Before a single wind turbine foundation is drilled or a single planning application is submitted, engineers and developers need to know whether the wind at a proposed location is good enough to justify the investment. Wind mapping — the science of measuring, modelling, and visualising wind resources across regions and oceans — provides the essential foundation for all of this. Wind atlases, the published products of large-scale wind mapping efforts, allow developers, governments, and researchers to quickly identify where the best resources exist before committing to costly site-level measurements.
Wind is not uniformly distributed across the Earth's surface. It varies with latitude, elevation, proximity to coastlines, terrain shape, land cover, and complex interactions in the atmospheric boundary layer — the lowest few kilometres of the atmosphere where the Earth's surface exerts a frictional influence on airflow. Mapping this variability accurately at the scale needed for wind energy planning requires a combination of physical measurements, numerical weather models, machine learning techniques, and satellite observations.
This guide explains how wind maps are constructed, what they show, how they are used in practice, and what their limitations are. It also introduces the major publicly available wind atlases and explains how site-level measurement campaigns complement and validate the broader picture these atlases provide. Whether you are studying wind energy or evaluating potential sites, understanding wind mapping is fundamental.
Why Wind Mapping Matters
The economics of a wind energy project depend overwhelmingly on the quality of the wind resource at the site. Wind power scales with the cube of wind speed — a site with an average wind speed 10% higher than another will produce roughly 33% more energy, all else being equal. Over a 25-year project life this difference compounds into a profoundly different economic outcome. Getting the wind resource estimate right therefore directly determines whether a project is financially viable.
Wind mapping at the regional scale serves several distinct purposes. Governments and energy planners use it to identify zones suitable for wind development and to set national capacity targets. Developers use it to prioritise which areas to explore in more detail before committing to expensive field measurement campaigns. Academic researchers use wind maps to study the atmospheric processes that drive wind variability and to project how climate change may affect wind resources in future decades.
Poor wind mapping has historically led to costly mistakes — projects built in locations where the wind resource turned out to be weaker than expected, leading to underperformance, financial distress, and sometimes early decommissioning. Improvements in mapping tools and techniques over the past two decades have substantially reduced this risk, but no remote sensing or modelling product fully replaces careful on-site measurement. The guide on wind resource assessment explains how site-level campaigns fit into the broader workflow that starts with wind maps.
The Atmospheric Boundary Layer and Why It Matters for Maps
The atmospheric boundary layer (ABL) is the lowest part of the troposphere — extending from the ground to roughly 1–2 km altitude in the daytime — where surface friction slows the wind and turbulence mixes air vertically. Above the ABL, the free atmosphere flows largely unconstrained by surface features. Within the ABL, wind speed increases with height (the wind profile or wind shear), turbulence intensity varies with terrain roughness and atmospheric stability, and complex flows develop around hills, forests, and coastal zones.
Wind energy is extracted within the ABL, which is why understanding boundary layer physics is central to wind mapping. A turbine with its hub at 120 metres may experience significantly different wind speeds than one at 80 metres on the same site, particularly over smooth terrain. Stable atmospheric conditions — common on calm, clear nights — suppress vertical mixing, creating strong wind shear and low turbulence near the surface. Unstable, convective conditions during warm sunny days mix the boundary layer vigorously, reducing wind shear.
Accurate wind maps must therefore represent not just the mean wind speed at a reference height but the full wind profile with height, the variability of wind direction, the frequency distribution of wind speeds, and ideally the turbulence characteristics. The wind speed guide introduces the key concepts of wind shear, wind profiles, and atmospheric stability in accessible terms. Use the Wind Speed Converter to translate between different wind speed units and reference heights.
Methods of Wind Measurement for Mapping
Ground-based meteorological masts (met masts) equipped with anemometers at multiple heights have historically been the gold standard for wind measurement, providing highly accurate, continuous records of wind speed and direction over periods of months to years. However, masts are expensive to install and limited in height to the length of the mast itself — rarely exceeding 100–120 metres even for large projects. As turbines grow taller, the wind at hub height cannot always be directly measured by a conventional mast.
Remote sensing technologies have transformed wind mapping over the past decade. Sodar (sonic detection and ranging) devices emit acoustic pulses upward and measure the Doppler shift in sound scattered by atmospheric turbulence to derive wind speed profiles up to 200 metres or more. Lidar (light detection and ranging) uses laser pulses in a similar way, providing higher precision at greater heights. Both sodar and lidar can profile the full rotor swept zone of modern tall turbines without the cost and complexity of erecting tall masts. Wind measurement instruments covers these technologies in depth.
Satellite observations provide a third source of wind data. Scatterometers and synthetic aperture radar (SAR) instruments aboard satellites measure wind speed over the ocean surface by observing the roughness pattern of the sea caused by wind. This provides global ocean wind coverage and has been crucial for identifying high-quality offshore wind resources in remote areas. Combining satellite data with numerical model output produces the global offshore wind maps that underpin commercial feasibility studies.
- Meteorological masts: highest accuracy; limited height; high installation cost
- Sodar: acoustic profiling; portable; good for lower and middle rotor zone
- Lidar: laser profiling; high precision; can profile full rotor height
- Satellite scatterometry: global ocean coverage; surface wind only
- Satellite SAR: high-resolution ocean surface wind; useful for offshore mapping
Numerical Weather Prediction and Reanalysis Data
Physical measurement networks, however extensive, cannot cover the Earth's surface comprehensively at the resolution needed for wind energy planning. Numerical weather prediction (NWP) models — computer simulations of atmospheric dynamics based on the laws of fluid mechanics and thermodynamics — fill this gap. NWP models divide the atmosphere into a three-dimensional grid and compute how temperature, pressure, humidity, and wind evolve over time by solving the governing equations on powerful supercomputers.
Global reanalysis datasets are created by running NWP models over historical periods, assimilating all available observations — weather station readings, radiosondes, satellite data — to produce a physically consistent, gridded record of atmospheric conditions stretching back decades. Major reanalysis products from national meteorological agencies provide wind speed records at multiple heights covering the world's ocean and land surfaces. These datasets have typical horizontal resolutions of 25–100 km, which is useful for regional wind climate characterisation but too coarse to capture local terrain effects.
Downscaling techniques bridge the gap between coarse reanalysis grids and the site-level resolution needed for project development. Statistical downscaling uses correlations between coarse-grid wind statistics and local measurements to infer fine-scale variability. Dynamical downscaling runs a higher-resolution regional NWP model nested inside a global model, allowing terrain features as small as a few kilometres to influence the simulated wind field. Modern wind atlases combine both approaches. You can explore how this affects project planning with the Wind Potential Checker.
What a Wind Atlas Contains and How to Read It
A wind atlas is a mapped dataset — usually presented as a set of coloured maps with an accompanying numerical database — showing wind resource characteristics across a geographic area. The most fundamental output is mean annual wind speed at a standard height above the ground, typically 50, 80, 100, or 150 metres, representing the range of hub heights in use today. Higher values on the map indicate more energetic wind resources.
Beyond mean wind speed, a complete wind atlas also includes the Weibull distribution parameters that characterise the statistical spread of wind speeds over the year (whether winds are relatively consistent or highly variable), wind direction frequency distributions (wind roses) averaged across grid points, wind power density (the energy content of the wind per unit area, in watts per square metre), and in some products, gross capacity factor estimates for a reference turbine.
Reading a wind atlas requires some care. Values are typically calculated for open, flat terrain at the specified height — actual conditions at a specific site may be quite different because of local terrain, vegetation, or coastal effects not captured in the atlas grid. The atlas is a starting point for site identification, not a substitute for site measurement. The step from atlas-level resource identification to bankable energy yield estimates involves multiple stages of progressively more detailed analysis, as explained in the guide on wind resource assessment.
Major Global and Regional Wind Atlases
Several major wind atlas products are available to researchers and developers, varying in geographic coverage, resolution, methodology, and public availability. The Global Wind Atlas, developed by the Technical University of Denmark (DTU) in partnership with the World Bank, provides free online access to wind resource maps covering the entire world at horizontal resolutions as fine as 250 metres. It combines ERA5 reanalysis data with mesoscale modelling and topographic downscaling and is widely used for preliminary site screening globally.
Regional and national atlases go into greater detail for specific areas. European wind atlases have been developed through large collaborative programmes, combining dense measurement networks with mesoscale modelling to produce validated wind resource maps for the continent. Similar efforts have produced high-quality atlases for North America, East Asia, India, and parts of Africa. Some atlases are freely available as public research outputs; others are proprietary commercial products sold to developers.
Offshore wind atlases require specific treatment because the sea surface, wave state, air-sea temperature differences, and coastal boundary layer effects all influence the wind field over water in ways that differ substantially from land. Dedicated offshore wind atlases incorporate satellite observations and high-resolution ocean-atmosphere simulations. These are essential inputs for the planning of large offshore projects, discussed further in the guide on offshore wind farms and the blog on how offshore wind turbines are installed.
Expert Insight: Machine Learning in Modern Wind Mapping
The application of machine learning to wind resource mapping represents one of the most significant methodological advances in the field in recent years. Traditional physical models solve fluid dynamics equations, which is computationally expensive and limited in resolution by available computing power. Machine learning approaches — particularly deep neural networks trained on large datasets of physical model output and real observations — can learn statistical patterns that bridge scales more efficiently.
One promising application is super-resolution: a neural network trained on pairs of coarse-resolution and fine-resolution wind maps learns to predict the high-resolution structure from a coarse input, producing detailed wind maps for new regions without running a full high-resolution dynamical model. Another application is bias correction: machine learning models can learn systematic differences between NWP model output and real measurements, applying corrections that improve the accuracy of modelled wind estimates across large regions.
These approaches are not a replacement for physical measurement or for physically based modelling — machine learning produces outputs that are only as trustworthy as the training data they are built from, and can produce physically implausible results in conditions outside the training distribution. But combined thoughtfully with physical understanding, they substantially enhance the resolution, accuracy, and coverage of modern wind atlases. The smart wind farms guide explores how machine learning is more broadly transforming wind energy operations and planning.
From Atlas to Site Assessment: The Chain of Analysis
A wind atlas is the starting point in a chain of increasingly detailed analysis that eventually yields a bankable energy yield estimate for a specific project. After atlas screening identifies promising areas, developers conduct site visits to assess terrain, accessibility, grid proximity, and land availability. The most promising sites then proceed to a measurement campaign, where met masts or lidar systems collect on-site wind data for a minimum of one year — and ideally two or more years to capture inter-annual variability.
Measured site wind data are correlated with a long-term reference dataset — typically a reanalysis product — to derive long-term corrected wind statistics that are more representative than the short measurement period alone (a technique called Measure-Correlate-Predict, or MCP). These long-term-corrected statistics are then input to a flow model that accounts for terrain and roughness effects across the specific proposed layout, yielding predicted annual energy production for each turbine position.
Uncertainty quantification is an important final step. Energy yield analyses report not just a single central estimate but a probability distribution reflecting meteorological uncertainty (how much does inter-annual variability affect the estimate?), model uncertainty (how accurately does the flow model represent local conditions?), and measurement uncertainty (how well does the met mast represent the broader site?). These uncertainties feed directly into the financial model, influencing debt financing terms and equity returns. The Energy Production Planner demonstrates how wind resource estimates translate into production projections.
- Atlas screening: identify high-resource areas at regional scale
- Site visit: assess terrain, access, grid, and land constraints
- Measurement campaign: collect on-site wind data for 12+ months
- MCP analysis: correlate site data with long-term reanalysis to derive long-term statistics
- Flow modelling: apply terrain and roughness corrections to the specific layout
- Energy yield estimate: compute predicted annual production with uncertainty bounds
Limitations and Common Pitfalls
Wind atlases and numerical models are powerful tools but carry inherent limitations that must be understood to avoid misuse. The most common pitfall is treating atlas-level wind resource estimates as site-specific values. Atlas grids typically represent horizontally averaged conditions over areas of a square kilometre or more — local terrain features, vegetation, buildings, or coastal effects can create significant deviations from the atlas value at a specific point within that grid cell.
Temporal representativeness is another concern. A wind atlas derived from 20 or 30 years of historical data may not accurately represent future conditions if the regional climate is changing. Research has documented that wind speeds over parts of Europe and North America have varied on decadal timescales by amounts significant enough to affect energy yield estimates. Long-term wind resource planning should account for this variability and for potential climate change effects on wind patterns in coming decades.
Model systematic errors — sometimes called model biases — can cause certain regions or terrain types to be consistently over- or under-estimated in atlas products. Users of wind atlases should always compare atlas values with available observational data from weather stations or measurement campaigns in the target region, and treat significant discrepancies as a signal to investigate further before committing to expensive development activities. The wind resource assessment guide and the guide on wind measurement instruments both address validation and quality control.
Wind Mapping for Offshore and Complex Coastal Environments
Coastal and offshore wind environments present particular challenges for wind mapping because the boundary layer transitions abruptly as air moves from rough land to smooth water. When cold air flows over warmer water — common in spring and summer in mid-latitude regions — convective instability develops, dramatically increasing turbulence and vertical mixing. When warm air flows over cold water, stability is suppressed and the boundary layer can become very shallow with very strong wind shear.
These coastal effects create strong gradients in wind resource quality over distances of just a few kilometres. A site 20 km offshore may have a very different wind climate from a site 5 km from shore, even if both are classified as 'offshore'. Capturing these gradients requires high-resolution mesoscale models run with careful treatment of sea surface temperature and coastal boundary layer dynamics — a computationally demanding and technically demanding undertaking.
As the offshore wind industry matures and moves into deeper waters and more remote locations — including the floating offshore wind projects that are emerging as of the mid-2020s — the need for accurate offshore wind mapping has intensified. The guide on floating offshore wind explains how resource assessment applies to these frontier environments. Blog readers can find additional discussion in the post on how engineers measure the wind, which explains the full range of observation techniques used in modern wind energy science.
| Data Source | Coverage | Typical Resolution | Key Strength | Main Limitation |
|---|---|---|---|---|
| Meteorological mast | Point measurement | Single location | Highest accuracy; direct observation | Expensive; limited height; sparse coverage |
| Ground-based lidar | Point/vertical profile | 10–200 m height profile | Profiles full rotor height; portable | Point measurement; needs calibration |
| Satellite scatterometry | Global ocean surface | 25–50 km | Global coverage; decades of data | Ocean surface only; coarse resolution |
| Satellite SAR | Ocean surface (coverage varies) | 1–10 km | High resolution; near-surface wind | Not continuous; limited to ocean surface |
| Reanalysis models | Global | 25–100 km | Long historical record; physically consistent | Too coarse for terrain effects |
| Mesoscale NWP downscaling | Regional | 1–10 km | Captures terrain effects; physically based | Computationally intensive; model biases possible |
| Machine learning super-resolution | Regional/global | 250 m – 1 km | High resolution at low cost | Depends on training data quality; extrapolation risk |
✅ Key takeaways
- Wind atlases provide essential regional-scale wind resource maps but are a starting point, not a substitute for site-level measurement campaigns.
- Wind power scales with the cube of wind speed, so even small errors in wind resource estimation propagate into large errors in energy yield and project economics.
- Reanalysis datasets provide globally consistent long-term wind records that are combined with downscaling to produce modern high-resolution atlases.
- Machine learning is increasingly used in wind mapping for super-resolution and bias correction, enhancing the detail and accuracy of atlas products.
- Coastal and offshore wind mapping requires specialised modelling of boundary layer transitions because the sea surface is fundamentally different from land.
💡 Interesting fact
Global reanalysis datasets incorporate decades of observations — from weather stations, radiosondes, ships, buoys, aircraft, and satellites — assimilated into physically consistent, gridded atmospheric records used for wind mapping worldwide.
💡 Interesting fact
A 10% increase in mean annual wind speed at a site translates to roughly 33% more extractable wind power, which is why even modest improvements in wind resource estimation accuracy have large economic consequences.
❌ Myth: A wind atlas gives you an accurate prediction of how much energy a specific wind farm will produce.
Reality: Wind atlases represent regional averages and are designed for area screening, not site-specific energy estimation. They operate at horizontal resolutions of kilometres to tens of kilometres and cannot capture the terrain, roughness, and coastal effects that determine actual wind conditions at a specific turbine location. Site-level energy yield estimates require on-site measurement campaigns, long-term correction using reference datasets, and detailed flow modelling — a process that takes months and carries its own uncertainty ranges.
Frequently asked questions
What is a wind atlas?
A wind atlas is a mapped dataset showing wind resource characteristics — typically mean annual wind speed, wind direction frequency, and wind power density — across a geographic area at one or more reference heights. Wind atlases are produced by combining meteorological observations with numerical weather model output and downscaling techniques. They range from global products covering the entire planet at coarse resolution to high-resolution regional atlases focused on specific countries or seas. Most major national and global wind atlases are available to the public for free. Try the Wind Potential Checker as a starting point for site exploration.
What is reanalysis data and why is it used in wind mapping?
Reanalysis data is produced by running a numerical weather prediction model over historical periods and assimilating all available observations — weather stations, balloons, satellites, aircraft reports — to produce a physically consistent, gridded reconstruction of atmospheric conditions. Major reanalysis products provide multi-decade records of wind speed, direction, temperature, and pressure at multiple heights across the globe. This gives wind resource analysts a long historical wind record at any location, even where direct observations are absent, which is essential for estimating long-term average wind speeds and inter-annual variability.
How accurate are wind atlases?
Accuracy varies significantly depending on the atlas, the region, and the terrain complexity. In open, flat terrain, modern atlases can estimate mean annual wind speeds with errors of a few percent relative to measured values. In complex terrain — hills, coastlines, river valleys — errors can be larger because local features are not resolved at atlas resolution. Forested areas are particularly challenging. Offshore, satellite-based atlases are generally reliable for open ocean conditions but less accurate near coasts where boundary layer transitions are strong. Atlas values should always be compared with available local observations before making development decisions.
What is the Measure-Correlate-Predict (MCP) method?
MCP is a statistical technique for extending a short wind measurement record into a long-term estimate. A wind measurement campaign of one or two years at a site is too short to represent the full range of inter-annual wind variability. MCP correlates the short on-site record with a long-term reference dataset (typically a reanalysis product) at a nearby location where both overlap in time, then uses the established relationship to predict what the long-term wind statistics at the site would have been over the full reference period. This corrected long-term estimate is more reliable for energy yield calculations than the raw short measurement period.
Can satellite data map wind over land as well as ocean?
The main satellite instruments used for wind mapping — scatterometers and synthetic aperture radar (SAR) — measure wind speed indirectly by observing sea surface roughness, so they only work over open water. Over land, the relationship between microwave backscatter and wind speed is obscured by vegetation, soil moisture, and terrain. However, satellite data from other instruments — synthetic aperture radar in special modes, GPS radio occultation, and upcoming dedicated lidar missions — are expanding the capability to profile wind over land. For now, over-land wind mapping relies primarily on surface station networks and numerical models rather than satellite observations.
How does climate change affect wind resource mapping?
Climate change is expected to shift atmospheric circulation patterns, which will alter regional wind climates. Research suggests that wind speeds over some regions may decrease while others increase, but projections vary significantly between climate models and scenarios. Wind resource planning over a 25-year project life should account for the possibility of inter-decadal variability and long-term trends, not just historical statistics. Some wind atlas products now include climate scenario outputs showing how wind resources might change under different emission pathways, providing developers with a more complete picture of long-term risk. See also the clean energy trends in 2026 guide.
What is wind power density and how is it shown on maps?
Wind power density (WPD) is the average power available per square metre of area perpendicular to the wind direction, expressed in watts per square metre (W/m²). It incorporates both mean wind speed and the shape of the wind speed frequency distribution (higher WPD for the same mean speed if the distribution has more frequent high-speed events). Wind power density maps are useful for comparing sites because a single number captures more information about the wind resource than mean speed alone. High WPD areas on maps are strong candidates for wind development, while areas below roughly 200–300 W/m² at hub height are generally considered too weak to be economically viable for utility-scale turbines.
How are wind maps used in offshore wind development?
Offshore wind development begins with review of wind atlases showing mean wind speed and power density over the sea at hub heights relevant to offshore turbines, typically 100–150 m. These maps help identify broad zones with the strongest resources and are combined with bathymetric (water depth) maps, shipping lane charts, and environmental protection area maps to screen for viable project areas. Promising zones then proceed to dedicated offshore wind resource assessment using buoy measurements, lidar on floating platforms, and high-resolution offshore modelling. The guide on floating offshore wind and the guide on offshore wind farms provide further context.
What tools are available for visualising wind resources?
Several web-based tools make wind resource data accessible without specialised software. The Global Wind Atlas website provides an interactive map of wind resources worldwide with selectable heights and downloadable data. National meteorological agencies often provide similar tools for their regions. On this site, the Wind Potential Checker offers a simple way to explore wind resource potential at a location, and the Wind Power Estimator translates wind speed estimates into approximate power output figures for educational purposes.
📚 Educational disclaimer
All content is provided for educational purposes only. Technical explanations are simplified for learning and should not replace professional engineering advice or official standards.