Wind Resource Assessment

Before a single turbine is ordered or a foundation designed, developers must answer a deceptively simple question: how much wind is actually at this site? The answer — reached through a systematic process called wind resource assessment — determines whether a project is financially viable, which turbines to select, how to arrange them, and what the bank financing terms will be. Get the assessment wrong and the entire project is built on faulty numbers.

Wind resource assessment draws on a combination of satellite data, numerical weather models, historical climate records, and crucially, direct on-site measurements. It is partly atmospheric science, partly statistics, and partly engineering judgement, and it is one of the most important technical activities in the wind energy development process. Major lending institutions require an independent assessment before they will finance a project.

This guide explains the full assessment process from start to finish: why it matters, what data sources are used, how instruments are deployed, how wind data is analysed statistically, how energy yield is estimated, and what the sources of uncertainty are. Understanding this process is essential for anyone studying wind energy development, evaluating project documents, or simply wanting to know how confident developers can be in their predictions.

Why Wind Resource Assessment Is Non-Negotiable

The financial case for a wind project depends almost entirely on how much electricity it will generate over its 25-year lifetime. Because wind power scales with the cube of wind speed, even a modest error in the estimated average wind speed can translate into a very large error in projected energy output — and therefore in projected revenue, debt serviceability, and equity return. A 10 % overestimate of annual energy production can turn a viable project into a loss-making one.

Banks and institutional investors who provide debt and equity financing to wind projects understand this dependency very well. Before committing funds, lenders routinely commission independent technical assessments — performed by specialist consultancies — that independently verify the developer's wind resource and energy yield estimates. These independent assessments typically apply conservative assumptions and characterise the uncertainty in the projections explicitly, allowing lenders to size loans to levels that can be serviced even in low-wind scenarios.

The consequences of inadequate resource assessment are not theoretical. Projects built on optimistic or poorly characterised wind resources have consistently underperformed their projections, with serious financial consequences. Conversely, careful, rigorous assessment that accurately characterises the wind resource — including its variability and uncertainty — gives projects the best chance of meeting or exceeding performance expectations.

Beyond financing, accurate resource assessment informs turbine selection, micrositing — the precise positioning of each turbine within the site — and wind farm layout optimisation. All of these decisions have significant impacts on lifetime energy production. For more on how layout decisions interact with the wind resource, see the guide to Wind Farm Layout.

Preliminary Screening: Identifying Candidate Sites

Resource assessment begins long before any instrument is deployed. The first step is desktop screening: using publicly available data to identify geographic areas with promising wind resources. Wind atlases — databases of modelled or measured wind speeds at standardised heights, covering large geographic areas — are the primary tool at this stage. Many countries and international agencies have published wind atlases based on numerical weather model outputs or satellite data.

At the global scale, reanalysis datasets — computationally intensive retrospective reconstructions of the atmosphere using weather model simulations constrained by historical observations — provide multi-decade wind speed records at any location. These datasets are coarser than measurements at a specific site, but they capture long-term climate patterns and can identify consistently windy regions for initial screening.

Once candidate areas are identified, developers overlay additional constraints: land ownership and availability, environmental designations, proximity to transmission infrastructure, visual and acoustic impact zones from settlements, aviation and radar constraints, and other planning considerations. This layered screening process rapidly narrows a broad geographic area down to a handful of specific sites that merit detailed investigation. The guide to Wind Mapping and Wind Atlases explains the data sources used in this stage.

Preliminary screening is inherently approximate — wind atlas data may not capture local terrain effects, and the resolution is too coarse to characterise specific sites precisely. But it is fast and cheap, allowing developers to focus expensive measurement campaigns on the most promising locations rather than wasting resources on clearly unsuitable sites.

• Global and national wind atlases for initial resource screening
• Reanalysis datasets for long-term climate patterns
• GIS overlays for constraints: ownership, environment, grid, planning
• Terrain and roughness analysis for expected local wind enhancement or reduction
• Shortlisting candidate sites for detailed measurement

Meteorological Mast Campaigns: The Gold Standard

Once a candidate site is identified, the traditional and most highly regarded method for characterising the wind resource is to erect a meteorological mast (met mast) — a lattice or tubular tower instrumented with anemometers, wind vanes, temperature and pressure sensors at multiple heights. Data from a well-sited met mast, collected over at least one full year, provides the most direct and bankable measurement of the wind resource at a specific location.

Standard met mast campaigns instrument the tower at several heights — typically including the expected hub height of the planned turbines and multiple lower reference levels. Multiple anemometers at the same height (redundancy) and at different heights (profiling) allow engineers to determine the wind shear exponent at the site. Wind vanes record the direction of wind from each compass sector, which is critical for understanding the prevailing wind direction and for micrositing turbines.

Sensors log data continuously at 10-minute averages, the standard interval in the industry. These 10-minute means — typically more than 50,000 per year per sensor — are then quality-checked, with obviously erroneous readings flagged and filtered. Gaps in the record from sensor failures, icing events, or power outages are noted and quantified, as the percentage of valid data ('data recovery rate') is itself an important quality indicator.

Met mast data is the foundation on which all subsequent analysis is built. However, masts have limitations: they are expensive to erect and maintain, can only measure wind at one horizontal location, and require planning permission and sometimes aviation lighting in their own right. These limitations have driven the adoption of remote sensing alternatives, discussed in the next section.

Remote Sensing: LIDAR and SODAR

Remote sensing instruments measure wind speed and direction at multiple heights without a tall mast. Two principal technologies are used in wind resource assessment: LIDAR (Light Detection and Ranging) and SODAR (Sound Detection and Ranging). Both are now widely accepted by lenders and independent assessors, and their use has grown rapidly as costs have fallen and confidence in their accuracy has increased.

LIDAR instruments emit laser pulses into the atmosphere and measure the Doppler shift in the light reflected back by aerosol particles — tiny natural particles like pollen, dust, and sea salt that are always present in the atmosphere. The magnitude of this shift is proportional to the particle velocity, which approximates the wind velocity. Modern scanning LIDAR systems can measure wind speed profiles from near the ground up to 200 m or more with high temporal resolution.

SODAR operates on the same Doppler principle but uses sound pulses rather than light. It emits acoustic signals and measures the frequency shift of echoes scattered from atmospheric turbulence structures. SODAR is generally limited to lower heights than LIDAR and can be affected by ambient noise, but is lower cost and portable. Both LIDAR and SODAR can be deployed on the ground, on floating offshore buoys, or on temporary platforms — making them much more flexible than masts, especially offshore.

The primary advantage of remote sensing is that a single instrument at ground level can simultaneously profile wind at all relevant heights — from below potential hub height to above it — enabling the wind shear profile to be characterised without multiple sensors at different mast levels. This is especially valuable for sites considering tall-hub turbines where a conventional mast would need to be impractically tall to reach hub height. More background on these instruments is in the guide to Wind Measurement Instruments.

• LIDAR: laser-based; profiles wind up to 200 m+; high accuracy
• SODAR: acoustic; lower cost; limited height and accuracy in noisy environments
• Floating LIDAR buoys: deployable offshore without fixed infrastructure
• Remote sensing reduces mast costs while maintaining acceptable data quality
• Both technologies are increasingly accepted by project lenders

Measure-Correlate-Predict: Extrapolating Short Records

Even a well-sited met mast measuring for a year captures only a snapshot of the wind climate. One year may be windier or calmer than the long-term average — a single year's measurement may deviate from the long-term mean by 5–10 % in either direction. Making a confident prediction about wind conditions over a 25-year project lifetime requires anchoring short measurement records to long-term reference datasets. This is the purpose of the Measure-Correlate-Predict (MCP) technique.

MCP works by identifying a reference dataset — typically a long-running nearby weather station (often 10–30 years of records) or a reanalysis dataset — and statistically correlating the site measurements with the reference over the period of concurrent overlap. If the site measurements and the reference data are well correlated (they tend to be windier or calmer at the same time), the statistical relationship can be used to 'translate' the long reference record into an equivalent prediction for the site, effectively extending the site record back in time.

The result is a long-term estimate of the site's annual mean wind speed, expressed with quantified uncertainty. This long-term estimate is the key input to energy yield calculations. Good reference datasets with long records and close proximity to the site produce tighter (less uncertain) MCP estimates. Poor-quality references or low correlations result in wider uncertainty bands — which translates into more conservative project financing terms.

MCP assumes that the statistical relationship between the site and the reference remains stable over time — a reasonable assumption in most cases, though changes in land cover (new forests, buildings) or unusual climate variability can introduce errors. Practitioners typically use multiple reference datasets and MCP methods, comparing results to build confidence in the long-term estimate.

Wind Flow Modelling and Micrositing

Measurements at one or a few locations on a site do not tell the developer what the wind is like at every point across the development area, which may cover several square kilometres. Wind flow modelling fills this gap by using numerical models to simulate how the wind resource varies across the terrain, accounting for hills, ridges, valleys, forests, and other features that deflect, accelerate, or slow the wind.

The most widely used tool for onshore wind resource modelling is WAsP (Wind Atlas Analysis and Application Program), developed by the Technical University of Denmark. WAsP uses a linearised flow model that accounts for terrain shape and surface roughness to predict wind conditions at any point, based on reference measurements at a nearby location. It is computationally fast and well-validated, though its linearisation assumptions break down in complex terrain — steep hills, large roughness changes — where more sophisticated computational fluid dynamics (CFD) models are sometimes applied.

CFD models solve the Navier-Stokes equations governing fluid motion numerically, capturing flow effects that simpler linear models cannot represent: recirculation in the lee of steep slopes, wake effects in complex terrain, and large roughness transitions. They are computationally expensive and require skilled practitioners to set up and validate, but increasingly powerful computing resources are making them more accessible.

The outputs of wind flow modelling — predicted mean wind speed, wind shear, and turbulence intensity at each proposed turbine location — are the inputs to micrositing: the optimisation of turbine positions to maximise total farm energy output while staying within turbine structural design limits for loads and turbulence. Micrositing is an iterative process that balances energy yield against foundation cost, cable cost, and structural integrity across the turbine fleet.

Energy Yield Estimation and Uncertainty Quantification

Once the wind resource at each turbine location is characterised, the energy yield calculation brings together the wind speed distribution and the turbine's power curve — the manufacturer-specified relationship between hub-height wind speed and power output. Integrating the power curve over the wind speed distribution at each turbine gives the gross energy yield: the energy that would be produced if the turbine operated in isolation with no losses.

From the gross yield, engineers subtract a series of losses to arrive at the net energy yield — the quantity that lenders and investors actually care about. Wake losses (the reduction in wind speed experienced by turbines downwind of others in the array), electrical losses in the cables and transformers, turbine availability (accounting for planned and unplanned downtime), and environmental losses (blade icing, soiling, curtailment for noise or wildlife) all reduce net output below the gross theoretical figure.

Wake losses within a wind farm can be substantial — typically 5–15 % of gross energy depending on turbine spacing, prevailing wind direction, and the number of turbines in rows aligned with the wind. Wake modelling has advanced considerably, with validated engineering models widely used in industry and more sophisticated CFD and large-eddy simulation approaches used for research and for complex arrays. You can explore turbine efficiency concepts further in the guide to Turbine Efficiency and the Betz Limit.

Uncertainty quantification is a critical output of the assessment. Rather than presenting a single best-estimate energy yield, professional assessments express yield at multiple probability levels — typically P50 (the median: a 50 % probability of exceedance), P75, and P90 (a 90 % probability of exceedance, i.e. the conservative estimate). Lenders typically size debt serviceability around P90 estimates to protect against downside scenarios.

• Gross yield: theoretical energy from turbine power curve × wind distribution
• Wake losses: typically 5–15 % deduction depending on farm layout and spacing
• Electrical losses: resistive losses in cables and transformers, typically 1–3 %
• Availability losses: planned and unplanned downtime deductions
• Environmental losses: icing, soiling, noise/wildlife curtailment
• P50/P75/P90: probability-of-exceedance levels used in financial modelling

Expert Insight: Understanding Uncertainty in Resource Assessment

A common misconception about wind resource assessment is that it produces a definitive, precise answer about how much energy a wind farm will generate. In reality, even the most rigorous assessment produces a range of probable outcomes, with stated confidence intervals. Understanding where uncertainty comes from — and how it is managed — is crucial for interpreting assessment reports and structuring project financing.

Uncertainty in energy yield estimates arises from multiple sources. Historical wind variability means that the next 25 years of wind may differ from the reference period used for long-term estimation. Measurement uncertainty in the instruments themselves contributes a small but non-negligible error. Model uncertainty — errors introduced by the approximations in wind flow models — can be significant in complex terrain. Wake model uncertainty is often one of the largest contributors, particularly for large dense arrays. And turbine performance uncertainty relates to whether the turbine's actual power curve will match the manufacturer's specification.

These individual uncertainties are combined (technically, added in quadrature if they are independent, or more carefully if correlated) to produce an overall uncertainty band on the P50 estimate. A well-characterised site with multiple measurement instruments, a long measurement campaign, and good reference data may have a total uncertainty of 5–8 % on the P50 energy yield. A poorly characterised site with a short measurement campaign and limited reference data might carry 15 % or more uncertainty — a range large enough to make financing very difficult.

Developers reduce uncertainty by measuring for longer periods, using redundant instruments, selecting better reference datasets, applying validated flow models carefully, and commissioning independent reviews. Each investment in reducing uncertainty has a direct financial payoff: tighter uncertainty bands allow more aggressive financing, which reduces the weighted average cost of capital and improves project returns. This is why rigorous resource assessment is not a cost centre but a value-creation activity.

Offshore Resource Assessment: Additional Challenges

Offshore resource assessment shares the principles of onshore assessment but adds significant logistical and technical challenges. Deploying instruments at sea requires specialised platforms: fixed met masts on monopile or jacket foundations (expensive and requiring permits), floating LIDAR buoys, or instruments mounted on existing offshore structures like oil platforms or lighthouses.

The offshore wind resource is generally better characterised at large scales — satellite scatterometry can provide wind speed estimates over the ocean — but at the local level, offshore assessments have historically had less data than onshore campaigns, partly because of the cost and difficulty of offshore measurement. Floating LIDAR technology has substantially improved this situation by providing hub-height data at modest cost without fixed infrastructure.

Offshore environments also introduce met-ocean parameters — wave height, current, and storm characteristics — that must be measured or modelled for foundation design, marine operations planning, and cable route assessment. These additional data requirements mean offshore resource assessment campaigns are more complex and expensive than onshore equivalents, often extending to two or more years to characterise seasonal variability adequately.

The strong offshore wind resource — consistently higher speeds than most onshore locations — generally justifies the additional assessment cost. The guide to Offshore Wind Farms covers how the resource assessment feeds into the full project development process at sea.

From Assessment to Project: Using the Results

A completed resource assessment is not the end of the process; it is the foundation on which the rest of project development is built. The energy yield estimate determines whether the project is financially viable at realistic electricity prices. The wind speed distribution and turbulence characterisation at each turbine position determine which turbine class — defined by international standards — is appropriate for the site conditions.

Turbine manufacturers specify their products for different wind regimes: higher wind-speed sites require more structurally robust turbines that can handle greater loads. Installing a turbine designed for a moderate-wind site in a high-wind or high-turbulence environment can lead to premature fatigue failure. Conversely, oversizing the turbine for a calm site wastes money on unnecessary structural capacity. Matching turbine class to site conditions is a direct output of the resource assessment.

The assessment also underpins the micrositing optimisation — the precise placement of turbines on the site map to balance energy capture, wake interactions, noise and visual constraints, access road costs, and foundation conditions. The Wind Farm Planner tool can help visualise how spacing and layout choices interact with the wind resource in educational scenarios.

Finally, the assessment is a key document in the planning and permitting process. Environmental impact assessments draw on the resource assessment's wind speed and turbulence data to model noise propagation, shadow flicker patterns, and — for offshore projects — wake effects on downstream sites. The guide to Wind Farm Planning and Permitting explains how these assessments feed into the regulatory process.

Frequently asked questions