Climate forecasting set to improve for wind

WORLDWIDE: Imagine the transformation that could be achieved by the industry if we could answer the question "So exactly how windy is it going to be over the lifetime of my wind asset?".

Mastermind… the Met Office's latest supercomputer has the equivalent power of 30,000 PCs
Mastermind… the Met Office's latest supercomputer has the equivalent power of 30,000 PCs

Google Translate

It would certainly help developers to plan a viable project and to secure financing for it.

Weather and climate modelling systems can offer some insight into these needs. Over the lifetime of wind projects, these numerical similations of the atmosphere can explain past climate variability, present-day risks and opportunities, and future resource.

The Met Office, the UK's national weather service, can already deliver site-specific wind forecasts to cover the next hour to the next week, used by many operators for short-term power predictions. It is now investing in research and development into longer-term future climate variability and change, as well as looking further back into history, to learn more about the long-term behaviour of wind.

Look back

Investigations into using current numerical models to re-analyse historical data stretching back over 140 years is being used to assess the variability in the longer-term wind climate. While recent experience in the UK might suggest some kind of trend - with higher winds in the 1990s and several extremely low-wind years more recently — by using the long-term data records we can see that these fluctuations are fully consistent with the natural variability seen in the past century.

average wind speeds 1871-2010

Looking only at the last two decades would also miss out some of the true variability in the wind speeds that are only apparent over longer periods. By incorporating information from these long records we can start to gain an understanding of what future wind speeds could be over a typical wind asset lifetime.

Already, developers make use of weather and climate research centres that run sophisticated models of the climate at a variety of resolutions. These models can predict across timescales and applications, ranging from "nowcasts" of the weather in the next few hours, which have resolutions of less than 2 kilometres, to multi-week ensemble forecasts, which cover resolutions of tens of kilometres. Other models look back and re-analyse the atmosphere through data that can stretch as far back as 140 years. These modelling systems have become more complex and capable as the supercomputing technology has improved.

Just three years ago, wind resource assessment of a wind project solely used data taken from a national network of 10-metre-high masts with anemometers, such as those that the Met Office uses to monitor the UK's climate.

While this observational data continues to play an important role, assessment is now greatly enhanced with the more frequent use of mesoscale modelling, particularly as observational data is difficult to acquire in many parts of the world.

Using weather modelling software already contributes to assessing wind resources in several ways. The flexibility in set-up means that an early assessment can be provided before money is spent on site surveys, measurement campaigns and site planning and design studies. Modelling can provide a long-term reference for measure-correlate-predict wind assessments using on-site measurements, particularly where long-term weather measurements are not available. It can supply data for a small project when an on-site measurement campaign is not economically viable, and traditional wind maps are not accurate enough.

As energy production, and therefore revenue generation, is highly sensitive to long-term wind resources, all stakeholders have much to gain from improving the accuracy and confidence of long-term wind assessments. New technologies and advances in science are assisting with this challenge. There is even greater potential if weather and climate modelling are included in the resource assessment mix.

It is possible to use very long-term historical re-analysis modelling to quantify climate variability and present-day risks. A major area of research at the Met Office is forecasting across a season or a decade. Although this is work in progress, leading scientists believe the next generation of models will be able to provide highly valuable forward projection, adding real value to wind resource assessments.

Science behind the solution

Increasingly, current numerical models are being used alongside on-site data, such as from a meteorological (met) mast or lidar light detection system, or from a met mast near the site. The ability to map current data against the historical data is particularly useful in locations where there is little observational data to provide the wind climatology information required as part of the resource assessment process. These data sets are low-resolution and designed to present a coherent picture of the larger-scale weather, so are not ideal to use in isolation to estimate asset-specific wind resource due to the highly siteand height-specific nature of wind.

Higher-resolution numerical weather prediction (NWP) model data, which currently covers much shorter timescales, can produce siteand height-specific resource assessments or forecasts of wind. For offshore energy projects, these can be used in conjunction with wave model data. The Met Office applies a downscaling process to its suite of high-resolution NWP models to produce resource assessments and forecasts that take into account the detailed characteristics of the site, including orography and land use, which are fundamental in determining the wind speeds at specific heights above the surface.

The Met Office wind prediction tool combines NWP models, re-analysis data sets and downscaling to capture and analyse all the wind information needed to define an accurate long-term wind climatology at a specific site and hub height. This improves the accuracy of the long-term mean wind speed and, as improved historical data sets cover longer periods and modelling capabilities improve, predictions will become more accurate, enabling more effective management of the impacts of weather at all stages of a project lifecycle.

Using NWP model data in this way also enables the uncertainty of resource estimates or forecasts to be quantified. Wind speed biases at a site are governed by the underlying skill of the model and the difficulty in modelling the site, whether through terrain and land use complexity, or local features not represented in the model. The accuracy of the wind data generated by combining these analyses has been extensively verified and the certainty in the wind data is defined according to the topographic complexity of individual sites.

The Met Office is researching the best way to combine on-site observations with model reference data taken at the same site, but at a different height, for example. Engagement across the industry will help to build understanding.

Reducing uncertainties

Traditional techniques may not be ideal as, like daily meteorological forecasts, there is a risk that constraining short periods of monitoring for co-located model reference data set risks removing significant climatological intelligence not observed in the monitoring period. To resolve this issue, a weighted algorithm must be added to the calculations, to combine the longer-term prediction data. This maintains the climatological signal contained in the modelled site wind record but incorporates the effects of local features observed at the site.

Given the high variability of winds across Europe it is crucial that advanced wind assessment methods are used to accurately define the long-term wind climatology, which in turn underpins long-term power yields and revenue projections. As development costs rise, the sensitivity of long-term profitability to wind speeds is increasingly important.

Experimental model designs that will generate future projections of wind speeds, on timescales ranging from months to decades, are being investigated. These are factoring in the needs of the global renewables industry — to ensure wind is a primary output, not a poor relation as it may have been in the past. As scientists progressively improve the skill of such models, it may be possible to factor in long-range forecast intelligence in wind-farm investment decision, rather than relying solely on the past to predict the future.

Rob Harrison is head of renewables at the Met Office

Have you registered with us yet?

Register now to enjoy more articles
and free email bulletins.

Sign up now
Already registered?
Sign in