grow shorter and solar energy decreases, yet the oceans, thanks to their enormous heat capacity, still retain plenty of warmth. This provides enough energy for the formation and further development of tropical cyclones. Statistically, September is their peak period, and it often brings exceptionally powerful storms, such as Hurricane Dorian (2019) or Super Typhoon Mangkhut (2018).

September 4, 2019; NASA

By September 8, 37 named storms had already formed in the Northern Hemisphere, slightly more than in an average year.

In the Atlantic, only six named storms developed, nearly two fewer than usual. The only major hurricane was Erin, which briefly reached Category 5 strength.

Hurricane Erin; Windy.com

In contrast, the Pacific has been busier. Fourteen storms formed in the eastern and central Pacific (about three more than average), eight became hurricanes, and four reached major hurricane status: Erick, Flossie, Kiko, and Iona. In the northwestern Pacific, 17 tropical cyclones developed, eight of which intensified into typhoons, including Danas, Co-May, Krosa, Podul, Kajiki, and Tapah. The North Indian Ocean has remained quiet, with no storms so far.

Northern hemisphere tropical cyclone activity 2025 (through September 8); Colorado State University

Meteorologists also track the Accumulated Cyclone Energy (ACE), which combines a storm’s strength (wind speed) and duration. In the Northern Hemisphere, it has so far reached only about 60% of the average, showing that although the number of storms is high, their overall intensity and lifespan remain below average. Overall, the season has been calmer than expected so far.

Why do devastating hurricanes sometimes form, while in other years the season remains unusually quiet? In the following text, we will look at the main conditions that determine the formation and development of tropical cyclones.

What is a tropical cyclone?

A tropical cyclone is a low-pressure system that develops in tropical regions over warm oceans. It appears as a destructive storm with a closed wind circulation around the center of low pressure.

Unlike mid-latitude lows, tropical cyclones are warm-core systems powered by latent heat released as water vapor condenses. Their warm core extends through the troposphere, giving them a symmetrical, vertically consistent structure that is not linked to fronts.

By contrast, mid-latitude cyclones are cold-core systems that typically form between 35° and 65° latitude. They draw their energy from horizontal temperature contrasts and are closely tied to cold, warm, and occluded fronts, which shape their development and weather patterns.

Extratropical lows and Tropical Cyclone Kiko (bottom right) on the NOAA Pacific Surface Analysis map; NOAA

The ingredients of tropical cyclone formation

Tropical cyclones are complex systems that need a steady supply of heat and moisture. Cyclogenesis begins when a storm becomes self-sustaining, which requires deep convection and enough low-level rotation of air to support further growth.

Formation of a tropical cyclone; Cléa PÉCULIER / Sophie RAMIS / AFP

For the development of deep convection, it is ideal when the sea surface temperature (SST) reaches at least 26.5 °C (about 80 °F) and warm water extends to a depth of at least 45 meters (about 150 feet). Warm ocean waters provide the necessary heat and moisture, which fuel tropical cyclones.

For air heated by the ocean surface to rise and form thunderstorm clouds, the atmosphere must cool sufficiently with increasing altitude. A moist mid-troposphere (at around 5 km) is also crucial, since excessively dry air at these levels can disrupt storm clouds and halt the development of circulation.

Thus, the development of deep convection requires:

• sufficient oceanic thermal energy (SST > 26.5 °C to 45 m depth)

• enhanced relative humidity in the mid-troposphere (700 hPa)

• conditional atmospheric instability

For tropical cyclones to form, sea surface temperatures (SST) must exceed 26.5 °C (80 °F). On the SST map, Tropical Storm Mario and Tropical Depression Blossom are marked; Windy.com

Air rotation (absolute vorticity) is influenced both by the actual swirling motion in the atmosphere (relative vorticity), for example due to the pressure field, and by the deflecting force of Earth’s rotation, the Coriolis force (planetary vorticity).

The Coriolis force is an apparent force that deflects motion to the right in the Northern Hemisphere and to the left in the Southern. Strongest at the poles and zero at the equator, it requires storms to form at least 500 km (300 miles, roughly 5° latitude) away for rotation to develop. As a result, hurricanes rotate counterclockwise in the Northern Hemisphere and clockwise in the Southern.

Wind shear refers to the change in wind direction and speed with height. In general, excessive shear (greater than 10–15 m/s (20-35 mph) between 850 and 200 hPa) can disrupt and weaken a tropical cyclone. In some exceptional cases, however, it may actually support storm development, though this is uncommon.

So, to sustain the rotation of a tropical cyclone, the following are required:

• a position at least 5° latitude away from the equator

• menhanced relative vorticity in the lower troposphere

• generally weak vertical wind shear at the location of formation

Meeting these conditions is necessary for a tropical cyclone to form, but not sufficient. A tropical disturbance must first develop, and its transition to a closed circulation remains under study. Even under seemingly ideal conditions, disturbances may fail to intensify, making this stage one of the greatest challenges in modern meteorology.

Areas of tropical cyclone occurrence and their regional names at ≥119 km/h (74 mph); BBC

Tropical disturbance

A tropical disturbance is an area of lower air pressure accompanied by extensive cloudiness and thunderstorm activity. The first signs of circulation may also appear, but a closed low-pressure center, characteristic of a tropical depression, has not yet formed.

Visualization of a tropical disturbance; UCAR

Tropical disturbances often form where conditions for thunderstorms align with initial atmospheric rotation. They most commonly develop in the monsoon trough, a low-pressure zone linked to seasonal monsoon flow. Unlike the Intertropical Convergence Zone (ITCZ), where trade winds converge with little rotation, the monsoon trough’s cyclonic flow makes vortex development more likely. The two are closely related, as the trough is essentially the monsoon-influenced part of the ITCZ.

Intertropical Convergence Zone and Monsoon Trough: their depiction on synoptic analysis; WMO

A tropical wave (also called an easterly wave) is an atmospheric disturbance that moves westward with the trade winds. In the Atlantic, African tropical waves play a key role: they form over eastern Africa, travel westward across the ocean, and often act as seeds for tropical cyclones.

The positions of monsoon troughs, the ITCZ, and tropical waves are shown on the Tropical Surface Analysis maps produced by NOAA’s National Hurricane Center (NHC).

NOAA Tropical Surface Analysis; NOAA

Development of tropical cyclones

If conditions are favorable, a tropical wave can intensify into a depression, the weakest stage of a tropical cyclone, and then into a tropical storm. With further strengthening and sustained winds above 119 km/h (74 mph), it becomes a hurricane (Atlantic, northeastern/central Pacific), a typhoon (northwestern Pacific), or a cyclone (Indian Ocean and near Australia). Learn more in our article Hurricane, Tropical Storm, Typhoon or Just a Tropical Depression?.

Infographic showing the structure of a tropical cyclone; Encyclopædia Britannica, Inc.

Under unfavorable conditions, cyclones weaken as convection fades, central pressure rises, winds slow, and the system may eventually dissipate.

What drives the movement of tropical cyclones

The movement of tropical cyclones over the oceans is the result of complex interactions with the surrounding atmospheric circulation.

In low latitudes (about 20–25°), trade winds on the equatorial side of the subtropical high steer them westward. This high is a semi-permanent area of high pressure near the Tropics of Cancer and Capricorn. At its western edge, tropical cyclones tend to curve, first poleward and then eastward, under the influence of the westerlies. These winds also govern the movement of extratropical cyclones in the mid-latitudes. Once tropical cyclones reach these latitudes, they gradually transform into extratropical lows.

Ex-tropical cyclones usually increase uncertainty in weather forecasts in the mid-latitudes, sometimes for several days ahead.

Life cycle of a North Atlantic hurricane; Encyclopædia Britannica, Inc.

Landfall

When a tropical cyclone strikes land, it quickly weakens as it loses its energy source, the warm and moist ocean. Dry continental air and friction further accelerate its decay. The rate depends on terrain: in mountainous areas like Haiti or the Philippines, circulation can collapse within hours, while over flatter regions such as Texas or India, systems may last one to three days.

Even after the circulation breaks down, a large amount of moisture remains in the atmosphere, often producing widespread rainfall. This can result in flooding hundreds of kilometers inland, and in mountainous terrain, dangerous landslides and mudflows.

Global distribution of annual person-days exposure to tropical cyclones in 2002–2019; Jing, R., Heft-Neal, S., Chavas, D.R. et al. Global population profile of tropical cyclone exposure from 2002 to 2019. Nature 626, 549–554 (2024)

Large-scale circulation phenomena and their influence on the formation of tropical cyclones

The conditions for tropical cyclone formation, described earlier, are strongly influenced by seasonal and interseasonal circulation phenomena, including trade winds, subtropical highs, monsoons, tropical upper tropospheric troughs (TUTTs), and climate oscillations like El Niño–Southern Oscillation (ENSO) or Madden–Julian Oscillation (MJO).

The strength of the trade winds determines whether tropical waves can organize and how quickly they move across the ocean.

The position and intensity of the subtropical high govern not only the tracks of storms but also the conditions for their formation.

Schematic of trade winds and the subtropical high in the wind layer; Windy.com

Monsoon circulation in the Indian Ocean and surrounding regions plays a crucial role. During the peak of the southwest summer monsoon (June to September), fewer cyclones form, but at its beginning and end, conditions may be more favorable.

The monsoon gyre is a large cyclonic circulation in the western North Pacific, typical of the summer monsoon season. At about 850 hPa (1.5 km / 0.9 mi), it can span thousands of kilometers and last up to two weeks. Convection and rainfall concentrate along its southern to southeastern edge, where tropical storms and typhoons often form. Its development is linked to the breakdown of the monsoon trough and thunderstorm growth, though its direct role in cyclogenesis is unclear. Although rare, often only once a year, monsoon gyres can spawn entire sequences of tropical cyclones.

The Tropical Upper Tropospheric Trough (TUTT) is a low-pressure area at 200–300 hPa (9–12 km / 6-8 mi), usually over the subtropics but extending into the tropics. It is essentially a “trough” of cooler air high in the atmosphere. For tropical cyclones, a TUTT can be either enemy or ally: bringing disruptive wind shear in unfavorable cases, or creating outflow channels that support intensification under favorable conditions.

The Madden–Julian Oscillation (MJO) is a disturbance in the upper atmosphere that travels through the tropics and circles the globe on a timescale of several weeks. In its positive phase, it creates favorable conditions for convection, while in its negative phase, it suppresses it. For developing tropical cyclones, this can either support or hinder their growth.

Warm waters of El Niño appeared in the tropical Pacific in June 2023; NOAA Climate

The El Niño–Southern Oscillation (ENSO) is a climate phenomenon that changes tropical wind circulation and ocean temperatures over the Earth, strongly influencing where and how cyclones form. Its impacts vary by region and by phase: during El Niño, fewer storms occur in the Atlantic, but activity increases in the central and eastern Pacific and shifts eastward in the northwest Pacific. In neutral years, activity aligns with the long-term average, while other phenomena like the Madden–Julian Oscillation (MJO), the monsoon trough, or the TUTT play a larger role in either supporting or suppressing cyclogenesis.

Multidecadal climate variability refers to long-term cycles lasting several decades (typically 20–60 years). These reflect natural variability in the ocean–atmosphere system, most evident in changes in sea surface temperature, circulation, and rainfall. They occur in all oceans: in the Pacific as the Pacific Decadal Oscillation (PDO), in the Atlantic as the Atlantic Multidecadal Oscillation (AMO), and in the Indian Ocean as the Indian Ocean Basin-Wide mode (IOBV).

Hurricane Henriette on August 8, 2025; Windy.com

When and where tropical cyclones form during the year

The conditions for the formation of tropical cyclones change throughout the year, and with them the likelihood and frequency of their development.

In the Atlantic, tropical cyclone formation is closely tied to African tropical waves. These disturbances form over eastern Africa and move westward across the ocean, often seeding hurricanes. The season peaks in September, when the warmest ocean conditions coincide with the weakest wind shear. Conversely, an influx of dry Saharan air, often carrying dust (the Saharan Air Layer), can hinder storm development.

Saharan dust over the Atlantic, June 29, 2018; NOAA

In the northeastern Pacific, activity peaks between June and October. Tropical cyclones mainly form with the Intertropical Convergence Zone (ITCZ) and the monsoon trough, which often lie close to Central America’s coast.

Seasonal cyclone activity: North Atlantic, Central and Eastern North Pacific; WMO

In the northern Indian Ocean, storm development is closely linked to the monsoon. During the peak of the southwest monsoon in July and August, there is a great deal of moisture over the ocean, but also strong vertical wind shear, which usually suppresses cyclogenesis. Tropical cyclones therefore form more often at the beginning of the season (May–June) and at the end (October–November), when the wind shear weakens.

The Northwestern Pacific has the world’s highest number of tropical cyclones, averaging 25–30 named storms yearly. With activity year-round, there is no official typhoon season. The quietest month is February, while the peak from August to October coincides with warm oceans and weak wind shear. Monsoon troughs and gyres foster conditions for multiple typhoons, and a smaller winter peak may occur at lower latitudes when the Madden–Julian Oscillation (MJO) enhances storm development.

Seasonal cyclone activity: North Indian Ocean, Northwestern Pacific; WMO

In the ocean waters around Australia and the southern Indian Ocean, the main cyclone season falls between December and March, coinciding with the summer monsoon in the Southern Hemisphere. They most often form in the monsoon trough, whose position shifts during the season.

Seasonal cyclone activity: around Australia and the Southwest Indian Ocean; WMO

Tracking tropical cyclones with Windy.com

Even before a tropical cyclone forms, Windy.com lets you monitor key parameters such as sea surface temperature, humidity, and air flow at the surface and aloft, helping identify areas where storms may develop.

Once a cyclone forms, real-time satellite imagery and, where available, radar data are provided.

Hurricane Tracker: forecast path of Tropical Storm Mario; Windy.com

To forecast its strength and track, you can use the hurricane tracker, which uniquely compares forecasts from multiple models (ECMWF, GFS, ICON, etc.) and reveals uncertainties. Potential impacts can then be explored across forecast layers, from wind and gusts to rainfall totals. Official warnings from national meteorological services are also available, helping you prepare in advance for dangerous weather.

Windy.com brings all the tools for tracking tropical cyclones into one clear and visually distinctive platform.

In early June 2025, people across many parts of Western and Central Europe enjoyed several days of beautifully colored sunrises and sunsets. This stunning display was linked to smoke from massive wildfires burning more than 6,000 kilometers (3,700 miles) away in central Canada.

Colorful sunset over Bibury, Gloucestershire, England (31 May 2025); Ed Robinson

How is it possible that smoke particles could travel such a vast distance across the entire Atlantic? The key to this long-range transport lies high up in the atmosphere, in a region known as the jet stream. This powerful high-altitude wind can carry fine particles thousands of kilometers across continents and oceans.

Smoke from Canadian wildfires captured in a GOES-19 satellite image (1 June 2025); NOAA

CAMS total aerosol optical depth analysis over the North Atlantic (30 May and 1 June 2025); CAMS

What is the jet stream?

The jet stream is a relatively narrow band of very strong winds, typically blowing from west to east. Its shape is often compared to that of a flattened tube with an approximately horizontal axis, stretching for thousands of kilometers, while its diameter is usually only a few hundred kilometers. It is located in the upper layers of the troposphere, often 1 to 2 kilometers below its upper boundary, known as the tropopause. This corresponds to an altitude of roughly 6 to 13 kilometers (about 4 to 8 miles) above the Earth's surface.

The jet stream can be imagined as a flattened tube with a nearly horizontal axis; NOAA (slightly modified)

According to some definitions, a jet stream is classified as airflow with a speed of at least 30 m/s (approximately 108 km/h or 67 mph). Jet streams commonly reach speeds of around 180 km/h (50 m/s, 112 mph), but in extreme cases can exceed 400 km/h (122 m/s, 273 mph).

The history of jet stream discovery

One of the earliest discoverers of the jet stream is often considered to be the Japanese meteorologist Wasaburo Ooishi, who in the 1920s used weather balloons to study upper-level air currents.

The term jet stream first appeared in 1939, in a scientific paper by German meteorologist Heinrich Seilkopf, who used the term Strahlströmung, meaning “beam flow”.

The intensive use of aircraft during World War II greatly expanded knowledge of upper-level air flow and atmospheric dynamics in the higher layers of the troposphere.

The Institute of Meteorology, Class of 1940–41 (C. G. Rossby: 3rd row, 5th from left), Portrait of meteorologist Carl-Gustaf Rossby; American Meteorological Society/Library of Congress

A significant contribution to understanding the origin and dynamics of the jet stream came from Swedish-American meteorologist Carl-Gustaf Rossby (1898–1957). He was one of the founding figures of the so-called Chicago School of Dynamic Meteorology, a group of scientists who, during the 1940s and 1950s, worked at the University of Chicago and studied the principles of general atmospheric circulation. Their research played a fundamental role in establishing the theoretical and physical foundations on which modern numerical weather prediction was later built.

Why do jet streams form?

Because of two essential ingredients: heating and rotation.

Average solar insolation in September 2013; NASA

The Sun does not heat the Earth evenly. Areas near the equator receive more solar radiation and warm up more than regions near the poles. When warm air masses meet cold ones, the warm air rises into higher layers of the atmosphere, while the colder air moves in to replace it from below. This movement creates air flow, in other words, wind.

Global atmospheric circulation: without Earth’s rotation (left), with rotation (right); Eastern Illinois University via Royal Meteorological Society

If the Earth did not rotate, the rising warm air from equatorial regions would flow directly toward the poles in the upper troposphere. However, because the Earth rotates, the Coriolis force deflects this air flow, to the right in the Northern Hemisphere and to the left in the Southern Hemisphere. As a result of this uneven heating and the planet’s rotation, three distinct circulation cells form in each hemisphere: the Hadley cell, the Ferrel cell, and the polar cell.

Cross section of the Northern Hemisphere showing jet streams and tropopause elevations; Atmospheric Sciences at Northern Vermont University

This three-cell atmospheric circulation system causes air masses with very different temperatures to meet in zones around 30° and 50°–60° latitude in both hemispheres. The greater the temperature difference, the stronger the resulting winds. These sharp horizontal temperature contrasts lead to the formation of intense high-altitude air currents, the so-called jet streams.

Jet stream distribution schematic; Windy.com

The subtropical jet stream occurs around 30° latitude, while the polar jet stream is found between 50° and 60°. The polar jet tends to be stronger than the subtropical jet due to the greater temperature contrast between cold polar air and warmer mid-latitude air, especially during winter, when the polar jet stream reaches its greatest strength.

The four main jet streams

In theory, jet streams encircle the Earth in four continuous bands: two polar and two subtropical. However, their actual shape and behavior result from a complex interplay of many factors, including the distribution of land and oceans and how differently they heat up, the position of pressure systems, seasonal variations in solar radiation reaching the Earth's surface, and more.

Jet Stream at 250 hPa (color) and Mean Sea Level Pressure (isolines), 14 June 2025; ClimateReanalyser.org

Jet streams meander, shifting in both altitude and latitude. At times, they split or merge, form eddies, and can even disappear entirely in one region or suddenly reappear in another.

Why care about something so high above us?

The jet stream has a direct impact on weather at the Earth’s surface in several important ways.

It steers the movement of pressure systems in the mid-latitudes, areas of high and low pressure, and therefore plays a key role in shaping surface weather.

Synoptic weather map, approximate jet stream positions (white arrows); Deutscher Wetterdienst

When the jet stream is strong and relatively straight, weather patterns tend to shift quickly. But when it’s weak or highly meandering, the movement of pressure systems can become blocked, causing a particular weather pattern, such as prolonged rain or an extended dry spell, to persist for days.

Definition of a jet streak; NOAA

A very important part of the jet stream is a jet streak, the area where winds blow the strongest. A jet streak is associated with zones of rising and sinking air. In the Northern Hemisphere, rising motion typically occurs in the right entrance and left exit regions of a jet streak, where upper-level divergence leads to compensating vertical upward motion. In the Southern Hemisphere, upward motion tends to occur in the left entrance and right exit regions.

Schematic of the cross circulation at the jet entrance and exit regions; UCAR/COMET

In the Southern Hemisphere, upward motion tends to occur in the left rear and right front quadrants. These upward air motions associated with jet streaks significantly contribute to the development and intensification of low-pressure systems, and also influence the strength and organization of thunderstorm systems.

Storm Barra – Analysis chart (18 UTC), 7 December 2021; Met Éireann

A well-documented example of how a jet streak and its upward motion zones can trigger explosive cyclogenesis is Storm Barra (December 2021). As the low-pressure system entered the left exit region of a strong jet streak over the North Atlantic on December 6–7, its central pressure dropped rapidly by 55 hPa in 24 hours. Barra reached peak intensity just before landfall and struck Ireland on December 7–8 with widespread damaging winds, gusting up to 135 km/h.

How does the jet stream affect air travel?

Airplanes often fly at the same altitudes where jet streams are typically found. When flying in the same direction as a strong jet stream, they can benefit from the fast-moving air to increase speed and save fuel. This is why flights from west to east are generally faster than those in the opposite direction.

Eastbound and westbound flight tracks over the North Atlantic (8–9 February 2020); Flightradar24.com

Within North America, flight time when traveling east across the continent can be reduced by about 30 minutes if an aircraft is able to ride the jet stream.

According to Flightradar24, a powerful jet stream helped British Airways flight BA112 cross the Atlantic from New York to London in a record time of just 4 hours and 56 minutes on February 9, 2020.

Record-breaking flight BA112 (9 February 2020); Flightradar24.com

Jet stream regions often experience rapid changes in wind speed and direction, both horizontally and vertically, a phenomenon known as wind shear. Wind shear can cause turbulence. When turbulence occurs in clear air, it is difficult to detect and predict. Such turbulence can disrupt flight smoothness, cause sudden altitude drops, and pose safety risks to passengers.

Jet stream pattern on 25 October 2021, highlighting key factors for upper-level turbulence; Aviation Weather Center

Track the Jet Stream

When forecasting the weather, it’s worth monitoring the position of the jet stream, as it can signal not only beautiful sunrises and sunsets caused by the transport of fine particles, but also the potential for dangerous intensification of stormy weather.

Jet streams are best identified on upper-air maps at the 250 hPa level. On Windy.com, simply select the wind forecasting layer and set the altitude to the 250 hPa pressure level. If you also enable the display of geopotential height isolines, you'll see the approximate altitude (in geopotential meters) at which the selected pressure level is located above sea level.

The U.S. National Oceanic and Atmospheric Administration (NOAA) announced the return of La Niña in early October 2025. A weak La Niña is expected to persist through February 2026, after which ENSO (the El Niño–Southern Oscillation) is most likely to return to neutral conditions, with a 55% chance. However, some models suggest a rapid transition to El Niño in boreal spring/austral autumn 2026.

La Niña conditions are present and favored to persist through December 2025–February 2026; NOAA Climate Prediction Center

Statistics show that La Niña generally has a cooling effect on the global mean temperature, and that in La Niña years, winters in western Canada, the northwestern United States (Pacific Northwest), and East Asia (including eastern China, Korea, and Japan) tend to be colder.

Typical winter pattern during La Niña in North America; NOAA Pacific Marine Environmental Laboratory

Whether La Niña actually brings colder weather also depends on many other factors (e.g., the North Atlantic Oscillation, Arctic Oscillation, and Indian Ocean Dipole) that can shape the outcome. Moreover, with ongoing climate change, the odds of truly severe winters are decreasing.

In this article, we examine what ENSO (El Niño/La Niña) is, how it operates, and its impact on seasonal weather patterns and climate in various parts of the world.

What is ENSO?

La Niña is part of the ocean–atmosphere circulation in the tropical Pacific known as the El Niño–Southern Oscillation (ENSO). ENSO is a manifestation of climate variability on time scales of months to years. It is characterized by changes in sea-surface temperature and associated shifts in the Walker circulation over the equatorial Pacific.

Weekly average sea surface temperature anomalies, November 2024–October 2025; NOAA Physical Sciences Laboratory

The Walker circulation is an east–west overturning of the atmosphere along the equator, which can be schematically described as follows: over the warmer part of the ocean, air rises (deep convection, lower pressure); in the upper troposphere it flows eastward (westerly flow); over the cooler part it descends (higher pressure); and near the surface the trade winds (easterlies) return westward, closing the zonal tropical circulation loop.

Generalized Walker Circulation (December-February) during ENSO-neutral conditions; NOAA Climate.gov

When the distribution of warm and cool water changes, the area of most intense convection shifts, the trade winds strengthen or weaken, and the pressure field across the Pacific is altered. This ocean–atmosphere coupling then influences seasonal weather patterns far beyond the Pacific.

The warm phase El Niño and the cool phase La Niña

ENSO fluctuates among three states: the warm El Niño, the cool La Niña, and neutral conditions.

The name “El Niño,” Spanish for “the little boy” (originally referring to the Christ Child), was adopted several centuries ago by fishermen in Peru and Ecuador for the unusually warm waters that reduced their catches just before Christmas.

Schematic of wintertime conditions during El Niño: SST (shaded), surface winds (vectors), and sea-level pressure (H/L); NOAA Physical Sciences Laboratory

The counterpart to El Niño is La Niña, “the little girl”, which denotes a widespread cooling of sea-surface temperatures in the same region along with a reversal of the usual atmospheric conditions.

Schematic of wintertime conditions during La Niña: SST (shaded), surface winds (vectors), and sea-level pressure (H/L); NOAA Physical Sciences Laboratory

How the ocean and atmosphere work together

Let’s begin in the eastern tropical Pacific off the coasts of Peru and Ecuador, a region where fishermen have put to sea for generations. These waters are some of the planet’s most prosperous and most productive, both ecologically and for fisheries, due to the unique interaction between ocean and atmosphere.

ENSO-neutral conditions across the tropical Pacific Ocean; NOAA Climate.gov

Along the equator under typical conditions, the trade winds blow from east to west. These winds push surface water away from the South American coast toward Indonesia. Hence, the mean sea level near Indonesia is approximately 1.5 feet (≈ 46 cm) higher than off the coast of Peru. Near Peru and Ecuador, surface water is displaced from the coast. The deficit is replenished by cold water, supplied from the south by the Humboldt (Peru) Current and from depth by coastal upwelling. Cold, nutrient-rich deep water reaches the surface more readily here because the thermocline (the boundary between warm surface water and the cooler water below) lies on average only about 30 m (≈ 100 ft) beneath the surface. Consequently, sea-surface temperatures near South America are about 8 °C (≈ 14 °F) lower than in the western Pacific.

El Niño conditions across the tropical Pacific Ocean; NOAA Climate.gov

When El Niño begins, the equatorial trade winds weaken; in some areas, they may even briefly reverse. Warm surface water that normally pools in the western Pacific near Indonesia then shifts east, sharply warming the central and eastern tropical Pacific. With weaker trades no longer effectively pushing surface water away from the South American coast, coastal upwelling of cold, nutrient-rich deep water diminishes. At the same time, the thermocline in the eastern Pacific deepens, allowing the cold water to remain farther down and reach the surface less readily.

La Niña conditions across the tropical Pacific Ocean; NOAA Climate.gov

During La Niña, by contrast, the trade winds strengthen and more effectively draw surface water away from the coasts of Peru and Ecuador, as well as along the equator. This drives more vigorous coastal and equatorial upwelling, resulting in the central and eastern equatorial Pacific cooling below normal. At the same time, the thermocline’s depth changes: in the eastern Pacific, it shoals toward the surface, making it easier for cold water to reach the surface, whereas in the western Pacific, it deepens.

How ENSO is monitored

There are several methods for monitoring and evaluating ENSO to determine its current phase. To describe the state of ENSO, so-called indices are used; these condense a highly complex ocean–atmosphere phenomenon into a single numerical value.

Locations of tropical Pacific regions used to monitor sea surface temperature; NOAA Climate.gov

The NOAA uses the Oceanic Niño Index (ONI) as its official indicator of ENSO. ONI is based on sea-surface temperature (SST) anomalies in the Niño 3.4 region of the central–eastern equatorial Pacific. These anomalies are computed relative to a 30-year rolling climatology and averaged into three-month running means. If those values remain at or above +0.5 °C (≈ +0.9 °F) or at or below −0.5 °C (≈ −0.9 °F) for at least five consecutive, overlapping three-month periods, NOAA classifies the episode as El Niño or La Niña.

Oceanic Niño Index (ONI): 3-month running mean of Niño-3.4 SST anomaly; NOAA National Centers for Environmental Information

Some agencies use different thresholds. For example, the Australian Bureau of Meteorology applies a threshold of about ± 0.8 °C (≈ ±1.4 °F) and also requires evidence of an atmospheric response (e.g., trades, pressure). The basic principle is the same: it must be a sustained state of both the ocean and the atmosphere, not just a brief fluctuation.

Locations of the Tahiti and Darwin stations used for the SOI; NOAA Climate.gov

The oldest indicator of ENSO is the Southern Oscillation Index (SOI). It is the standardized difference in sea-level pressure between Tahiti and Darwin, Australia. Fluctuations in pressure between these two locations capture the atmospheric component of ENSO (the Southern Oscillation), which was first described in the early 20th century. During El Niño episodes, the SOI tends to be negative because pressure over Tahiti is lower than usual, while over Darwin it is higher. Conversely, during La Niña, the SOI is typically positive, reflecting the opposite pressure pattern and stronger trade winds. A key advantage of the SOI is the availability of long station records (in some cases extending back to the late 19th century), which allows for the analysis of the relationship between ENSO and climate in a longer-term context.

Southern Oscillation Index (SOI): standardized Tahiti–Darwin SLP difference; NOAA National Centers for Environmental Information

Conversely, one limitation of the SOI is that both Tahiti and Darwin lie south of the equator (Tahiti 18° S, Darwin 12° S), whereas ENSO itself is strongest right along the equator. For this reason, the Equatorial Southern Oscillation Index (EqSOI) is also used today. It is based on area-averaged sea-level pressure over two broad equatorial regions (from 5°S to 5°N): one centered over Indonesia and the other over the eastern equatorial Pacific. This index directly captures the atmospheric component of ENSO in the equatorial zone, but data for it are available only from approximately 1949 onward.

Comparison of ONI and SOI; NOAA Climate.gov

Sea-surface height is also a representative indicator of the ENSO phase and is monitored by satellites. During an El Niño episode, sea level in the eastern Pacific is above average, whereas during a La Niña episode, the enhanced upwelling of cold, deep water cools the surface layer, resulting in a decrease in sea level.

Sea surface height measured by satellite altimeters; NASA

Many other indices are also used to monitor ENSO (e.g., wind indices or outgoing longwave radiation indices). Taken together, they help describe the behavior of the complex, dynamic ocean-atmosphere interaction in the tropical Pacific.

Impacts of ENSO

ENSO is among the most crucial climate drivers on Earth. Although it originates in the tropical Pacific, its influence on large-scale circulation patterns (teleconnections) has far-reaching effects on seasonal weather patterns and climate across many regions of the world.

The distribution of warm surface water along the Pacific equator determines where evaporation and the release of heat and moisture to the atmosphere are most significant. Where the ocean is exceptionally warm, vigorous ascent and deep tropical convection develop. This strengthens the ascending branch of the Walker circulation, which in turn changes the pressure pattern over the Pacific and also influences the path of the jet stream in the upper levels of the atmosphere.

Typical impacts of El Niño (warm phase)

• Warmer-than-normal water shifts into the central and eastern Pacific. This moves the atmosphere’s main “convective heat engine” eastward, along with the position of the Pacific jet stream. The result is that the winter storm track tends to lie farther south, delivering moist, heavy rainfall to the southern United States (California and the Gulf Coast). At the same time, the northern U.S. and Canada are more likely to experience warmer and drier conditions.

• In the tropics, El Niño often brings extreme downpours and flooding to coastal Peru and Ecuador as unusually warm water arrives together with deep convection. By contrast, Australia, Indonesia, and parts of Southeast Asia tend to experience heatwaves and droughts, which can lead to water shortages and increased wildfire risk. Drought risk also increases in southern Africa and parts of Brazil, while southern South America (for example, Uruguay and southern Brazil) and certain regions of East Africa may experience above-normal rainfall.

• El Niño often weakens the Indian summer monsoon (lower average rainfall and a higher risk of drought), but the relationship is not one-to-one; in some years, even with a drier seasonal mean, the likelihood of extreme local downpours in certain regions can increase.

All-India Summer Monsoon Rainfall (1871–2023): drought +10% (dark blue); El Niño/La Niña during monsoon shown by red/blue dots; Climate Research Lab

• Globally, El Niño typically nudges the planet’s average temperature upward, so strong El Niño years rank among the warmest on record (for example, 2023–2024 featured record heat, severe heatwaves, and widespread coral bleaching).

• In the Atlantic, El Niño tends to suppress hurricane activity by increasing vertical wind shear over the tropical basin. The enhanced shear hinders storm intensification and, in years with moderate-to-strong El Niño, can reduce the number of hurricane days in the North Atlantic by around 60%. By contrast, the eastern and central Pacific often experience more active tropical cyclone seasons.

• El Niño is generally associated with a weaker, less coherent polar vortex and a greater likelihood of disruptions.

Typical El Niño impacts on global seasonal patterns (December–February); NOAA

Typical El Niño impacts on global seasonal patterns (June–August); NOAA

Typical impacts of La Niña (cool phase)

• La Niña essentially does the opposite: the trade winds strengthen, cold water rises to the surface in the eastern Pacific, and the warmest ocean waters remain near Indonesia and northern Australia. That concentrates ascending convective regions there, bringing heavier rains to Southeast Asia, Indonesia, and Australia. The flip side is drier conditions in parts of East Africa and along the eastern Pacific near the South American coast. In past strong La Niña years, Australia and Southeast Asia have experienced widespread flooding, while the Horn of Africa (Somalia, Ethiopia, eastern Kenya) has suffered extreme drought.

• In North America, La Niña tends to shift the winter storm track farther north: the U.S. Pacific Northwest and western Canada are usually wetter and cooler, while the southern and southeastern United States tend to have drier and warmer winters. This is the opposite of El Niño.

• La Niña is typically associated with a more active Atlantic hurricane season. The primary mechanism is reduced vertical wind shear over the tropical Atlantic during La Niña, which increases the likelihood that tropical storms organize and intensify into hurricanes. As a result, La Niña years often feature more and stronger Atlantic hurricanes, raising the risk of damage in the United States, the Caribbean, and the Gulf of Mexico.

• La Niña generally has a cooling effect on the global mean temperature. In today’s warming climate, however, even La Niña episodes do not return temperatures to the “cool” levels of past decades; recent observations show that most of the world remains above the long-term average even during La Niña.

• La Niña is often linked to a stronger, more coherent polar vortex; there are exceptions, though. In some winters, a sudden stratospheric warming (SSW) disrupts the vortex, followed by widespread cold-air outbreaks. For example, in 2017/18, the so-called “Beast from the East” brought Arctic air into Europe from the northeast.

Typical La Niña impacts on global seasonal patterns (December–February); NOAA

Typical La Niña impacts on global seasonal patterns (June–August); NOAA

Each ENSO episode is unique; El Niño and La Niña never unfold the same way. Their impacts on seasonal weather depend on several variables, including the intensity of the episode, the spatial pattern of temperature anomalies across the Pacific, the time of year, and interactions with other major climate modes (for example, conditions in the Indian Ocean or circulation patterns over the Atlantic).

ENSO is also unfolding against the backdrop of ongoing climate change, with the global mean temperature rising. Classic teleconnections (for example, “El Niño = a drier Indian monsoon”) may still occur, but record warm oceans and the increased water vapor content of the atmosphere can modify them.

Global surface temperature: Warmest and coldest years per decade marked with circles: red for El Niño, blue for La Niña; NOAA Climate.gov

The role of ENSO in seasonal outlooks

On monthly to seasonal timescales, ENSO is among the most predictable components of the climate system. Changes in sea-surface temperature in the tropical Pacific are translated, with a lag of several months, into distinct patterns of precipitation, pressure, and temperature in various parts of the world. Therefore, ENSO phases are routinely used in seasonal forecasts and serve as a tool for meteorologists and climatologists to estimate the probabilities of departures from normal conditions (e.g., drier/wetter, warmer/cooler conditions, or changes in thunderstorm activity).

U.S. Seasonal temperature and precipitation Outlook (December 2025–February 2026); NOAA Climate Prediction Center

The Climate Prediction Center (CPC), part of the U.S. National Oceanic and Atmospheric Administration (NOAA), regularly issues ENSO analyses and outlooks, tracks indicators such as Pacific SST, and incorporates them into precipitation and temperature forecasts.

The India Meteorological Department (IMD) monitors SST in the Pacific and Indian Oceans as part of its models for predicting the onset and intensity of the summer southwest monsoon.

Australia’s Bureau of Meteorology (BoM) monitors ENSO alongside other indicators, such as the Indian Ocean Dipole (IOD), because these phenomena influence Australian rainfall, monsoons, and temperatures.

ECMWF seasonal forecast (SEAS5): probabilities of 2 m air temperature anomalies for December 2025–February 2026 (51-member ensemble; horizontal resolution ≈ 36 km); ECMWF

Windy is a weather expert

Windy is designed to display weather, particularly current conditions and short- to medium-range forecasts. While you won’t find climate phenomena such as ENSO, which rely on long-term datasets, directly in Windy, you can track their current signals in the atmosphere and ocean: sea surface temperature, trade winds, cloud and convection patterns, and the pressure field over the Pacific.

Information on the state of the climate, specific climate phenomena, and long-term climate outlooks is provided by specialized sources, including the WMO Global Seasonal Climate Update, NOAA’s Climate Prediction Center (CPC), and ECMWF/Copernicus C3S.

As the strength of the Sun’s rays decreases and the polar night beyond the Arctic Circle expands day by day, winter is approaching in the Northern Hemisphere. Astronomically, it does not begin until the moment of the winter solstice, which this year falls on Sunday, 21 December 2025 at 15:03 UTC. Meteorological winter arrives earlier, on 1 December.

Even though autumn still officially reigns in the calendar, the first wintry days with freezing weather and, in places, a generous blanket of snow have already arrived in the Northern Hemisphere during the 2025/2026 season.

Solar Power: forecast solar radiation at Earth’s surface (ECMWF), 23–25 November 2025; Windy.com

The Northern Hemisphere has already experienced two notably cold episodes. The first came in the first half of November 2025, when Arctic air spread into the eastern two-thirds of the United States. The second affected western and central Europe and parts of southern Europe in the second half of November, bringing an early taste of winter weather with below-average temperatures and snow not only in the mountains but also in some places at lower elevations.

Forecast maximum temperatures and temperature anomalies for 11 November 2025; NOAA National Weather Service

According to current forecasts, another outbreak of cold Arctic air is on the horizon around the turn of November and December 2025, this time for much of Canada and probably part of the United States.

Forecast 2 m air temperature anomaly (°C) over successive 24-hour periods from 22 November to 1 December 2025; ECMWF

With winter on the way, terms like Arctic air outbreaks, the polar vortex, or sudden stratospheric warming will appear more and more often in weather forecasts. In this article, we will explain them in simplified terms, show how they are connected, and why they can influence the weather from North America across Europe to the Arctic.

Polar vortex

At its most basic, the polar vortex is a ring of strong westerly winds that, in winter, circulates the cold polar air near the pole. It forms as part of the general circulation of the atmosphere and, in terms of its size, is incomparable with any other vortical phenomenon on our planet.

Schematic of the stratospheric and tropospheric polar vortex over the Northern Hemisphere; NOAA Climate.gov

In the media, the term polar vortex began to appear frequently during the 2013/2014 winter season, in connection with the freezing winter in North America. However, it has been used as a technical meteorological term for a long time; it appeared as early as 1853 in the magazine Littell's Living Age.

Temperature anomalies in winter 2013/2014 show an extreme east–west contrast across the United States. The central and eastern regions were much colder than average due to repeated Arctic air intrusions, while the West, especially California, was significantly warmer than average; NOAA Climate.gov

In meteorological practice, we often distinguish between the stratospheric and tropospheric polar vortex. They are two layers of the same system, but their evolution over time need not be synchronized.

Stratospheric polar vortex

If only the term polar vortex is used, it usually refers specifically to the stratospheric polar vortex, that is, the circulation in the stratosphere, roughly between 100 and 1 hPa (about 15–50 km). You can think of it as a strong, relatively symmetric band of westerly winds encircling the pole, which is clearly visible, for example, on geopotential height maps at 10 hPa.

Animated map of Northern Polar region showing the polar vortex; NOAA Climate.gov

Tropospheric polar vortex

The tropospheric polar vortex, on the other hand, refers to the circulation of cold air and westerly flow in the troposphere, typically in the 500–200 hPa layer (the jet stream region), and also in the near-surface pressure field as a deep polar trough or a broad area of low pressure.

Schematic of the position of the polar night jet and the jet stream within the polar vortex; Met Office

The boundary between cold polar air and the warmer air of the mid-latitudes in the upper troposphere is formed by the polar jet stream. The polar jet stream (hereafter simply the jet stream) is a narrow, elongated band of powerful westerly winds at altitudes of roughly 8–12 km. It extends approximately between 50 ° and 60° latitude. This strong temperature and pressure contrast between cold and warmer air is the main reason why the jet is so fast: wind speeds in its core can exceed 200 km/h.

On a large scale, the jet stream is one of the main factors controlling weather in the mid-latitudes. It strongly influences the movement of low- and high-pressure systems, thereby determining the overall character of the weather. When it is relatively straight and strong, smooth westerly flow predominates; when it becomes strongly meandering, its troughs and ridges can bring deep intrusions of cold polar air towards the equator and, conversely, carry warm air far into the cold polar regions.

Schematic of a meandering jet stream and the positions of low- and high-pressure systems; NOAA Climate.gov

How the Polar Vortex Forms

In winter, the polar regions are under polar night conditions. The Sun does not rise at all for weeks to months. As a result, the air over polar regions cools strongly through radiative cooling. At lower latitudes, the stratosphere at middle and upper levels, roughly 30–10 hPa (about 25–30 km, or 16–19 miles), is significantly warmer. This creates a sharp temperature contrast between the warmer low-latitude air and the colder air over the polar regions.

The atmosphere tries to balance this difference. Warmer air tends to flow from the tropics toward the pole, but Earth’s rotation deflects it to the right in the Northern Hemisphere and to the left in the Southern Hemisphere. The closer the flow gets to the pole, the stronger this effect becomes. The result is a band of strong westerly winds (called the polar night jet) blowing from west to east around the cold air over the polar regions. These westerly winds form the stratospheric polar vortex.

As long as the pole remains much colder than the lower latitudes in the stratosphere, the ring of westerly winds is maintained.

Observed (bold blue) and forecast (red) polar vortex wind speeds at 60°N compared with the 1991–2020 average and the range of natural variability. Around 9 March 2025, a significant disruption of the polar vortex occurred; NOAA Climate.gov

Meteorologists most often monitor the zonal-mean westerly wind component at about 10 hPa and 60°N to determine when the stratospheric polar vortex has formed. Once a persistent westerly wind sets in during autumn and its speed stabilizes at values of at least 10–20 m/s (about 36–72 km/h, 22–45 mph) for several days, the stratospheric polar vortex is considered to have formed. Other parameters are also monitored.

Strength and stability of the polar vortex

When the westerly circulation in the polar regions is strong, the stratospheric polar vortex is strong and relatively stable. The jet stream in the troposphere tends to flow closer to the polar areas and has fewer large meanders. As a result, cold air remains confined over the polar regions (the Arctic during the Northern Hemisphere winter and Antarctica during the Southern Hemisphere winter), and winters in the mid-latitudes are usually milder, without pronounced cold outbreaks.

When the stratospheric polar vortex is strongly weakened or breaks down, the wintertime westerly circulation in the stratosphere is disrupted. The westerly winds weaken, or may even temporarily reverse to easterly. The polar vortex can be displaced away from the pole (a displacement event) or split into two or more parts (a split event). As a consequence, the jet stream in the troposphere often becomes highly meandering, facilitating deep intrusions of cold polar air into the mid-latitudes.

For example, when the polar jet stream dips far south over the United States, cold polar air can reach as far south as Texas and the Gulf Coast. Conversely, where the jet stream flows northward into higher latitudes, warm air is carried into those regions.

When the Arctic polar vortex is strong and stable, it keeps the polar jet stream farther north, allowing the coldest air to remain in the Arctic. When the vortex weakens, shifts, or splits, the jet stream becomes highly wavy, allowing warm air to enter the Arctic and polar air to spill into the mid-latitudes; NOAA Climate.gov

Causes of Polar Vortex Disruption

The westerly circulation around the cold polar air in the stratosphere can be disrupted for several reasons.

This regularly occurs due to the seasonal change in radiative conditions. With the onset of the polar day, ozone in the stratosphere above the pole once again begins to absorb UV radiation from the Sun. The polar stratosphere gradually warms, and the temperature difference between the pole and lower latitudes decreases until, at certain levels, the temperature gradient reverses and the westerly flow gradually changes to easterly (the so-called final warming).

During the polar summer, temperatures in the middle and upper stratosphere (roughly 30–10 hPa, that is, from about 25–30 km/16–19 miles upwards) above the pole gradually increase to the point where they eventually become warmer there than over the tropical latitudes. It is precisely this reversal of the temperature gradient in the middle and upper stratosphere that explains why easterly winds prevail there in summer instead of the westerlies typical of winter.

Rossby waves are large bends in the jet stream that turn an otherwise straight west-to-east flow into a wavy pattern; Wikimedia Commons

During the winter season, strong planetary Rossby waves are among the most common causes of disturbances to the stratospheric polar vortex. On upper-level height maps at 500 hPa or 300 hPa, they appear as large meanders of the jet stream in the form of deep troughs and pronounced ridges. They arise primarily as a consequence of Earth’s rotation and the fact that the effect of the Coriolis force changes with latitude. Their shape and strength are further influenced by mountain ranges, land–sea contrasts, and significant differences in heating between different regions, which distort the flow and can lead to the development of pronounced meanders.

When they are strong enough, they can propagate vertically upward across the tropopause into the stratosphere, where they gradually dissipate and transfer their energy and momentum to the surrounding flow. In doing so, they distort the wind field, slow down the zonal (westerly) winds at high latitudes, and thereby enhance the Brewer–Dobson circulation.

Average temperature profile in the lower atmosphere (left) and schematic of the Brewer–Dobson circulation (right); NOAA Jet Stream, EULIAA

The Brewer–Dobson circulation is the main large-scale circulation of air in the stratosphere. It provides a slow but crucial transport of air and trace gases (such as ozone). In this circulation, air rises from the troposphere into the stratosphere above the tropics and then, at higher altitudes, flows predominantly toward the poles (i.e., meridionally), where it gradually descends. This circulation exists year-round but is most pronounced during winter in the respective hemisphere. The concept of the Brewer–Dobson circulation was derived by Gordon Dobson and Alan Brewer from the distribution of ozone and water vapour in the stratosphere in the mid-20th century; it was only later confirmed by direct measurements.

If Rossby wave activity becomes very strong, the downward motion of air over the polar region intensifies. As the air descends, it warms rapidly adiabatically, and a sudden stratospheric warming can occur.

Sudden stratospheric warming

Sudden stratospheric warming (SSW) is a phenomenon in which the winter stratosphere over the polar region warms by several tens of degrees within just a few days. Most often, SSW is discussed in the context of the Arctic.

At levels around 10–30 hPa (about 20–30 km, roughly 12–19 miles above the surface), the temperature can jump by about 20–40 °C (35–70 °F). At the same time, the westerly zonal winds in the polar vortex weaken. During a so-called major SSW, the ring of westerly flow around the pole at 60°N and 10 hPa even temporarily reverses to easterly. The vortex then often shifts off the pole or breaks up into several parts, typically into two lobes.

Evolution of the stratospheric polar vortex: on 4 March 2025 (left) it is compact and more tightly centred over the pole, while by 10 March (right) it has been displaced off the pole, warm air is entering the Arctic, and easterly winds appear around 60°N, a hallmark of a sudden stratospheric warming; NOAA Climate.gov

At the surface, the effects of an SSW usually appear with a time lag of about 1–3 weeks. As the altered circulation in the stratosphere gradually propagates downward, the jet stream in the troposphere often weakens and becomes more wavy, and blocking high-pressure systems become more frequent in the polar regions. This increases the likelihood of significant outbreaks of cold Arctic air into the mid-latitudes (Europe, North America, Asia) and, conversely, intrusions of warm air into the polar regions. It is not the case, however, that every SSW automatically brings severe cold to a specific area in the mid-latitudes; rather, it increases the overall likelihood of cold polar air outbreaks in the weeks that follow.

Suppose an SSW and the subsequent weakening of the polar winds occur during winter, when solar radiation does not yet reach the polar regions directly (for example, in January or February). In that case, the polar stratosphere will cool radiatively again, become colder than the stratosphere at lower latitudes, and the westerly winds and thus the polar vortex will be re-established.

Average daily temperatures in the polar stratosphere (10-millibar pressure level) of the Northern Hemisphere from late 2022 into early 2023 (dark purple line); NOAA Climate.gov

Suppose the disruption occurs after the polar night ends and sunlight returns to the polar regions (for example, in March). In that case, the polar stratosphere is already beginning to warm, and the temperature difference between lower latitudes and the pole is weakening. In this situation, easterly flow tends to persist, and the polar vortex decays. It does not re-establish itself until the following autumn. Such an SSW is referred to as a final warming.

How to identify the polar vortex on meteorological maps and graphs

To monitor and describe the stratospheric polar vortex, several types of meteorological products are used, which together provide a picture of its strength, shape, and stability.

A basic indicator is the time series of the zonal-mean westerly wind at 60°N at 10 hPa. This graph shows how the average strength of the westerly flow in the mid-stratosphere changes over the course of winter: a strong, cold, and compact vortex corresponds to large positive values, while a weakened or disturbed vortex appears as a pronounced drop in wind speed or even a reversal of the wind direction.

Mean zonal wind at 10 hPa - Sub-seasonal range forecast; ECMWF

Equally important are maps of geopotential height and stratospheric flow, most often at the 10 hPa level. They clearly show whether the polar vortex is nearly circular and centred over the pole, or instead deformed, displaced toward lower latitudes, or split into several smaller vortices. Temperature and temperature anomaly maps then show how cold or warm the air inside the vortex is compared to climatology; a cold, low, “compressed” stratosphere indicates a strong vortex, whereas warming and positive anomalies over the pole signal that it is weakening.

For a comprehensive description of the state of the polar vortex, various indices (such as the Northern Annular Mode, NAM) and also “polar-cap” averages are often used. These are averages of quantities (for example, temperature anomalies or geopotential height) over the entire 60–90°N latitude band. These indices and polar-cap averages condense the complex spatial fields of pressure and temperature into a simple numerical time series, from which it is immediately apparent whether the polar vortex is powerful, near normal, or, conversely, significantly weakened.

How temperature affects the height of pressure; NOAA Jet Stream

Vertical cross-sections of the atmosphere are also very useful, often showing anomalies of geopotential height, either as latitude–height plots at a given time or as time–height plots over the polar cap (e.g., 60–90°N). This makes it possible to see how deeply into the troposphere the influence of a strong or weakened polar vortex extends, how its vertical structure changes throughout the winter, and whether a sudden stratospheric warming occurs.

Geopotential height anomalies over the polar cap during the 2024–25 polar vortex season. Positive atmospheric thickness anomalies over the Arctic are gradually propagating downward from the stratosphere into the troposphere; NOAA Climate.gov

When an SSW occurs over the polar region (typically shown on a vertical cross-section as an average over the polar cap 60–90°N), the air column in the middle stratosphere warms significantly, “puffs up,” and the 10–30 hPa pressure levels are shifted to higher altitudes. On vertical cross-sections of geopotential height anomalies, this appears as a column of positive anomalies over the pole, initially strongest in the upper stratosphere. Over time, this signal propagates downward into the lower stratosphere and the troposphere, eventually reaching the surface, where blocking anticyclones often form, allowing frigid air to spill into the mid-latitudes around them.

Daily geopotential height anomalies (17 pressure levels, last 120 days), normalized relative to the 1979–2000 daily climatology and averaged over the polar cap north of 65°N; NOAA Climate Prediction Center

Polar Vortex on Windy.com

You can easily view outbreaks of cold polar air in Windy in the Temperature or Wind (surface) layers. These views show where the cold air is coming from, how it spreads into the mid-latitudes, and allow you to switch on additional layers such as Clouds, Rain, thunder, or New snow. This gives you a much more complete picture of how the cold-air outbreak will affect the weather in your area.

If you are interested in the polar vortex itself, switch to higher levels of the atmosphere and display, in the same layers, for example, 150 or 10 hPa. In the standard atmosphere, the 10 hPa level corresponds to an altitude of roughly 30 km (19 mi) above sea level, i.e., the boundary between the middle and upper stratosphere. For more precise information about the altitude at each pressure level, select Geopotential height in the isoline settings.

Temperature (shading) and geopotential height (contours) at 10 hPa on 24 November 2025 and 8 December 2025. During this period, the displaced and weakened polar vortex undergoes a slight strengthening; Windy.com

Vertical cross-sections of the atmosphere (especially in the troposphere) are also a handy tool for visualizing temperature stratification and airflow. You can find them in the Airgram mode. In Windy, you can display them in two ways. The first option is to view the evolution of temperature and wind above a single point in the detailed forecast for a given location. The second option is a cross-section along a chosen route, which you can find in the Distance & planning tool. Draw a track, then select Airgram mode again.

On Windy.com, you can monitor the polar vortex and how it is forecast to change over time, from the Earth’s surface all the way up to the stratosphere – all in one place and with just a few clicks.

Vertical cross-section of the troposphere along the Distance & Planning route (Airgram mode); Windy.com

Summer in the Northern Hemisphere is now well underway. June 2025, the season’s first month, ranked as the third warmest June ever recorded, according to NOAA. Global surface temperature records date back to 1850, meaning this result is based on 176 years of measurements.

June 2025 Selected Climate Anomalies and Events Map; NOAA

The average global temperature in June was 0.98 °C (1.76 °F) above the 20th-century average, and 0.46 °C above the 1991–2020 baseline. Only June 2023 (+0.51 °C vs 1991–2020) and the record-breaking June 2024 (+0.67 °C vs 1991–2020) were warmer globally.

From a regional perspective, Europe and Asia experienced their fifth warmest June on record, North America the eighth, and Africa the ninth.

Surface air temperature anomaly for June 2025 relative to the June average for the period 1991-2020; Copernicus

In June 2025, extreme temperatures and heatwaves affected many regions across the Northern Hemisphere. Several episodes brought intense heat to large parts of North America (including Mexico and the southern and eastern United States), southwestern and western Europe, the Middle East (with a potential Asian temperature record of 54.3 °C currently under verification in Kuwait), as well as parts of South Asia (Pakistan) and East Asia (including Siberia, China, and Japan).

Definition of a Heatwave

There is no globally standardized definition of a heatwave, and the criteria for declaring one vary depending on the local climate, geography, and social conditions of each region or country. The World Meteorological Organization (WMO) defines a heatwave as “a period of unusually hot weather, statistically unusual, lasting several days and nights.”

Different meteorological services apply their own specific thresholds. The duration of extreme temperatures required to declare a heatwave is often set at 3 days (e.g., in the United Kingdom, Australia).

Impacts of Extreme Heat and Heatwaves

Extreme heat sets off a chain of complex, interconnected processes in both nature and society, ranging from wildfires and droughts to storms, infrastructure failures, and shortages of water and food.

Interrelated and compounding impacts resulting from extreme heat; Climate Central, 2024 (supplemented with text)

Extreme heat during heatwaves places significant thermal stress on the human body, not just due to air temperature, but also because of humidity, solar radiation, wind, surrounding environment, and physical activity.

When the body cannot release excess heat, the risk of heat exhaustion or heatstroke increases. High thermal stress strains the heart and kidneys and worsens chronic illnesses such as cardiovascular, respiratory, mental, or diabetic conditions.

Number of days in 2024 during which at least ‘strong heat stress’ was experienced, based on the daily maximum feels-like temperature exceeding 32°C; Copernicus, 2025

Globally, extreme heat causes more deaths than any other type of extreme weather. Between 2000 and 2019, around 489000 people died each year due to high temperatures, with most deaths occurring in Asia (45%) and Europe (36%).

However, these numbers are only estimates, as heat-related deaths are often underreported. Moreover, heatwaves can also raise mortality indirectly through drought, poor water quality, wildfires, or infrastructure failures.

Examining heat-related deaths during the 1995 Chicago heat wave; EPA

The July 1995 Chicago heatwave caused 465 confirmed heat-related deaths in Cook County, though studies suggest heat may have contributed to hundreds more.

Impacts of heatwaves in cities

Heatwaves and extreme temperatures have particularly strong impacts in urban environments. Cities are affected by the so-called urban heat island effect, where temperatures are significantly higher than in the surrounding rural areas. Dense development, asphalt surfaces, lack of greenery, and high population density all contribute to greater heat accumulation during the day and reduced cooling at night.

A heat map of Vienna shows temperatures several degrees higher in the city center than in parks and surrounding areas; meteoblue

Urban areas can be 5 to 10 °C (9 to 18 °F) warmer than the surrounding countryside, intensifying heatwaves and increasing related risks. In 2023, the temperature difference between the surface of London and the surrounding rural landscape reached up to 7 °C (12,6 °F). The greatest differences occurred during the summer months, when the heat accumulated during the day could not dissipate sufficiently over the short night.

Surface Urban Heat Island Index for maximum (red) and minimum (orange) temperatures for London in 2023; Urban Climate Explorer via Copernicus, 2023

An example of how extreme temperatures can affect the functioning of society in an urban environment can be illustrated by the following scenario:

High temperatures increase thermal stress, which can lead to exhaustion, heatstroke, or worsening of chronic illnesses, especially among vulnerable groups such as children, the elderly, pregnant women, or people working outdoors. Heat also reduces work productivity and can result in economic losses.

Rising demand for cooling increases the load on the power grid, which can become overloaded and lead to blackouts, precisely at the time when electricity is needed most.

Intense sunlight and traffic emissions worsen air quality and increase ground-level ozone concentrations, which cause or exacerbate respiratory illnesses.

Heatwaves are often accompanied by drought and high evaporation, which reduce water availability for personal use, industry, and public services. The risk of wildfires also increases. Extreme temperatures can also damage infrastructure, such as deforming roads, rail tracks, or building structures.

These phenomena interact with one another and can lead to disruptions in urban services, significant economic losses, and increased illness or even mortality among the population.

Diagram of protective stratospheric ozone and harmful tropospheric ozone; Climate Central, 2019

The Dynamic Evolution of Cities

Cities and settlements are constantly evolving. Their populations grow or decline, economic activities expand or contract, and political priorities shift. The risks that cities and their inhabitants face, now and in the future, are shaped both by changes within the urban environment itself and by climate change.

Growth rate of urban agglomerations between 1990 and 2018; United Nations, Department of Economic and Social Affairs, Population Division (2018)

According to the United Nations, 55% of the world’s population lived in cities in 2018 (with urbanization rates of 82% in North America, 65% in China, and only 43% in Africa). Between 2015 and 2020, the global urban population grew by more than 397 million people, with over 90% of this growth occurring in less developed regions. By 2050, the share of the urban population is expected to rise to 68%. This increase must also be viewed in the context of overall global population growth.

As the number of urban residents increases, more people will be exposed to the urban heat island effect. At the same time, the effect itself is expected to intensify due to climate change, according to the Intergovernmental Panel on Climate Change (IPCC).

Global temperature increase

The year 2024 was the warmest year on record. The average global temperature reached 15.10 °C (59.18 °F), which is 0.72 °C (1.30 °F) above the 1991–2020 average and 0.12 °C (0.22 °F) higher than the previous record set in 2023. Compared to the pre-industrial period (1850–1900), 2024 was 1.60 °C (2.88 °F) warmer. This marks the first time the symbolic threshold of 1.5 °C (2.7 °F) of global warming has been exceeded.

Surface air temperature anomalies in 2024, relative to the average for the 1991–2020 reference period; Copernicus, 2025

The WMO warns that global warming is currently occurring at an exceptionally rapid pace. Every year from 2015 to 2024 ranked among the ten warmest years on record, and global temperatures are expected to continue rising.

Global surface air temperature (ºC) increase above the average for the 1850–1900 designated pre-industrial reference period; Copernicus, 2025

As global temperatures rise, the frequency, intensity, and duration of extreme events such as heatwaves, heavy rainfall, droughts, and tropical cyclones are also increasing.

Observed and attributed changes in regional hot extremes; IPCC, 2023

According to the WMO, there is an 86% probability that at least one of the next five years will be more than 1.5 °C (2,7 °F) warmer than the average for the period 1850–1900. There is also an 80% probability that one of these years will surpass 2024 as the warmest year on record.

The Sixth Assessment Report of the IPCC uses five key scenarios to estimate future changes in global temperature and related climate phenomena, such as changes in precipitation, extreme weather events, or sea level rise. These scenarios are based on different assumptions about the future development of greenhouse gas emissions.

Global surface temperature changes relative to 1850–1900 under various greenhouse gas emissions scenarios (left), and the relationship between cumulative CO₂ emissions and associated global temperature increase (right); IPCC, 2023

The IPCC scenarios consider different developments in CO₂ emissions, ranging from a rapid decline and the achievement of net-zero (or even negative) emissions around mid-century in the “very low” scenario, through gradual reductions in the “low” scenario and stabilization in the “intermediate” scenario, to continued growth in the “high” scenario and the fastest increase in the “very high” scenario, where emissions remain high even at the end of the century.

Observations and climate models indicate that global warming occurs with varying intensity across the globe. Land areas are warming more than oceans, and mid- to high-latitude regions more than the tropics.

Impact of global temperature rise on regional temperature distribution; IPCC, 2023

Global Climate Shifts of Cities

An analysis by Climate Central aims to determine how dramatically the climate of major cities could change if the current pace of greenhouse gas emissions is not slowed (corresponding to the IPCC “very high emissions” scenario). For selected cities, average daily summer maximum temperatures at the end of the century were calculated and compared to cities that currently experience similar temperatures. Future warming could effectively "shift" the climate of some cities to resemble that of entirely different regions of the world.

Expected climate shift in selected cities by 2100 without mitigation efforts; Climate Central, 2025

By the year 2100, London’s summer climate could resemble that of present-day Milan, while Milan may shift toward the climate of Port Said in Egypt. Baghdad and other cities in the Middle East are projected to face such extreme warming that their future climate will have no current equivalent anywhere in the world.

Expected climate shift in Boston by 2100 without mitigation efforts; Climate Central, 2025

In the United States, the summer future of 247 major cities was analyzed. By 2060, warming of 2 °C (3.6 °F) is expected, and by 2100, an increase of 4.4 °C (7.9 °F). On average, summer conditions will shift approximately 700 kilometers (435 miles) southward.

For 16 U.S. cities, however, no comparable climate will exist within the continent, as their future conditions will resemble those of present-day Pakistan, North Africa, or the Arabian Peninsula. The greatest warming is expected in the city of Mitchell, South Dakota, where by 2100, summer could resemble today’s Wichita Falls, Texas (+6.2 °C / 11.1 °F).

The five largest U.S. cities, which are home to a combined population of over 19 million people, are projected to experience the following changes by the end of the century:

• New York: warming of 4.2 °C (7.6 °F), with a climate resembling that of Columbia, South Carolina
• Los Angeles: +3.2 °C (5.8 °F), similar to Túxpam, Mexico
• Chicago: +5.1 °C (9.1 °F), similar to Montgomery, Alabama
• Houston: +3.6 °C (6.4 °F), similar to Lahore, Pakistan
• Phoenix: +4.0 °C (7.2 °F), similar to Al Mubarraz, Saudi Arabia

The future of cities

Heatwaves are among the most dangerous natural meteorological phenomena. According to the IPCC, the number of people exposed to these extremes is expected to increase as global temperatures rise. If greenhouse gas emissions are not reduced and no adaptation measures are taken, the impacts of rising temperatures and more frequent heatwaves will continue to worsen.

Global surface temperature: Observed (1900–2020) and projected (2021–2100); IPCC, 2023

Rapid urbanization, a lack of climate-sensitive planning, and the persistent urban heat island effect increase the vulnerability of urban environments and critical infrastructure.

Future climate change and its impact on human health; IPCC, 2023

As climate change continues to intensify, the urgency of adaptation measures is growing as well. Mitigation strategies and adaptation efforts are essential for protecting society as a whole.

Thoughtful urban planning and the implementation of so-called Heat Action Plans (HAPs) can play an important role in protecting cities from extreme heat. These plans combine short-term responses, seasonal preparedness, and long-term strategies aimed at safeguarding the population, infrastructure, and public health. The first city to introduce HAPs was Ahmedabad in India. While such plans are already in place in many cities across North America, Europe, Australia, and India, they are still lacking in large parts of Africa, the Middle East, South America, and small island states, even though these regions are among those most in need due to rapid urban growth.

The urban adaptation gap. The urban adaptation gap is revealed when levels of achieved adaptation fall short of delivering ‘no risk’; IPCC, 2023

Warning systems also play a key role by providing timely information to the public about approaching heatwaves and enabling the implementation of preventive measures. However, according to UN data, only 54% of national meteorological and hydrological services issue warnings for extreme heat, significant heat stress, or heatwaves.

As part of efforts to raise public awareness about the risks of extreme heat and to encourage institutions to adopt adaptation measures, proposals have recently emerged to name heatwaves similarly to hurricanes. Some cities and organizations are already considering or piloting this approach. The first officially named heatwave in the world was Heat Wave Zoe, which struck Seville, Spain, in July 2022.

How to track extreme temperatures and heatwaves using Windy.com

Windy.com offers a wide range of tools for tracking extreme heat, including official warnings issued by national meteorological services, custom temperature alerts, and specialized forecast layers. In addition to the standard Temperature layer, users can make use of the Wet-bulb temperature and Extreme forecast layers (under the Temperature section), which help identify high-risk situations in advance.

Forecast layer wet-bulb temperature on Windy.com; Windy.com

Wet bulb temperature (WBT) represents the lowest temperature to which air can be cooled through the evaporation of water. It depends on temperature, humidity, and wind. In high humidity, the body loses its ability to cool itself through sweating, which raises the WBT. If the WBT exceeds 35 °C (95 °F), the human body can no longer cool itself, which can lead to thermoregulatory failure and death. This threshold is considered the upper limit of human survivability.

The Extreme forecast (Temperature) layer compares ensemble temperature forecasts from the ECMWF model for a given location with the model’s long-term climatology. It shows how much the expected temperatures deviate from normal values.

City Heat Maps by meteoblue are also available on Windy.com; meteoblue

A unique product is also the City Heat Maps, developed by meteoblue. It displays the distribution of temperatures in urban environments and allows users to observe how heat accumulates and spreads across different parts of a city.

Weather radar is one of the most powerful tools in modern meteorology. It allows us to look inside precipitation systems and measure their intensity, structure, movement, and development, details that satellites and ground stations alone cannot provide. This article explains how weather radar works, what it measures, what radar reflectivity is, and how to interpret the data.

Typhoon Co-May (July 2025) brought heavy rain and strong winds to the Philippines, the Ryukyu Islands, and eastern China. It developed into a long-lasting typhoon with winds up to 130 km/h; Windy.com

Where did RADAR come from?

RADAR is an acronym for RAdio Detection And Ranging, meaning “radio detection and distance measurement.” The word “radio” refers to the radio waves transmitted and received by radar.

Just like many great inventions, radar was born out of chance discoveries. In 1922, the steamer Dorchester on the Potomac River disrupted a Navy radio experiment, hinting that radio waves could be used to detect ships under cover of darkness. The idea was forgotten, but in 1934 a passing airplane once again disturbed a radio signal, this time sparking the breakthrough concept of using short energy pulses to locate targets (R. E. Rinehart, Radar for Meteorologists, 2004).

Precipitation clouds on a radar display unit (left); Routine plot of hourly precipitation intensity (center), and radar reflectivity (right), 1986; CHMI via Root

The first principles of radar technology began to take shape as early as the 1930s. Initially, development was focused mainly on military applications, particularly the detection of aircraft and ships. During World War II, however, operators noticed that radar screens often also displayed bands of precipitation, which made it more difficult to track intended targets. This “side effect” later became the impetus for the postwar scientific use of radar in meteorology.

Radar reflectivity (left) and radar velocity (right) around time of an EF1 tornado west of Marion, Wisconsin; NSSL

In the 1950s and 1960s, the first radars designed specifically for weather monitoring were developed, capable of displaying the location and intensity of precipitation. In the 1970s, Doppler radars came into use, able to measure the velocity of moving precipitation particles. Another major advance came in the 1990s with the introduction of dual polarization, which made it possible to distinguish the shape and type of detected particles (rain × snow × hail).

Base reflectivity showing light precipitation over North Texas on February 14, 2021 (left) and the corresponding Differential reflectivity from dual-polarization at the same time (right); NOAA

Today, weather radars are an integral part of both national and international networks. They make it possible to monitor the weather with high spatial and temporal resolution and to provide timely warnings of hazardous phenomena, from heavy rainfall and hail to tornadoes.

How radar measures

The basic operating principle of a weather radar can be illustrated using a pulsed Doppler radar. This type of radar is now widely used and employed by national meteorological services around the world.

Beneath the radar’s circular dish (left) and the Laverton radar near Melbourne (right); BOM

A radar consists of several main components: a transmitter that generates a high-frequency signal; an antenna that transmits this signal into space and simultaneously receives its echoes; a receiver that amplifies the returned signal; and a display system that makes it possible to clearly interpret the results.

The radar transmits a pulse and listens for a returned signal from a target; NOAA

The first weather radars transmitted a continuous signal and therefore required two separate antennas: one for transmitting and one for receiving. Modern radars, however, work differently. They use a single antenna and emit short pulses of energy instead of a continuous beam. After each pulse, the radar waits for the echo to return before sending the next. Because radio waves travel at the speed of light, the echoes return within fractions of a second, even from targets hundreds of kilometers away. In each hour, the radar spends just over 7 seconds transmitting pulses. The remaining 59 minutes and 53 seconds are devoted to listening for the echoes.

Most radars used in operational meteorology operate in a volume-scanning mode, in which both the azimuth and the elevation angle of the antenna system can be varied. The antenna repeats circular scans at 10 to 20 different elevation levels, usually ranging from 0.3° to 20°. The entire process takes about 4 to 6 minutes and provides a three-dimensional picture of the atmosphere.

The WSR-88D scans a predetermined set of elevation angles to complete a volume scan. During active weather, Volume Coverage Pattern (VCP 12 or 212) is used. This VCP includes 14 elevation angles and takes about 4.5 minutes to complete; NOAA

Once the radar receives the reflected signal, the receiver must first amplify it. To illustrate: the WSR-88D radar used in the U.S. NEXRAD network transmits a pulse with a power of 750,000 watts, but only a tiny fraction of that returns. Only then does the display system take over, processing the data further and converting it into visual form.

How a radar beam propagates through the atmosphere

For simplicity, we can imagine a radar beam as a cone of light from a flashlight. The strongest signal is in the center, while it weakens and gradually spreads toward the edges. Close to the radar, even fine details can be seen, but at greater distances they fade, and the radar’s resolution decreases.

Depiction of the WSR-88D radar beam; NOAA

As the distance from the radar increases, its beam gradually rises above the surface due to the curvature of the Earth. At greater distances, the radar can no longer capture conditions close to the ground, which are the most important from a weather perspective.

Diagram of radar beam propagation through the atmosphere under different meteorological conditions; NOAA (slightly modified)

In a vacuum, a radar beam would propagate in a straight line, but in the atmosphere it bends slightly due to changes in air density and refractive index. Under normal conditions, it curves gently toward the Earth. However, if the vertical profile of temperature and humidity deviates significantly from the norm (for example during a temperature inversion), anomalous propagation may occur, causing the beam to bend more strongly toward the Earth, or away from it.

What does a weather radar see?

A weather radar measures the power of the electromagnetic signal reflected from particles in the atmosphere. This power is influenced both by the technical parameters of the radar (e.g., transmitted power, wavelength, pulse length, antenna gain, or beamwidth) and by the reflectivity of the target, and it decreases rapidly with distance. This relationship is described by the so-called radar equation.

Weather radars are tuned to best capture the backscatter from precipitation particles, such as raindrops (from drizzle to large drops in heavy showers), snowflakes, or hail.

Electromagnetic spectrum with the radio frequency range and the S, C, and X bands highlighted; NASA (with radar bands added)

S-band radars operate at wavelengths around 10 cm. These waves penetrate even heavy precipitation well, which is why they are often used in regions with severe thunderstorms, tropical downpours, and hurricanes.

C-band radars operate at wavelengths around 5 cm and are suitable for areas where heavy rain is less frequent, so signal attenuation is not as significant.

X-band radars operate at wavelengths around 3 cm. They can detect smaller cloud particles and light precipitation, but their signal is quickly weakened in heavy rain.

In addition to precipitation particles, and sometimes clouds, radars also register so-called non-meteorological echoes, such as those from birds, insects, aircraft, ships, or the Earth’s surface.

How far from the radar is it raining?

The distance from which the reflected signal returned can be determined easily: measure the time between transmission and reception, multiply it by the speed of light, and divide by two, because the pulse traveled the path twice, out and back.

Radar emits pulses of electromagnetic energy, part of which is scattered back toward the radar; UCAR MetEd COMET

The waiting time before the radar sends the next pulse determines its maximum range, i.e., the distance within which the reflected signal can be uniquely assigned to the transmitted pulse. A longer delay between pulses allows the radar to detect more distant targets, but it also reduces temporal and spatial resolution. In addition, the beam weakens as it travels through the atmosphere and rises higher above the Earth’s surface with distance. Setting the delay is therefore always a compromise between range and measurement quality.

Radar reflectivity

Radar reflectivity, derived from received power, is one of the primary weather-radar products and is likely the most familiar to the public. It represents the intensity of the backscattered signal. Reflectivity depends mainly on the size of precipitation particles (it is strongly weighted toward larger drops) and, to a lesser extent, on their concentration.

The base unit of reflectivity is mm⁶/m³, as given by the radar equation. Raw values span many orders of magnitude: tiny water droplets are about 0.001 mm⁶/m³, while hail in severe storms may reach 36 million mm⁶/m³. For practicality, reflectivity is expressed on a logarithmic scale as decibels of reflectivity (dBZ). Typical measured values of radar reflectivity range from –35 dBZ to +85 dBZ; in general, higher dBZ corresponds to heavier precipitation.

General reflectivity guidelines; NOAA (color scale by Windy.com)

Base reflectivity

Base reflectivity shows the intensity of rain or snow as seen by the radar at a low angle above the horizon (around 0.5°). It gives a clear picture of where precipitation is falling and how strong it is.

This product is especially useful for spotting details inside storms, like hook echoes, gust fronts, or new cells forming near the ground. Its limitation is that it only shows the lowest slice of the atmosphere, so the full vertical structure of storms can be missed. Close to the radar, the image may also be distorted by echoes from terrain or buildings.

Base reflectivity shows a hook echo near Oklahoma City on April 19, 2023 (left); Velocity data confirm tornado rotation (right); NOAA

Composite reflectivity

Composite reflectivity is one of the most commonly used radar products. It displays the highest reflectivity value above a given location. It is useful for a quick overview of the strength and distribution of convection and for detecting the first signs of thunderstorm development. It also allows monitoring of larger-scale systems, such as squall lines or multicell storms.

Base Reflectivity shows weak scattered storms over Dallas-Fort Worth, while Composite Reflectivity reveals their stronger structure; NOAA

For detailed storm analysis, however, it is less suitable, since maximum reflectivity values cannot be assigned to a specific altitude and finer storm features (e.g., hook echoes) are often obscured.

Instantaneous precipitation intensity

Radar reflectivity can be converted into instantaneous precipitation intensity (mm/h, in/h) using empirically derived relationships between drop size distribution and rainfall rate. These conversion formulas are known as Z–I relationships.

Why radar images can be misleading

Radar images are an invaluable tool for monitoring precipitation and storms. Their interpretation, however, is not straightforward, since the captured signal can be influenced by a variety of interfering factors. Some of these effects can be partially filtered out. The most common issues include:

Ground and object echoes
In addition to precipitation, radar can also capture unwanted echoes from ground targets such as buildings, hills, or trees. These may occur when anomalous propagation of the radar beam happens due to very stable atmospheric layering. Such “ground clutter” echoes can be partially filtered out. Radar can also detect echoes from other non-meteorological targets.

Range folding
Sometimes so-called “second-trip echoes,” or “range folding,” appear in the image. These are reflections that return only after the next pulse and are therefore displayed in the wrong location.

Possible range folding: false echoes that are typically narrow and elongated along the radar beam axis; Windy.com

Attenuation of the radar beam
As a radar beam travels through the atmosphere, its intensity decreases due to absorption and scattering by gases, water vapor, clouds, and precipitation. Passing through heavy rain or hail, part of the signal’s energy is lost, which can cause precipitation behind the storm core to be underestimated.

Interference from external signals
Radar can also be affected by other sources operating at similar frequencies, such as telecommunication transmitters or even the Sun. In such cases, strange lines or streaks may appear on the radar images.

External interference; Windy.com

Conversion of reflectivity to rainfall
Errors in estimating rainfall intensity arise already from the use of the Z–I relationship itself. It is based on an assumed drop size distribution of precipitation particles. In reality, however, these distributions are highly variable and cloud-specific, which makes it impossible to accurately convert radar reflectivity into rainfall intensity.

Presence of hail
In radar measurements, it is assumed that the radar detects only particles much smaller than the signal’s wavelength, so the echoes follow Rayleigh scattering. Hailstones, however, are much larger and scatter the signal in a different way. Another assumption is that all particles have the same composition, but in reality reflectivity depends on what they are made of and whether they are liquid or frozen. Because of this, hail often appears on radar with unusually high reflectivity values.

Small particles follow Rayleigh scattering, large particles follow Mie scattering; UCAR MetEd COMET

Enhanced reflectivity at the freezing level (Bright band):
In the melting layer of solid precipitation below the freezing level, enhanced radar reflectivity is often observed, typically increasing by 5 to 15 dBZ. This effect results from changes in the composition of precipitation particles. Enhanced reflectivity is mainly associated with stratiform precipitation. By contrast, strong updrafts within convective clouds tend to disrupt the horizontal stratification that is essential for the formation of this band.

Concentric rings around the radar in stratiform rain are caused by enhanced reflectivity at the freezing level (the bright band) and by the geometry of the radar scan; WIndy.com

Processes below the radar beam:
Evaporation of precipitation below the beam (virga) - If the air near the surface is dry, raindrops or snowflakes may evaporate (or sublimate) before reaching the ground. In such cases, the radar shows precipitation that never actually reaches the surface.
Increase in precipitation intensity - On the windward slopes of hills, rainfall may intensify due to orographic effects, causing the radar to underestimate the actual amounts.

Virga; WMO

Radar on Windy.com

Windy.com now provides radar data from approximately 1,000 weather radars worldwide, and the number continues to grow. To keep the data as accurate as possible, removable measurement errors are automatically filtered out.

In the Radar layer, you’ll find the Composite reflectivity product, which you can display either in dBZ or as the current precipitation intensity (mm/h or in/hr).

Color scale of reflectivity and converted precipitation intensity (the scale can be adjusted as needed); Windy.com

In the Radar+ layer, composite radar reflectivity data are combined with satellite observations, providing a clear, detailed view of the formation and evolution of clouds and precipitation and enabling more reliable identification of false echoes by comparing the radar reflectivity field with satellite imagery.

Typhoon Kajiki made landfall along Vietnam’s coast near Vinh on Monday, August 25, 2025.; Windy.com

From the radar echoes, Windy also generates a short-term alerts of precipitation movement for the next 60 minutes using interpolation, so you can see not only where it’s raining or snowing right now, but also what’s on the way.

And so a downpour never catches you off guard, you can now enable rain alerts. Windy will notify you in advance when heavy rain is approaching.

All of these terms refer to different forms of tropical cyclones—rotating, organized systems of clouds and thunderstorms that occur in tropical and subtropical regions.

A tropical cyclone is a low-pressure system that forms over warm ocean waters, most commonly between 5° and 25° latitude. In the Northern Hemisphere, it rotates counterclockwise; in the Southern Hemisphere, it spins clockwise.

Global tropical cyclone tracks from 1985 to 2005; Wikimedia Commons

What Makes a Tropical Cyclone Unique?

Unlike mid-latitude low-pressure systems (also known as extratropical cyclones), which typically occur between 35° and 65° latitude, a tropical cyclone is symmetrical, has a warm core, lacks frontal boundaries, is generally smaller in size, and may exhibit significantly lower central pressure. Because of the steep pressure gradient over a relatively short distance, tropical cyclones can produce extremely strong and destructive winds.

The largest and smallest recorded tropical cyclones, shown relative to the size of the United States; NOAA

To put it into perspective: the lowest central pressure ever recorded in a tropical cyclone was 870 hPa, and the largest observed diameter reached 2220 km (1380 miles). Both of these records were set by Supertyphoon Tip, which swept through the northwestern Pacific in October 1979.

Tropical Weather Outlook showing a hurricane, two tropical storms, and no depressions or disturbances; NOAA

From Disturbance to Monster

The development of a tropical cyclone begins with a tropical disturbance. At this early stage, cumulus clouds begin to form, but the wind field does not yet contain a closed cyclonic circulation.

If conditions are favorable and the circulation becomes closed around the system’s center, a tropical cyclone forms. Its classification then depends on the strength of its sustained winds:

If winds are below 39 mph (62 km/h), it is called a tropical depression.

At 39–73 mph (63–118 km/h), it is classified as a tropical storm and given a name.

Once winds exceed 74 mph (119 km/h), the system is known as a hurricane, typhoon, or severe tropical cyclone, depending on the region.

Tropical cyclone naming by geographic region; MetOffice

From Category 1 to 5: How Hurricanes Are Rated

Hurricanes are further classified by wind speed using the five-level Saffir-Simpson Hurricane Wind Scale. This scale considers only the storm’s sustained wind speed and does not account for other hazardous impacts such as storm surge, flooding from heavy rainfall, or tornadoes. At Category 5, the hurricane’s sustained winds exceed 157 mph (252 km/h), and the result is often catastrophic destruction.

A storm is referred to as a major hurricane if it reaches Category 3 or higher.

Saffir-Simpson Hurricane Wind Scale; NOAA

Types of Damage Caused by Hurricane Winds; NOAA/UCAR

In the case of typhoons, the category of a super typhoon is defined when sustained winds exceed 150 mph (241 km/h).

An Unimaginable Amount of Energy

In just one day, an average tropical cyclone can release approximately 5.2 × 10¹⁹ joule (J) of energy through the condensation of water vapor, which is equivalent to about 14400 terawatt-hours (TWh). According to EMBER, the world’s total electricity production in 2024 was 30850 TWh. That means a single tropical storm can, in one day, generate nearly half of the world’s annual electricity output, and we're talking about an average cyclone, not an extreme case like Supertyphoon Tip.

Hurricane Tracker on Windy.com

Stay Informed with Windy

Windy and its Hurricane Tracker tool let you follow tropical cyclones from their formation through to the latest forecast updates. It displays the storm’s past track, projected path, and expected changes in intensity. Combined with satellite and radar imagery, as well as layers showing wind speed and precipitation, you’ll always have a clear overview of the risks a tropical cyclone may pose in your area.

A new hurricane season is underway. It officially begins on May 15 in the Eastern Pacific. Then, on June 1, the season also starts in the Atlantic basin (including the Atlantic Ocean, Caribbean Sea, and Gulf of Mexico) as well as in the Central Pacific region (including the Hawaiian Islands). In all three regions, the official season runs through November 30.

Basins of tropical cyclone activity in the Atlantic and Northeastern Pacific, NOAA

The official hurricane season marks the time of year when tropical cyclones are most likely to develop over the warm waters of the Atlantic and Eastern Pacific. According to the National Hurricane Center (NHC), more than 97% of tropical cyclone activity in the Atlantic occurs during this designated period.

However, tropical cyclones can also form outside of this timeframe. The earliest Atlantic hurricane ever recorded formed on March 7, 1908.

For the National Hurricane Center (NHC) in Miami and the Central Pacific Hurricane Center (CPHC) in Honolulu, the start of the official season also signals the beginning of regularly scheduled Special Tropical Weather Outlooks, issued four times a day.

Average annual number of named storms and hurricanes, NOAA

An Average Hurricane Season

According to NHC statistics, the highest number of named storms (meaning tropical cyclones that reach tropical storm or hurricane strength) forms in the Eastern Pacific, with an average of 15 per year. The Atlantic produces around 14 named storms annually. However, tropical cyclone activity in the Atlantic basin poses a greater threat to the U.S. and nearby coastal regions, as those storms are more likely to make landfall.

Tropical cyclones typically track west to northwest in tropical and subtropical latitudes (between 5° and 30° N). In the Atlantic, this movement often steers storms toward the eastern shores of North America. Meanwhile, in the Northeastern Pacific, storms usually track out to sea, staying well away from the western coast.

Daily tropical cyclone activity in the Atlantic and Eastern Pacific (May–Dec), NOAA

Tropical cyclone activity usually reaches its highest levels between August and September. In the Atlantic basin, the season typically peaks around September 10. In the Eastern Pacific, the most active period often occurs a bit earlier, generally toward the end of August.

Forecast number of named storms and hurricanes for the 2025 hurricane season, CSU/SMN/NOAA

2025 Hurricane Season Forecast

Current projections indicate that the 2025 hurricane season will be more active than usual in the Atlantic and Northeastern Pacific, while the Central Pacific is expected to see near- or below-normal activity.

In the Atlantic, Colorado State University (CSU) forecasts 17 named storms, including 9 hurricanes and 4 major hurricanes (category 3 or higher). Although activity is expected to be above average, it may still fall slightly below the extremely active 2024 season. CSU will update its forecast on June 11, July 9, and August 6.

In the Northeastern Pacific, Mexico’s meteorological service (SMN) predicts 16 to 20 named storms, 8 to 11 hurricanes, and 4 to 6 major hurricanes.

In the Central Pacific, the National Oceanic and Atmospheric Administration (NOAA) expects 1 to 4 tropical cyclones. Unlike forecasts for other regions, this outlook includes all storm types, from tropical depressions to hurricanes.

Windy.com Hurricane Tracker, Windy.com

Stay Safe

Be prepared, no matter the forecast. Have a plan for you and your family in case a hurricane threatens. Find safety tips at ready.gov/hurricanes.

Stay one step ahead with Windy.com. The Hurricane Tracker lets you follow storms from their formation all the way through to forecast updates. You can also explore radar and satellite imagery, plus detailed weather layers to help you stay informed in real-time.

Media contact: press@windy.com

The southwest monsoon, also known as the summer monsoon, has reached southern India, marking the official start of the rainy season. According to the India Meteorological Department (IMD), the monsoon arrived in the state of Kerala on May 24, 2025, which is eight days earlier than the usual onset date of June 1. This is the earliest onset in the past 16 years.

Average date of monsoon onset in Asia; Encyclopædia Britannica

The summer monsoon follows a period of extreme heat and drought, bringing welcome relief in the form of cooler temperatures and, more importantly, life-sustaining rain essential for agriculture. However, the intense and persistent rains also come with risks: flash floods, widespread inundation, landslides, and damage to infrastructure.

NASA’s visualization of precipitation and soil moisture over South and Southeast Asia on July 7, 2014; NASA's Goddard Space Flight Center

Typically, the monsoon season in India lasts from June 1 to the end of September. The monsoon usually covers the entire country by around July 8. During this period, India receives the majority of its annual rainfall.

In the coming days, monsoon rains will advance inland. They typically reach the capital city of Delhi between late June and early July, and around the same time, moist monsoon winds begin to affect the Himalayan foothills.

What Is a Monsoon?

A monsoon refers to a seasonal shift in wind patterns that occurs twice a year and significantly alters weather conditions, bringing either a rainy season or a dry season to the affected region.

The word monsoon originates from the Arabic word mausim, meaning "season" or "time of year."

The most well-known monsoons occur in South and Southeast Asia, but monsoon-like patterns are also found in other tropical and subtropical regions, such as West Africa, Australia, and parts of North and South America, as well as East Asia (e.g., China and Japan).

How seasonal heating differences between land and ocean trigger monsoon circulation; NOAA/JPL

What Causes the Monsoon?

The monsoon arises from differences in how land and ocean absorb heat. In spring and summer, land heats up faster than the surrounding ocean waters. Warm air over the land rises, creating a low-pressure zone. As air moves from high-pressure to low-pressure areas, moist air from the ocean is drawn inland. This inflow of moisture-laden air forms the summer monsoon.

When this moist air cools, it condenses into clouds and rain, leading to months of heavy rainfall. The heaviest precipitation typically occurs on the windward slopes of high mountains.

Global distribution of monthly mean and anomaly of sea-level pressure; DWD

In contrast, during the winter months, the land cools faster than the ocean, resulting in a high-pressure system over the continent. Cold, dry air then flows from the land toward the ocean. This winter monsoon brings dry weather to most of the region. However, in some island and coastal areas, this air may pick up moisture over the ocean and cause rain or snow.

How Much Rain Does India Receive During the Monsoon?

The average total rainfall across India during the southwest monsoon season (based on 1971–2020 data) is approximately 870 mm. However, there are significant regional differences:

• In northwestern and southeastern India, average rainfall for the season is less than 500 mm
• In contrast, some parts of the southwestern coast and the northeastern states often receive over 1500 mm

Average rainfall totals during the monsoon season (June to September); IMD

One of the most famous regions for extreme rainfall is Mawsynram and Cherrapunji in the Himalayan foothills of northeastern India. According to the World Meteorological Organization (WMO), Mawsynram holds the world record for the highest average annual rainfall, which is 11872 mm between 1941 and 1979, with most of that falling during the southwest monsoon.

Average monthly temperature (°C) and precipitation (mm) in Mawsynram; meteoblue

When it comes to rainfall extremes, Cherrapunji holds the record for the highest annual rainfall ever recorded globally, with 26461 mm falling between August 8, 1860 and July 7, 1861.

2025 Monsoon Forecast

The India Meteorological Department (IMD) forecasts that the 2025 southwest monsoon season will bring above-average rainfall. Total rainfall for the season is expected to reach approximately 106% of the long-period average. During June to September 2025, most parts of the country are very likely to receive normal to above-normal rainfall, except for some regions in Northwest and East India, as well as many areas in Northeast India, where below-normal rainfall is expected.

2025 Southwest Monsoon likely to be above normal (Long Range Forecast); IMD

This forecast is based on several key climate indicators, including sea surface temperatures in the equatorial Indian and Pacific Oceans, and associated phenomena such as the Indian Ocean Dipole (IOD) and the El Niño–Southern Oscillation (ENSO). The IMD also takes into account the extent of snow cover in the Northern Hemisphere, particularly over Eurasia.

• The IOD is currently neutral, but a weak negative phase may develop during the southwest monsoon, potentially leading to reduced monsoon intensity and slightly lower rainfall in parts of India.

• The ENSO is also neutral, but atmospheric patterns resemble La Niña, a phase that typically enhances monsoon rainfall. If this trend continues, it may slightly strengthen the monsoon.

• Eurasian snow cover over the past three months has been below normal, a condition that tends to favor stronger summer monsoons in India.

Illustrative image showing the Rain accumulation forecast layer on Windy.com; Windy.com

The monsoon brings both relief and challenges

Above-average rainfall can benefit crops and replenish water supplies, but it also carries the risk of heavy downpours and flooding.

With Windy.com, you can stay on top of both current and upcoming rainfall:

• Radar+ shows where it’s raining right now, how intense the rain is, and even provides a very short-term forecast with rainfall interpolation for the next 60 minutes

• The Rain, Thunder forecast layer displays expected rainfall for up to 15 days ahead, in 3-hour intervals (1-hour intervals for Premium users)

• With the Rain Accumulation layer, you can easily see how much rain is forecast to fall over your selected time period

Barbara, Ema, Lorenzo… Do those names sound familiar? It’s no surprise. The tropical storms that carried them have impacted the lives of many along the coasts of the Atlantic and eastern Pacific in the past. And this year, these names might appear again. Tropical storms in these regions don’t get random names, they’re selected from pre-determined, rotating lists.

Hurricane Barbara track positions, 30 June–5 July 2019; NOAA

Why do tropical storms get names?

Giving a storm a name is a simple and practical way to make communication clearer and avoid confusion. When several cyclones form over the ocean at the same time, having a distinct name helps meteorologists, the media, emergency responders, and people in affected areas clearly identify the specific storm being discussed.

GOES-16 Sees Three Hurricanes in the Atlantic 2017; NOAA

History of Storm Naming

Historically, tropical cyclones have been given names for a long time, but the process was unstructured. Storms were often named after the saint’s day on which they struck. For example, San Felipe (the first) and San Felipe (the second) hit Puerto Rico on September 13 in 1876 and 1928, respectively.

Another example is Hurricane Antje, which got its name from the ship HMS Antje, whose mast was broken by the storm on August 30, 1842, in the western Atlantic.

Gradually, efforts to bring more order emerged. Storms were identified by geographical coordinates or letters of the phonetic alphabet.

Synoptic map of Hurricane Galveston, named after the city it hit hardest; NOAA

The storm-naming system as we know it today was established in the second half of the 20th century. In 1953, the United States began using female names for storms in the Atlantic. In 1978, both female and male names were introduced into the storm lists for the eastern North Pacific. A year later, in 1979, names of both genders also began to be used for the Atlantic and Gulf of Mexico.

Are all tropical cyclones named?

Tropical cyclones are only given a name once they reach tropical storm status. That is, when their sustained wind speed reaches at least 39 mph (63 km/h).
At that point, the storm receives a name from the current list designated for that season and region.

A Tropical Weather Outlook displaying one named hurricane (Teddy), one named tropical storm (Wilfred), and one numbered depression (Twenty-Two); NOAA

Across the broad hurricane region, three separate name lists are used each year:

• for the Atlantic (including the Caribbean and Gulf of Mexico),

• for the eastern North Pacific,

• and for the central North Pacific.

Each of these lists is created with consideration for the language and cultural context of the region it covers.

Who chooses the names?

The lists of names for tropical storms in the North and Central American region are prepared by the Hurricane Committee of Regional Association IV, which operates under the World Meteorological Organization (WMO).

WMO’s Five Regional Tropical Cyclone Bodies; WMO

When selecting new names, the committee follows several guidelines. Names must be short, easy to pronounce, and understandable across the various languages spoken in the region. They must not duplicate names used elsewhere in the world, and they cannot refer to specific individuals, among other criteria.

Names are listed alphabetically, but certain letters (like Q) are skipped due to a lack of suitable names.

When is a name retired?

If a storm causes exceptional damage or results in significant loss of life, its name can be permanently retired from the list at the request of any member state. The decision is made during the committee’s meeting held after the end of the season. The retired name is then replaced with a new one of the same gender and starting with the same letter. Retiring a name serves as a mark of respect for the victims.

The devastation caused by Hurricane Katrina (2005) led to the retirement of its name from the list; National Wildlife Federation/NOAA

What if there are more storms than names?

This has happened only twice in the Atlantic so far, during the exceptionally active seasons of 2005 and 2020. Once the list of names was exhausted, the Greek alphabet (Alpha, Beta, etc.) was used according to established rules.

However, it became clear that some Greek names were difficult to pronounce or too similar to one another. Moreover, there was no official way to retire a Greek name, even when the storm it represented was especially destructive.

As a result, the Hurricane Committee decided to discontinue the use of the Greek alphabet altogether and replace it with a standard supplemental list of names. This allowed two very powerful hurricanes from 2020, Eta and Iota, to be retired.

Visualization of tropical cyclone activity during the 2020 hurricane season; NASA

Since 2021, two backup lists have existed — one for the Atlantic and one for the eastern North Pacific. These lists are only used if the main list is exhausted.

In the eastern Pacific, however, the main list has never been exhausted. It contains more names than the Atlantic list and has always been sufficient, even in highly active seasons.

Names for Caribbean Sea, Gulf of Mexico, and the North Atlantic

There are six main name lists and one supplemental list used for the Atlantic basin. Each main list is assigned to a specific year and contains 21 names arranged in alphabetical order.

In 2025, the same list is being used as in 2019, 2013, 2007, 2001, and earlier. This means the first named storm of the season will be Andrea, followed by Barry, Chantal, and so on.

Hurricane Dorian moving along the Southeast U.S. coast on September 5, 2019 (satellite and radar imagery); NOAA

However, one change has been made since 2019: the name Dorian has been replaced with Dexter. Why? In 2019, Dorian reached Category 5 strength and devastated Grand Bahama and Great Abaco, later also impacting the Outer Banks of North Carolina. After the season, the name was retired and replaced due to the storm’s severe impact.

According to the latest forecast from Colorado State University (CSU), 17 named storms are expected this year. If that prediction holds true, the last named storm of the season will be Rebekah.

Tropical Storm Names 2025: Caribbean Sea, Gulf of Mexico, and North Atlantic; WMO

Names for the Eastern North Pacific

In the eastern North Pacific, the naming of tropical storms works similarly to the Atlantic region.
There are also six main name lists and one supplemental list. Each list contains 24 names in alphabetical order, and storms are named sequentially, one after another, following the current list.

In 2025, the first tropical storm is named Alvin, followed by Barbara and the next names in alphabetical order, just as it was in 2019. No names were retired after the 2019 season, so the list remains unchanged.

According to the forecast by Mexico ’s national meteorological service (Servicio Meteorologico Nacional, SMN), up to 20 named storms could form this season. If that prediction comes true, the last named storm will be Velma.

Tropical Storm Names 2025: the Eastern North Pacific; WMO

Names for the Central North Pacific

The naming system in the Central Pacific differs from the two previous regions. There are four alphabetical name lists used in rotation, regardless of the calendar year. A new list is only used once all the names from the previous one have been exhausted.

In the 2024 season, the last named tropical storm was Hone. Therefore, in 2025, the naming will continue with the next name on the list, Iona.

Stay Ready, Stay Informed

The 2025 hurricane season is forecast to be above average. Timely information about a storm can give you a crucial head start, and a chance to protect yourself and those around you.

Follow the development of tropical cyclones through satellite and radar imagery on Windy.com. The Hurricane Tracker tool also lets you compare different forecasts of the storm’s path and intensity.

In January, winter climatically peaks in the Northern Hemisphere. Around 22–24 January, the Northern Hemisphere mean daily air temperature (1991–2020 climatological normal) falls to about 8.6 °C, the lowest daily average of the year.

Daily surface air temperature in the Northern Hemisphere; Climate Reanalyzer

Snow is an inseparable part of winter weather. In the following sections, we take a closer look at how a snowflake forms and its “journey” through the atmosphere. To do so, we briefly explore the basics of cloud and precipitation microphysics.

Modelled snow depth, 16 January 2026, 08 UTC (ECMWF); WIndy.com

Cloud Particle Formation

Our story begins in a cloud. For a cloud to form, that is, for tiny cloud droplets or ice crystals to begin forming, the air must first cool, most commonly through ascent.

Schematic diagram of a rising and sinking air parcel; Climate Water Project

For simplicity, let us imagine an air parcel, a small volume of air that we follow as it rises. As it ascends into lower-pressure levels, it expands and therefore cools. The amount of water vapor changes little initially, so the water vapor mixing ratio remains nearly constant, whereas relative humidity increases. This is because the saturation vapor pressure decreases rapidly with temperature (as given by the Clausius–Clapeyron relation), so colder air reaches saturation with less water vapor. At a certain altitude, the air cools sufficiently to reach saturation (100%), and water vapor can begin to condense or deposit to form cloud particles. In the real atmosphere, this process is strongly facilitated by tiny aerosol particles.

Warmer air reaches saturation with more water vapor, so it can hold more water vapor than colder air.

Homogeneous nucleation, the formation of cloud droplets and ice particles directly from water vapor without aerosol particles, is extremely rare in the real atmosphere. It is an energetically demanding and statistically unlikely process that would require very high supersaturation (on the order of hundreds of percent).

The relative size of water molecules to condensation nuclei; NOAA

In practice, therefore, condensation and, in colder clouds, ice formation occur mainly on aerosol particles, which are abundant in the atmosphere and provide surfaces where cloud particles can readily form. For droplet formation, we refer to cloud condensation nuclei (CCN); for ice crystal formation, we use the term ice-nucleating particles (INP). These processes are known as heterogeneous nucleation.

A single aerosol particle can act as both a condensation nucleus and an ice nucleus. The most effective ice nuclei are typically mineral dust particles (e.g., clays and desert dust), soil particles, volcanic ash, and certain biological particles (e.g., pollen, spores, and specific bacteria). By contrast, other aerosols, such as sea salt and many organic particles, often serve as effective condensation nuclei for droplets but are less effective at initiating ice formation.

Schematic illustration of ice-nucleating particle formation from mineral dust aerosols; EGU

Formation of Ice Crystals in Clouds

Because low temperatures alone do not automatically induce the freezing of water droplets, and suitable ice nuclei are often absent, cloud droplets at temperatures below 0 °C (32 °F) are frequently supercooled and remain liquid. As long as ice nuclei or other freezing triggers are not present, such as direct contact with ice, they can persist in the liquid phase well below the freezing point. Particularly small and very clean droplets can remain liquid even at temperatures tens of degrees below freezing. Only at very low temperatures, typically below about −38 °C (−36 °F), do supercooled droplets freeze spontaneously even without ice nuclei (homogeneous ice nucleation).

Occurrence of supercooled liquid water droplets and ice crystals in clouds as a function of air temperature; COMET via NOAA

Current meteorological understanding suggests that, in real clouds, ice particles form primarily by freezing supercooled droplets. The most common pathways are immersion freezing and condensation freezing. In immersion freezing, an ice-nucleating particle is immersed inside a supercooled droplet; once the temperature is low enough, ice forms on the particle and the droplet freezes. In condensation freezing, an aerosol particle first acts as a cloud condensation nucleus on which a droplet forms, and then serves as an ice nucleus that triggers freezing.

Schematic illustration of how MPs (microplastics) can promote atmospheric ice nucleation (a).Experimental images of a gradually cooled MP particle, showing the immersion-freezing process (b). The scale bar in (b) is 20 µm; Philip Brahana 2024

An ice crystal can sometimes grow directly from water vapour without an intermediate liquid-water stage; this is known as deposition nucleation. However, it is relatively uncommon because it typically requires supersaturation with respect to ice and highly effective ice-nucleating particles.

Another possibility is contact freezing, in which a supercooled droplet freezes upon contact with a suitable particle; its importance depends on the frequency of such contacts within the cloud.

Finally, there is homogeneous freezing, which we described above. This process typically occurs in very cold clouds.

Ice nucleation occurs through four mechanisms responsible for forming primary ice crystals in the atmosphere; Wikipedia

From Cloud Particles to Precipitation

For cloud particles to become precipitation particles, i.e., droplets or ice crystals that fall out of a cloud, they must first grow sufficiently. Only when their fall speed, determined by the balance between gravity and air resistance, exceeds the speed of the updrafts do they begin to precipitate. Precipitation particles are typically considered to have radii greater than about 0.1 mm.

However, the available supply of water vapor is not sufficient for all cloud particles to grow to precipitation size, because a liter of cloudy air can contain hundreds of thousands to millions of droplets and ice crystals. Only a fraction of them grow large enough to become precipitating particles, often at the expense of the others.

The primary mechanism by which small cloud particles grow is the diffusion of water vapor. In supersaturated air, water vapor deposits on their surfaces, causing the particles to grow. By contrast, in an unsaturated environment, water evaporates from their surfaces (or sublimates from ice), and the particles shrink and may even disappear entirely.

An illustrative example of the Bergeron–Findeisen process is a fallstreak hole (cavum). It forms when ice crystals develop locally within a thin cloud layer composed of supercooled water droplets, for example, after an aircraft passes through the cloud; NOAA

In clouds where ice crystals and supercooled droplets coexist, the Bergeron–Findeisen process takes place (a diffusion-driven growth process). Because the saturation vapor pressure over ice is lower than over liquid water at the same temperature, the air can be supersaturated with respect to ice while remaining unsaturated with respect to liquid water. As a result, ice crystals grow, while the supercooled droplets gradually evaporate.

Once ice crystals grow large enough for their fall speed to exceed the speed of the updrafts, they begin to fall out of the cloud. During their descent, they can clump together (aggregation) and collide with supercooled droplets, which freeze onto them almost instantly. The crystals then become coated with a layer of frozen droplets, a process known as riming, which can obscure their original shape. Some ice particles can also fragment, increasing their number within the cloud.

Sounding examples: rain, a sufficiently deep warm layer near the surface (left), and snowfall, with temperatures below freezing throughout the profile (right); Windy.com

If the air remains below the freezing point throughout its fall, the precipitation reaches the ground as snow; if it passes through a warmer layer above 0 °C (32 °F), it begins to melt and may fall as rain.

The Remarkable Variety of Snowflakes

A snowflake can consist of a single ice crystal or, more often, an aggregate of ice crystals that forms within a cloud and falls to the ground, where it accumulates as snow. Although snowflakes form through broadly similar processes, which we have largely described above, each one is unique—no two snowflakes are exactly alike.

Because the crystal lattice of ice is hexagonal, ice crystals naturally grow with sixfold symmetry, typically as plates or columns. In plates, growth tends to be strongest at the six corners, while in columns it is strongest along the six edges.

The morphology diagram shows how snow-crystal growth depends on temperature and water-vapour supersaturation; the water saturation line indicates liquid-water supersaturation relative to ice (a). Classic thin, flat stellar-dendrite snowflakes are typically photographed within a narrow range around −15 °C, while slender columns and needles form mainly near −5 °C (b); Kenneth G. Libbrecht 2017

What an ice crystal looks like depends mainly on the temperature and humidity in the cloud: sometimes delicate, highly branched dendrites dominate, while at other times plates, columns, needles, or grains are more common.

As a crystal grows, water vapor deposits onto it, and the crystal retains its hexagonal structure. During its fall, other ice crystals often attach to it, or supercooled droplets freeze onto it. Once a crystal or aggregate becomes visible to the naked eye, it is called a snowflake and can continue to grow. The average diameter of snowflakes is around 5 mm (≈0.2 in), and the largest documented size is reported to be 38 cm (≈15 in).

Close-up of a real snowflake; Photo: Jaroslav Fous

Cloud Types That Most Often Produce Snow

In temperate and subpolar latitudes (and, more rarely, even in the subtropics), snowfall most often comes from mixed-phase clouds containing both supercooled droplets and ice crystals, where snow formation is highly efficient due to the Bergeron–Findeisen process and subsequent aggregation. This is particularly true in the dendritic growth zone, roughly between −12 and −18 °C (about 10 to 0 °F), where crystals grow rapidly and readily aggregate into large flakes.

Cloud type: Significant snowfall can be produced by Ns, Cu con, Cb, and occasionally Sc; Valentin de Bruyn via Wikimedia Commons

Snow most often falls from the following cloud types:

• Nimbostratus (Ns) most commonly produces long-lasting, widespread snowfall. It occurs mainly along warm and occluded fronts and within the broad precipitation area of mid-latitude lows, including the wrap-around region (the “comma head”).

• Stratus (St) and stratocumulus (Sc) usually bring only light snow or brief snow showers, often referred to as low-cloud snowfall.

• Altostratus (As) is often a sign of an approaching warm front. Altostratus itself may produce light snow, but it often gradually transitions into nimbostratus, with precipitation intensifying and typically lasting longer.

• Cumulus congestus produces snow showers. It occurs mainly along cold fronts or in the cold air behind them. During lake-effect snow, it can form long bands (“cloud streets”), often together with stratocumulus (Sc).

• Cumulonimbus (Cb) is associated with very intense snow showers and, in rare cases, thundersnow, which can occur along strong cold fronts.

A striking lunar halo formed by diamond-dust ice crystals; Photo: Jaroslav Fous

Snow can also fall from ice clouds, but less often and usually less intensely than from mixed-phase clouds. These are mainly high-level clouds composed of ice crystals, especially cirrus (Ci), cirrostratus (Cs), and cirrocumulus (Cc). From these clouds, precipitation is typically limited to very light snow or isolated ice crystals, often as virga or diamond dust during dry, very cold weather. Heavy snowfall is rare because ice clouds typically have low ice-water content, the air beneath them is often dry, and they lack the efficient growth and aggregation processes typical of mixed-phase clouds.

Note: In one of our upcoming articles, we will focus on snowfall-favourable situations, covering both synoptic setups and locally driven mechanisms (e.g., orographic snowfall, lake-effect snow, and industrial/anthropogenic snow).

Snowfall and Snow Cover on Windy.com

On Windy.com, you can get a quick overview of current conditions and the expected evolution of snowfall in just a few clicks, or explore a more detailed analysis.

A meteorological radar is available, clearly showing not only precipitation coverage and intensity, but also, more recently, an estimated precipitation type.

Lake-effect snow over the Great Lakes as seen on Windy.com radar (10 November 2025); Windy.com

In addition, you can browse forecasts from numerical weather models. For selected models, you can display layers such as Snow depth and New snow, which represent accumulation over a time period of your choice.

From the meteogram for a specific location, you can track how forecast snowfall accumulation evolves over time, either at an hourly resolution for premium users or in three-hour intervals.

If you want to dive deeper into snowfall forecasting, Windy also offers tools for assessing the atmospheric temperature profile. You can use both measured soundings and model-based (forecast) soundings.

Freezing altitude layer; WIndy.com

A practical feature is the display of the freezing level (0 °C isotherm), either as a standalone layer or as isolines. Because the freezing level often determines whether precipitation falls as snow or rain, it provides a quick way to estimate the snowline. If multiple freezing levels appear in the profile, Windy displays the topmost one.

In mid-January, the far south of Russia’s Kamchatka Peninsula was affected by a series of deep cyclones. The associated winter storms produced exceptional snowfall. In Petropavlovsk-Kamchatsky, 131.6 mm (≈ 5.2 in) of precipitation was recorded from 13–16 January 2026 (120% of the monthly normal), with snow depth at the meteorological station reaching 170 cm (≈ 67 in). According to available records, this was the city’s highest snow depth in 55 years.

A snow-covered Kamchatka Peninsula after mid-January winter storms, 17 January 2026; Windy.com

In Canada’s largest city, Toronto, Toronto Pearson International Airport recorded a daily total of 46 cm (≈ 18.1 in) of fresh snow on Sunday, 25 January, setting a new one-day snowfall record. Observations at the airport date back to 1938.

Observations from Toronto Pearson International Airport, 26 January 2026 (rule of thumb: 1 mm liquid precipitation ≈ 1 cm snow); Windy.com

Even parts of northwestern and western Japan have experienced record-breaking snowfall this winter. Above-average snow depths are the result of frequent January snowfall, which affected Hokkaido in particular and the western part of Honshu.

In Sapporo, Hokkaido, snow depth reached 110 cm (≈ 43.3 in) on 26 January. The last time it exceeded the one-metre mark was in January 2005, 21 years ago.

Aomori captured by a webcam (3 February 2026). In early February, snow depth here reached 183 cm (≈ 72 in), the highest value in the past 40 years. Windy.com; webcam source: rab.co.jp

In Aomori City, in northern Honshu, snow depth reached 183 cm on Sunday, 1 February 2026. According to the Japan Meteorological Agency, this is the highest value in the past 40 years and the first time since 1986 that snow depth has exceeded 180 cm (≈ 70.9 in). The all-time record in Aomori is 209 cm, observed on 21 February 1945; records at this site date back to 1893.

Key facts about snow; WMO

Extreme snowfall in these regions was linked to winter storms during frigid Arctic air outbreaks. In addition to synoptic-scale mechanisms such as low-pressure systems, local and regional processes also played an important role in further enhancing snowfall totals.

In this article, we examine the processes that determine how much snow from a cloud actually reaches the ground and highlight selected historical episodes of extreme snow accumulation. Loosely building on the article Winter Weather: From Water Vapour to Unique Snowflake, we shift from snowflake formation to the conditions that allow snow to accumulate at the surface.

Typical Synoptic Setups for Snowfall

Snowfall most often occurs during synoptic setups that typically produce precipitation. If the atmosphere is sufficiently cold and temperatures remain below 0 °C (32 °F) throughout the entire vertical profile from the cloud layer down to the surface, precipitation will fall predominantly as snow. This includes various parts of low-pressure systems, from frontal boundaries (warm fronts, cold fronts, and occlusions) to the broad precipitation shields of cyclones, such as wrap-around/deformation zones (e.g., the comma head).

Schematic of a developing low-pressure system with the precipitation area highlighted; NOAA

In addition, some situations can trigger snowfall or enhance it locally, such as industrial snowfall, orographic snowfall, and lake-/sea-effect snow.

Schematic vertical cross-section of a cold front and a warm front with typical cloud types and precipitation; Met Office via Future Learn

Industrial Snowfall (Anthropogenic Snow)

Industrial snowfall most often occurs at low temperatures (often around −5 °C, 23 °F), under calm or light winds, with low clouds or fog, and with a high concentration of both condensation nuclei and ice nuclei.

Schematic of industrial snow formation; Velle Toll 2024

The low cloud layer typically has a base near the ground (below ~150 m, ≈ 500 ft) and a thickness of about 200–300 m (≈ 650–1,000 ft); higher-level clouds are often absent above it. Nocturnal radiative cooling of the upper part of the cloud/fog layer promotes mixing and the formation of supercooled droplets. Additional aerosols, moisture, and waste heat from industrial sources can further enhance mixing, accelerate ice particle growth, and trigger light snowfall, most often just before sunrise.

Usually, only a thin layer of snow falls, often less than 2 cm (≈ 0.8 in), but documented cases exist, for example, in West Virginia, USA (near the John E. Amos power plant, December 1975) or in the Texas Panhandle, USA (February 2014), where around 10 cm (≈ 4 in) of snow accumulated overnight.

Localized industrial snow over the Netherlands, observed by Landsat 8 (19 January 2017); NASA Earth Observatory via Edward Graham 2024

Orographic Snowfall

Orographic snowfall occurs in mountainous terrain when moist air is forced to rise along the windward side of a mountain range. As the air ascends, it cools, condenses, and forms clouds, from which snow can fall when the atmospheric column is sufficiently cold. The key factor here is not atmospheric instability, but the mechanical forcing of upward motion as airflow passes over the terrain.

Schematic of upslope snowfall and snow depth over a mountain range (areas B and C not included); Rebecca Mott 2018

A pronounced spatial variability is typical: tens of centimeters of snow (several inches to over a foot) can fall in a short time on windward slopes, while on the leeward side, where air descends and dries out, a so-called snow shadow forms with much lower totals, sometimes with no precipitation at all.

Another characteristic is persistence: under steady airflow and with sufficient moisture, snowfall can continue for hours or even days, often leading to very high accumulations in mountainous areas.

Mean daily precipitation: statistically, the highest totals are found in the tropics and mountainous regions; Copernicus Interactive Climate Atlas

Orographic snowfall is therefore responsible for the highest long-term snowfall totals in many mountainous regions worldwide, including the Alps, the Rocky Mountains, the Cascade Range, the Canadian Coast Mountains, the Japanese Alps, and New Zealand’s mountain ranges. Rapid snowpack buildup during persistent orographic snowfall also directly affects avalanche danger and hydrological conditions in mountain watersheds.

Orographic snowfall can also enhance precipitation that is already present. It can significantly intensify snowfall along fronts or amplify lake-effect snow in areas where snow bands encounter mountainous terrain.

Schematic of Lake-Effect Snow; NOAA

Lake-Effect Snow in the Great Lakes Region

In the North American Great Lakes region, lake-effect snowfall can develop as heat and moisture from the lake surface are transferred into the atmosphere. It occurs when very cold Arctic air flows over the relatively warm water, warming and moistening as it passes over the lakes.

Great Lakes surface-water temperatures and air temperature isotherms. A strong air–water temperature contrast supported the season’s first significant lake-effect snow event (10 November 2025); Windy.com

A commonly used rule of thumb is that for lake-effect snow to develop, the temperature difference between the lake surface and the air at about 1.5 km altitude (≈ 1 mi, ≈ 850 hPa) should reach at least 13 °C (23 °F). This promotes instability and the formation of cumulus (Cu) and stratocumulus (Sc) clouds, and in more extreme cases, even cumulus congestus (Cu con) and locally cumulonimbus (Cb).

Wind direction is also important because it determines the length of the airflow path over open water (fetch). The longer the fetch, the more heat and moisture the air can pick up from the lake. To produce long snow bands and “cloud streets,” a fetch of at least about 100 km (≈ 60 mi) is usually needed, although the exact value depends on the temperature contrast between the water and the air, atmospheric instability, and wind shear.

Lake-effect snow on radar: precipitation bands over Lake Michigan affecting the Chicago–South Bend area (10 Nov 2025).; Windy.com

A typical feature of these bands is pronounced spatial variability. They are often very narrow (only a few kilometers wide), so while more than 75 cm (≈ 30 inches) of snow may fall in one area, only a few centimeters may accumulate just a few kilometers away.

One of the most remarkable multi-day episodes occurred from 3 to 12 February 2007 across the Tug Hill Plateau near Redfield, New York, when up to 141 inches (≈ 358 cm) of snow fell over 10 days.

The largest reported 24-hour lake-effect total is often cited as 77 inches (≈ 196 cm) in Montague, New York, during 11–12 January 1997; however, because standard measurement procedures were not followed, this value was not accepted by an NWS review committee as an official 24-hour record.

Lake-effect storm (Feb 3–12, 2007): Max snowfall – Lake Ontario 141 in (≈358 cm; Redfield); Lake Erie 42 in (≈107 cm; East Aurora); NOAA

Lake/Sea/Ocean-Effect Snowfall

Lake-effect snow is not limited to the Great Lakes region, but can occur wherever very cold air flows over a sufficiently large, relatively warm body of water and then moves onshore. When the water body is a sea or an ocean, the same mechanism is referred to as sea-effect or ocean-effect snow. A classic example is Japan’s Sea of Japan coast, where the winter northwesterly monsoon produces heavy snowfall, often further enhanced by orographic lifting. Similar conditions occur, for instance, around the Black Sea (especially in northern Turkey, including Istanbul) and around the Caspian Sea, particularly along the southwestern coast of northern Iran.

Sea-effect snowfall over Japan (25 January 2026); Windy.com

Blizzards Are Formed by a Combination of Strong Winds and Snow

Snow plays a major role in a blizzard, but it does not necessarily have to be snowing during the event. The U.S. National Weather Service (NWS) defines a blizzard as a situation in which sustained winds or frequent gusts reach at least ~35 mph (≈ 56 km/h), and significant falling and/or blowing snow occurs, often reducing visibility to 1/4 mile (≈ 400 m) or less for at least 3 hours.

Weather warning layer: Blizzard Warning for parts of Minnesota and North Dakota (17 January 2026); Windy.com

A blizzard is typically associated with a deep low-pressure system (an intense winter storm), with the worst conditions often occurring in its rear sector and in the wrap-around/deformation zone (the comma head), where strong winds combine with snowfall. Another common scenario occurs behind a strong cold front, with gusty winds and snow showers. Hazardous conditions can also develop during a ground blizzard, when dry, powdery snow is lofted and blown around between a departing low and a strengthening high-pressure system.

During a blizzard, the main hazard is a rapid drop in visibility, sometimes to whiteout conditions, when falling and blowing snow makes it difficult to maintain orientation even over short distances. Strong winds quickly redistribute snow and create drifts, so even with relatively small new snowfall totals, roads can become impassable, vehicles can get stranded, and some areas may become isolated. The wind also greatly increases heat loss from the body (wind chill), raising the risk of hypothermia and frostbite.

Iran was hit by the world’s worst snowstorm in 1972: rescue operations (left); life in Tehran on 11 February 1972 (right). Ettela'at newspaper via Wikipedia

The early February 1972 blizzard in Iran (3–9 February) is often cited as one of the worst and, in some accounts, among the deadliest in modern history. Over the course of about a week, the country was hit by a series of exceptionally intense snowstorms accompanied by severe cold. In some areas, snowfall totals reached several metres, and snowdrifts in the south were estimated at up to 7.9 m (≈26 ft). The town of Ardakan and nearby villages were among the hardest hit, with no survivors reported in Kakkan and Kumar. During the cold spell, temperatures in some areas dropped to −25 °C (−13 °F). Estimates suggest the disaster claimed around 4,000 lives.

The Great Blizzard of 1978: Snow removal in Indiana (photo by Argil Shock / The Fort Wayne News-Sentinel), a car stranded in large snowdrifts in the Ohio Valley, and the synoptic setup over the eastern United States on 26 January 1978. NOAA; NOAA

In the United States, one of the most destructive blizzards was the Great Blizzard of January 1978, which occurred from 24 to 28 January. An exceptionally deep cyclone hit mainly the Ohio Valley and the Great Lakes region: its central pressure dropped to 955.5 hPa, and wind gusts in Cleveland reached 82 mph (132 km/h). The storm brought intense snowfall, massive snowdrifts, and widespread whiteout conditions. It claimed around 90 lives and caused an estimated $100 million in damage (in 1978 dollars).

A North Dakota DOT employee next to the top of a utility pole during a blizzard (March 9, 1966). Collection of Dr. Herbert Kroehl, NGDC via NOAA

One last interesting fact: Yuki-no-Otani

Yuki-no-Otani is a snow corridor along the Tateyama Kurobe Alpine Route, a mountain route across the Northern Japanese Alps that is world-famous for its giant snow walls. In the Tateyama massif, especially near the highest point of the route at Murodo (about 2,450 m (≈ 8,000 ft)), an exceptional amount of snow accumulates during winter. This is due to a combination of the winter northwesterly (monsoonal) flow, which brings moist air from the Sea of Japan, and strong orographic enhancement on the windward side of the Japanese Alps.

The Tateyama Kurobe Alpine Route runs through the Japanese Alps. Kyodonews/ZUMA Press via CNN

In winter, the route is closed due to the heavy snowfall and reopens in early spring, when heavy machinery must literally cut through the snow to create the corridor. In April 2025, the snow walls reached roughly 16 m (≈ 52 ft), while the historical maximum of around 20 m (≈ 66 ft) was recorded in 2000.

Snow walls on the Tateyama Kurobe Alpine Route; JNTO

Snowfall and Snow Cover on Windy.com

Windy.com offers a wide range of features to help you monitor snowfall, estimate how precipitation will evolve over the next few hours, and assess how much snow may accumulate on the ground.

View from a webcam in Livigno: Belvedere (1 February 2026); Windy.com; webcam source: livigno.panomax.com

For an up-to-date picture of current conditions, use radar and satellite observations, weather-station measurements, radiosonde observations, and webcams, which give a direct view of what is happening on the ground.

Sounding examples: rain, a sufficiently deep warm layer near the surface (left), and snowfall, with temperatures below freezing throughout the profile (right).

For the forecast, Windy.com includes multiple model layers. The key layers for snow conditions are New snow (forecast snowfall accumulation over the selected time period) and Snow depth (forecast snow cover on the ground). In general, the most useful level of detail is obtained by using the highest-resolution model available for a given area.

A vertical cross-section of the atmosphere along a selected route, using the Distance & Planning feature (VFR mode), showing flight levels, precipitation (blue for rain, cyan for snow), visibility, and freezing altitude (blue line); Windy.com

To estimate at what altitude snow may turn into rain, you can use the Freezing altitude information. It shows the approximate altitude of the highest freezing level (0 °C/32 °F) and can be viewed as a map layer, as isolines, or as a vertical cross-section along a selected route in VFR mode.

Meteorological spring began in the Northern Hemisphere on Sunday, 1 March. Astronomical spring follows with the vernal equinox on 20 March. Across many parts of Europe, the first low concentrations of alder pollen are already being observed, and as temperatures rise and days grow longer, pollen levels are expected to increase rapidly.

Bee collecting pollen on a dandelion; photo by Nadine Eggenberger

After winter, flowering plants are a welcome sign of the new season for many of us, but for allergy sufferers, they also mark the start of symptoms. Pollen grains can trigger hay fever, asthma, or conjunctivitis. It is estimated that about one in four Europeans suffers from pollen allergy, making it one of the most common chronic conditions.

Scanning electron micrograph showing pollen grains of paper birch (Betula papyrifera) on the left and common ragweed (Ambrosia artemisiifolia) on the right; Cellimagelibrary.org

The timing of flowering and the amount of pollen in the air are closely linked to the weather. Meteorological conditions determine when plants start to bloom, how much pollen they release, how far the wind carries it, and when it gradually settles back to the ground. If we can describe when and how much pollen is released, numerical atmospheric models can be used to forecast pollen concentrations.

Pollen forecasts help allergy sufferers plan everyday activities and also allow them to start or adjust treatment in time, following medical advice, thereby reducing the risk of symptoms worsening during the season.

On Windy.com, you can view pollen concentration forecasts for selected pollen types at any location in Europe. These forecasts are provided by the Copernicus Atmosphere Monitoring Service (CAMS).

In this article, we will take a simplified look at how this pollen forecast is produced.

What is CAMS, and what does it provide?

The Copernicus Atmosphere Monitoring Service (CAMS) is part of the Copernicus programme and focuses on atmospheric composition monitoring. It is operated by the European Centre for Medium-Range Weather Forecasts (ECMWF). Its goal is to provide high-quality, well-documented, and openly accessible information on air pollution and other atmospheric components at global and European scales.

Example of a CAMS product: aerosol forecast; Windy.com

CAMS produces near-real-time analyses and forecasts of atmospheric composition, as well as long-term reanalyses that enable assessment of past developments and trends. Key products include forecasts of major air-quality pollutants such as particulate matter (PM₁₀ and PM₂.₅), nitrogen dioxide (NO₂), and ozone (O₃), as well as aerosol forecasts (including Saharan dust and wildfire smoke), estimates of fire emissions (GFAS), and information on the ozone layer and UV radiation. On the European scale, CAMS also provides pollen forecasts.

Why are CAMS pollen forecasts available only for Europe?

CAMS pollen forecasts are part of the CAMS Regional Production System. This system is designed specifically for the European domain, with a horizontal resolution of 0.1° × 0.1° (about 10 km). It uses detailed European vegetation datasets and a dense observation network, and is therefore not part of the global CAMS system.

European hourly air quality forecast of alder pollen concentrations; CAMS

The CAMS Regional Production System is based on 11 European chemical transport models for air quality: CHIMERE (INERIS, France), EMEP (MET Norway, Norway), EURAD-IM (Jülich IEK, Germany), LOTOS-EUROS (KNMI and TNO, the Netherlands), MATCH (SMHI, Sweden), MOCAGE (Météo-France, France), SILAM (FMI, Finland), DEHM (Aarhus University, Denmark), GEM-AQ (IEP-NRI, Poland), MONARCH (BSC, Spain), and MINNI (ENEA, Italy).

Ground-level alder pollen concentrations; CAMS

Coordination is provided by the Central Regional Production Unit (CRPU), led by Météo-France. Each model produces its own pollen forecast. Forecasts are run at 00 UTC and are available hourly up to four days ahead (T+96). The CRPU then combines the individual outputs into an ensemble product, typically as the median at each grid point.

On the Copernicus website, both the individual model outputs and the combined ensemble product are available.

Which pollen types are included in the forecast?

Six pollen types are included in the forecast: alder, birch, olive, grasses, mugwort, and ragweed.

These are among the most important pollen allergens in Europe, both because of their wide geographic distribution and because they can trigger allergic reactions in a substantial part of the European population.

Distribution of black alder (Alnus glutinosa) in Europe and the frequency of records in national forest inventories. Native range: EUFORGEN; European Commission Tree Atlas

In early spring, alder and birch are among the most significant pollen sources, especially across northern and central Europe. In many regions, alder can start the pollen season in the winter months, while birch dominates spring and is among the most allergenic tree pollens. Pollen from these trees most commonly causes allergic rhinitis and conjunctivitis, and in more sensitive individuals, it can also worsen asthma symptoms. Birch pollen is also well known for cross-reactivity with certain foods, which can trigger symptoms after eating some types of fruit or nuts. In Europe, around 8–16% of the population is sensitised to birch pollen.

Distribution of silver birch (Betula pendula) in Europe and the frequency of records in national forest inventories. Native range: EUFORGEN; European Commission Tree Atlas

From late spring into early summer, grasses dominate and are widespread across almost all of Europe. They are the most widespread source of airborne pollen and one of the most common triggers of pollen allergy. Typical symptoms include hay fever, allergic conjunctivitis, and, often, asthma-like symptoms. Grass pollen is recognised as one of the principal causes of pollen allergy in Europe, with national sensitisation rates of up to ~26% reported in some countries, although values vary by region and study population. In the Mediterranean region, olive tree pollen is another important spring and early-summer allergen, and in some areas it can be the main driver of symptoms, with high sensitisation rates.

Distribution of olive tree (Olea europaea) in Europe and the frequency of records in national forest inventories. Native range: EUFORGEN; European Commission Tree Atlas

Late summer and autumn are typically associated with mugwort and ragweed pollen. Mugwort is native to Europe and widely distributed, while ragweed is an invasive species that is spreading, especially in central and south-eastern Europe. Ragweed pollen is among the most potent aeroallergens and is often linked to severe allergic symptoms and asthma exacerbations. Sensitisation rates to ragweed pollen in Europe vary markedly by region, with some national estimates in the ~10–20% range (e.g., ~16% in Germany). These late-season allergens can extend the pollen season well into autumn.

Birch and alder pollen exposure in Europe. Maps were created using data from the Medical University of Vienna, Vienna, Austria, and represent medium to very high pollen levels; T. Biedermann, 2018

How is a pollen forecast produced?

In simplified terms, producing a pollen forecast can be broken down into a few key steps that together determine the resulting pollen concentrations in the air. First, emissions are estimated, that is, the spatio-temporal pattern of pollen release. Next comes atmospheric transport, dispersion, and deposition to the surface. Finally, the forecast is evaluated by comparing it with observations.

The individual operational models in the CAMS system may differ in how they numerically solve transport (also known as advection) and deposition, but the emission schemes are, to some extent, harmonised and coordinated across the system.

Common ragweed (Ambrosia artemisiifolia) in the early flowering stage, a highly allergenic invasive species in Europe; ECMWF

How does the model estimate pollen emissions?

In the model, the pollen season is not determined by real-world plant development, but by the accumulation of temperature above a threshold, expressed as temperature sums (degree days). The CAMS system uses two degree-day thresholds. Once the accumulated temperature sum exceeds the first threshold, the season starts. After the second threshold is reached, the season is considered to have ended.

The threshold values are derived from real-world measurements. However, these measurements are not ingested into the model operationally every day; instead, they are used retrospectively for validation and, if needed, for tuning parameters (such as the degree-day thresholds) during calibration. The operational forecast, therefore, relies on the meteorological forecast and predefined, calibrated parameters.

Map showing the temperature-sum thresholds for the onset and termination of the birch pollen emission season; M. Sofiev, 2013

Because in reality the season does not start abruptly, the model applies a gradual, probabilistic ramp-up and ramp-down of emissions. As the temperature sum increases, an increasingly larger fraction of the seasonal pollen reservoir is released.

Next, an emission flux is calculated for each grid cell, that is, the amount of pollen released into the air at a given moment (the number of pollen grains released per unit area per unit time). This requires input data on the distribution of individual pollen types (alder, birch, grasses, etc.) in the landscape and on their production potential.

How much pollen is actually released on a given day depends on the current weather. Emissions are therefore adjusted based on meteorological conditions. Rain and high humidity suppress release (pollen becomes moist and remains on the plant), whereas wind and turbulence enhance it by mechanically liberating pollen into the air. The wind effect is capped so that under very strong winds, emissions do not increase without limit but instead gradually level off.

At the same time, information on production potential is used to determine how much pollen is available for a given pollen type and season in a given area (the seasonal reservoir). Once this reservoir is depleted, pollen is no longer released, even if, according to the temperature thresholds, the season has not yet formally ended.

The output of the emission module is the pollen emission flux at each grid point and time step.

Photo by Alex Jones

What controls pollen transport in the atmosphere?

After emission, transport determines how pollen is distributed in the atmosphere. In the model, pollen is treated as a chemically inert particle, meaning it does not undergo chemical changes. It is carried horizontally by the wind (advection) and mixed within the atmospheric boundary layer by turbulence, which drives both vertical and horizontal dispersion. Turbulent mixing is often enhanced by convection, especially during daytime surface heating, when updrafts and downdrafts distribute pollen more efficiently across different heights. Transport is therefore mainly governed by meteorological conditions, such as the wind field, atmospheric stability, and the depth of the boundary layer.

In parallel, pollen is removed from the air by deposition. Dry deposition includes gravitational settling as well as turbulent impaction and interception on surfaces. For example, birch pollen typically has a settling velocity of a few centimetres per second.

Wet deposition is also very important, namely, precipitation scavenging within and below clouds. Precipitation has a major impact on pollen concentrations, and even relatively small rainfall amounts can substantially reduce airborne pollen levels.

Generation of the individual forecasts

By combining emissions, transport, and deposition, each of the 11 European chemical transport models produces its own forecast of pollen concentrations across the grid and over time, expressed in pollen grains per cubic metre.

These individual outputs are then combined into a single representative ensemble product, usually computed as the median value at each grid point.

Daily mean concentrations of birch pollen grains from the CAMS regional ensemble forecast; Copernicus

This ensemble approach improves forecast stability and reliability by reducing the influence of extreme errors from individual models and by blending their different representations of emissions, transport, and deposition into a more robust result.

Are pollen concentration forecasts compared with real-world conditions?

The final and crucial step in producing a pollen forecast is validation. Modelled pollen concentrations are retrospectively compared with real measurements from aerobiological monitoring stations of the European Aeroallergen Network (EAN). The evaluation considers, for example, overall forecast error, systematic bias, and whether the model correctly captures the onset of the season, its development, and peak concentrations.

Comparison of observed birch pollen concentration (MeteoSwiss/EUMETNET/Autopollen) with the CAMS forecast for the same period in Zurich; ECMWF

The EAN is a pan-European database of pollen and fungal spore measurements. Operating since 1988, it now includes data from more than 600 stations in 38 countries. It serves as a central platform for collecting, standardising, and sharing both long-term and near-real-time pollen data from national networks and laboratories.

CAMS Evaluation Report 2024: RMSE of birch, alder, and mugwort pollen forecasts at individual EAN stations (model error, pollen grains m⁻³); Copernicus

These data allow continuous assessment of model performance and the identification of weaknesses. Validation results support further model development and are used to adjust parameters and schemes in subsequent versions, including phenological thresholds, temperature sensitivity, the influence of humidity and precipitation, and source maps. In this way, forecasts can be progressively refined in future seasons, both at the level of individual models and of the overall ensemble product.

Pollen forecast on Windy.com

On Windy.com, you can view CAMS pollen forecasts in the detailed forecast for a specific location by switching to Pollen & Air Quality. If higher pollen levels are expected over the next four days, the chart will show the forecast concentrations for the relevant pollen types. The displayed value represents the CAMS ensemble median.

Pollen is shown alongside air quality to better assess the overall allergy load. Air pollution can worsen symptoms and, in some cases, interact with pollen grains to increase their allergenic potential. Pollutants such as particulate matter (PM₁₀ and PM₂.₅), ozone (O₃), nitrogen dioxide (NO₂), and sulfur dioxide (SO₂) can lead to more intense reactions than pollen alone. As a result, days with high pollen combined with poor air quality can feel more challenging than high-pollen days with cleaner air.

Residents and travelers across the central Mediterranean region should prepare for dangerous weather conditions as low-pressure system Ines moves across the area. The center of the system is gradually shifting from northwestern Africa toward the Ionian Sea and is expected to continue northeastward into Eastern Europe.

Track of low-pressure system Ines (ICON EU), windy.com

What to Expect

The storm is bringing intense rainfall and strong thunderstorms. Rainfall totals may reach around 60 mm, with some areas, such as Sicily, Calabria, Apulia, and northwestern Greece, potentially seeing over 100 mm. This significantly increases the risk of flash flooding. Thunderstorms may also bring brief periods of strong wind gusts.

Total precipitation until Sunday 18th, 00 UTC (ICON EU), Windy.com

On Wednesday, Ines affected northwestern Africa. Northern Algeria was hit by severe thunderstorms that, according to the European Severe Weather Database, produced hailstones up to 2.5 cm in diameter.

Progression of Ines

Stormy weather is forecast to affect Tunisia, Sicily, and southern Italy on Thursday, bringing intense rain, strong winds, hail, and a high risk of flash flooding. An orange alert is in effect for Sicily and parts of southern Italy.

The center of the low-pressure system will move over the Ionian Sea on Friday and continue toward Greece, with severe weather expected to impact the southwestern Balkans.

As the weekend progresses, the system is expected to gradually weaken while shifting northeast. The heaviest rainfall will move from northeastern Greece to eastern Bulgaria. Meanwhile, weather conditions in the central Mediterranean should steadily improve.

Stay Weather-Ready

Stay updated by following weather warnings from local meteorological services (available via MeteoAlarm or Windy.com – Weather Warnings layer). For a detailed forecast in your area, including expected rainfall and wind gusts, check the meteogram (e.g., Windy.com – meteogram for Catanzaro).

Stay tuned for further updates as this dynamic weather system progresses across the region.

Media contact: press@windy.com

I residenti e i viaggiatori in tutto il Mediterraneo centrale dovranno prepararsi a condizioni meteorologiche avverse, poiché nei prossimi giorni è previsto l'arrivo di un potente sistema di bassa pressione, denominato Ines, che porterà forti piogge, temporali e venti violenti.

Totale delle precipitazioni fino a domenica 18, ore 00 UTC (ICON EU), Windy.com

Cosa ci si aspetta

Si prevede che il totale delle precipitazioni raggiungerà circa 80 mm, con oltre 130 mm di pioggia possibili in aree come l'Algeria settentrionale, la Sicilia, la Calabria e la Puglia. Ciò aumenterà il rischio di alluvioni improvvise e frane. Saranno inoltre probabili raffiche di vento fino a 70 km/h, con raffiche che portanno essere più forti durante i temporali.

Traiettoria del sistema di bassa pressione Ines (ICON EU), Windy.com

Sviluppo della tempesta

Il sistema si sta muovendo verso nord-est dall'Africa nord-occidentale e continua a intensificarsi. Mercoledì, forti temporali hanno colpito l'Algeria settentrionale, con piogge occasionali che hanno raggiunto anche le Isole Baleari e la Sardegna meridionale.

Giovedì le condizioni temporalesche si estenderanno alla Tunisia, alla Sicilia e all'Italia meridionale, portando piogge intense, forti raffiche di vento e grandine, con il rischio di inondazioni improvvise.

Entro venerdì, il centro della bassa pressione si sposterà sul Mar Ionio e continuerà a muoversi verso la Grecia, portando un forte maltempo nei Balcani sud-occidentali.

Sabato, si prevede un graduale indebolimento del sistema verso nord-est. Tuttavia, l'Albania e la Grecia settentrionale potrebbero ancora registrare oltre 100 mm di pioggia in alcune zone. Nel frattempo, si prevede un miglioramento delle condizioni meteorologiche in Sicilia e nell'Italia meridionale.

Meteogramma per Sciacca, windy.com

Tenetevi aggiornati sulle condizioni meteorologiche

Rimanete aggiornati seguendo gli avvisi meteo dei servizi meteorologici locali (disponibili tramite MeteoAlarm o Windy.com – layer Avvisi meteo).

Per le previsioni dettagliate nella vostra zona, comprese le precipitazioni previste e le raffiche di vento, consultate il meteogramma (ad esempio, Windy.com - meteogramma per Sciacca).

Restate sintonizzati per ulteriori aggiornamenti mentre questo sistema meteorologico dinamico si sposta.

Contatto per i media: press@windy.com

@lexolutha Hi, for now, the forecast always opens in the basic view by default. However, in the future, we plan to add an option to customize the default view. Thanks for the suggestion, Jari

@JohnPiepers Thank you for your contribution.

You're absolutely right that the climate of our planet is constantly and dynamically changing. Estimates of global temperature evolution in Earth's geological past indeed show that the planet's average temperature has been higher than it is today for long periods of time.

However, this article focuses on the current challenges and needs of our society within a timeframe of several human generations.
Its main message is that it is essential to adapt to extreme temperatures and heatwaves, which, within this context, appear to be becoming more frequent and intense, and to implement measures that can mitigate their impact on society.

The article aims to summarize the current state of extreme heat and heatwaves on Earth (see the introduction and the chapter Global temperature increase), their impact on human society, particularly in urban environments (chapters Impacts of extreme heat and heatwaves and Impacts of heatwaves in cities), and to highlight the anticipated developments in the future, both in terms of urbanization (chapters The dynamic evolution of cities and The future of cities) and in terms of expected global temperature rise (chapters Global temperature increase, Global climate shifts of cities, and The future of cities).

The climate of our planet is the result of complex and interconnected processes whose effects can either amplify or offset one another. It is possible that forecasts of future global temperatures, based on our current level of understanding, may not fully capture all variables and feedback mechanisms.

Warm regards, Jari

@Mathew1998 Hi, just so that the article wouldn’t be too long and complicated. There is room for explanation here. So I apologize for the unsolicited explanation.

The Convective Condensation Level (CCL) is the height at which cumulus clouds begin to form. So like the Lifting Condensation Level (LCL), it represents the cloud base, but differs in the mechanism by which an air parcel reaches it: in the case of the LCL, the parcel is lifted by an external force (a so-called triggering mechanism, such as a front), whereas in the case of the CCL, it rises on its own due to buoyancy.
A dry air parcel heated by the warm surface becomes warmer than the surrounding air and begins to rise due to buoyant force. As it ascends, it cools at the dry adiabatic lapse rate (approximately 10 °C per kilometer, following the dry adiabat) and its relative humidity increases. At the CCL, the relative humidity reaches 100 %, and the parcel becomes saturated. At this point, condensation of water vapor begins, forming a visible cloud. With further ascent, the air cools at the moist adiabatic lapse rate (following the the moist/wet adiabat).

@robobo1 Hi, I hope you'll be pleased to hear that Windy.com has decided to add smoke forecasting to the app. Thank you very much for your suggestion. It’s thanks to posts like yours that we started considering the inclusion of this feature. We truly appreciate your help in shaping an app that’s as useful as possible for you. Warm regards, Jari