Current State of Wind Energy Prediction
Relevant Expertise in Wind Energy Prediction
Wind Power Prediction
Boundary-Layer Processes That Affect Wind-Energy Production
Why are Accurate Wind Energy Predictions Important?
Image of NCAR's Auto–Nowcast System. This system provides short–term predictions (0–2 hours) of the local surface weather conditions.
Risks associated with energy supply and demand revolve around market dynamics, financial conditions, politics, technology choice, fuels, environmental quality, and weather. A major challenge in charting a course to stable, secure energy services is "seeing the whole" – understanding the complex interactions and magnitudes of risk factors across the range of energy supply, demand, and price issues that are often unique to specific cities and regions.
The recent policy trend to require a larger fraction of the energy portfolio devoted to renewable energy sources such as wind and solar puts additional strain on the energy industry as these have to date been less predictable than traditional generation sources.
The influence of significant weather events on the energy industry has increased with diminishing reserve margins to meet peak loads. In the Western U.S., weather factors have not only included unusually hot summer and cold winter events, but also low precipitation that reduces hydropower capacity and increases electrical demand for irrigation pumping and groundwater withdrawal.
Improved weather prediction and precise spatial analysis of mesoscale weather events are crucial to both short–and long–term energy management. There is a need to further develop and implement advanced weather observation and prediction technologies for the energy industry that benefit both public and private sectors.
Weather data and information are crucial to infrastructure planning and management, prediction of energy demand, management of energy supply, energy pricing and markets, energy system operations and regulatory compliance, and economic risk minimization. The energy industry has expressed a need for:
- Improved weather forecast accuracy
- Improved resolution of forecast information and diagnostic data in time and space (e.g., hourly heating and cooling degree day calculations, hourly and shorter term wind and solar predictions)
- A better understanding of the uncertainty that is inherent in the forecasts
- Increased attention to understanding customer requirements
- Improved understanding of user capabilities to utilize information for specific applications
- Understanding of barriers to the use of improved information in decision making
- Additional weather observations at strategic locations as determined by quantitative methods
Improved weather information facilitates improved decision making by the energy industry. Tailored weather decision support products support:
- System load forecasting
- Increased efficiency in pricing for hourly and bulk markets, weather–indexed energy commodity transactions, and energy commodity trading
- Precision management of demand in the local distribution system
- Analysis of potential environmental and societal impacts related to both short- and long-term supply strategies
- Improved anticipatory response management planning for high impact weather
- Improved air quality and greenhouse gas management strategies
- Increased efficiency in wind, solar, and hydropower energy generation
- Improved efficiency of coal and petroleum product distribution
- Increased efficiency of gas pipeline operations
- Renewable energy planning, development, and operations
Detailed knowledge of the current and future weather and climate conditions is critical for minimizing the impact of significant weather events on energy operations (e.g., unanticipated drops or rises in wind speeds impacting wind energy production, periods of unanticipated temperature extremes, severe weather, etc.). On a day–to–day basis, better prediction of wind speeds in time spans of a few minutes to a few hours would be helpful for gauging the output of wind energy facilities. These points were made very clear during an electrical energy workshop titled "Increasing the Value of Weather Information in the Operation of the Electrical Power System" held at the National Center for Atmospheric Research (NCAR) in November 2003. The workshop focused on the needs for improved weather information for the electrical power industry and was attended by energy officials and weather experts representing numerous interests including electrical power generation, transmission, wind energy, load forecasters, energy research institutes, public utilities commission, Department of Energy (DOE), DOE laboratories, Department of Defense (DOD), private sector weather companies, universities, NCAR, and the National Oceanic and Atmospheric Administration (NOAA).
Back to top
Current State of Wind Energy Prediction
Over the last decade, awareness of the need for improved weather services for the renewable energy industry has increased significantly. A coordinated national effort highlighted the unmet weather needs of the renewable energy industry and helped to identify research requirements and implementation strategies for such improved weather services. As a result, the atmospheric science community (public, private, and academic sectors) is now actively engaged with the renewable energy industry. This partnership will help accelerate technological improvements designed to benefit the renewable energy industry. What is required – and is often missing from the research community – is the ability to move weather technologies from the workbench to operations. NCAR has extensive technology transfer experience and prides itself on its ability to work cooperatively with industry to develop and deliver operational decision support technologies that will advance the state–of–the–art in wind energy prediction and provide economic benefits to the industry.

Comparison of nacelle wind speeds vs. wind speeds generated from the WRF model at various model grid spacing configurations. Approximately 275 wind turbines are plotted. The finer grid spacing of theh model more closely represents the wind speed variance at the wind farm.
One of the biggest limitations of weather prediction is obtaining information on the current state of the atmosphere. The lack of a dense worldwide observation network means that critical details about atmospheric properties (e.g., fronts, moisture, temperature gradients, wind, etc.) can be lost when the current state of the atmosphere is analyzed to create the initial state for weather models. If the analyzed initial state is not a true representation of the atmosphere, then forecast accuracy will suffer. Weather models can only utilize atmospheric state variables; that is weather models can only be initialized with temperature, winds, water vapor, and pressure observations.
In the last ten years, the research community, and particularly NCAR, has focused significant resources on data assimilation research. Data assimilation is a process whereby disparate observations are ingested, quality controlled, translated to atmospheric state variables, and objectively analyzed to produce a physically balanced state of the atmosphere on global, regional, and local scales. Data assimilation is proving to be a critical component of the weather forecast process as it provides the ability for weather prediction models to utilize new data sets including Doppler weather radar, satellite data fields, lidar, measurements obtained from regional aircraft, local sensor data including meteorological towers, and others. Through a variety of research grants from mostly military sponsors, NCAR has arguably become a world leader in data assimilation technologies, particularly targeted to the regional and local scales.
NCAR also has significant experience in developing statistical post–processing systems that optimize wind and wind energy forecasts. Given weather forecast information and wind farm output data, optimized wind energy forecasts can be generated.
NCAR is also a world leader in automated nowcasting technologies. Local in situ and remote weather observations are combined to diagnose the local wind environment and short–term modeling techniques are applied to predict the surface weather conditions. Knowledge of the current and predicted local wind conditions can be translated into short–term wind energy predictions when wind facility data are included in the calculations.
Advanced and unique data assimilation, nowcasting, and statistical post–processing technologies are key ingredients of advanced wind energy prediction systems.
Back to top
Relevant Expertise in Wind Energy Prediction
NCAR is a world–renowned atmospheric science research and development center. Its Research Applications Laboratory (RAL) specializes in applied research and technology transfer to mission agencies and sponsors. NCAR/RAL has successfully developed and transferred to operations weather decision support technologies to the aviation community (e.g., airlines, Federal Aviation Administration (FAA)), National Weather Service (NWS), international governments (Taiwan, Hong Kong, Korea, Australia, United Arab Emirates, Singapore, and others), private sector companies (e.g, Xcel Energy, The Weather Channel, and Telvent/Meteorlogix), Army, Air Force, Defense Threat Reduction Agency (DTRA), Pentagon Force Protection, National Ground Intelligence Center (NGIC), Department of Homeland Security (DHS), Department of Transportation (DOT), National Aeronautics and Space Administration (NASA), and other clients.
NCAR has played a key role in advancing weather capabilities of the NWS and private sector weather companies. As a national center, NCAR is able to utilize advancements developed not only at NCAR, but at research centers, institutes, universities, and national laboratories worldwide. NCAR is leading an international effort on the development of the next generation numerical weather prediction model called the Weather Research and Forecasting (WRF) model. This state–of the–art model is designed to be efficient, flexible, and extensible. It contains advanced physics packages and has the ability to be configured as a global model or as a very high–resolution local model with grid spacing down to 500 meters. A large eddy simulation (LES) model version is capable of making predictions on scales of 10s of meters.
NCAR is also leading a large effort on data assimilation and has developed the WRF 3–Dimensional Variational Data Assimilation (3DVAR) system and a continuous Real–Time 4–Dimensional Data Assimilation (RTFDDA) system. NCAR has also ported its "observation-nudging" assimilation technique to the WRF model, which is considered the best method of incorporating local observations into a model that is running at high–resolution, such as the type that is being proposed herein for Iberdrola Renewables. The WRF RTFDDA, which is a continuously cycling data assimilation system built onto WRF, allows for smooth and uninterrupted assimilation of new observations between forecast cycles; i.e., with no cold–starts that incur model spin–up effects that are simply discarded by forecasters because of the inherent inaccuracies, which usually last for the first 6–12 hours of each forecast, depending on the state of the atmosphere.
NCAR also has considerable experience in the development of environmental analysis and prediction systems, plume dispersion, hydrology, hydraulics, nowcasting, and fire behavior models, which require detailed predictions of the surface or near surface conditions. Most of the efforts have involved the design, development, and deployment of operational systems for private and public–sector organizations. The following examples are representative, and illustrate NCAR's capabilities to deliver long–and short–range weather prediction decision support capabilities.
- NCAR is one of the lead organizations developing the WRF Model and has a lead role for distributing the code to the community. WRF features multiple dynamical cores, a growing suite of data assimilation systems, and a software architecture allowing for computational parallelism and system extensibility. WRF is suitable for a broad spectrum of applications across scales ranging from meters to thousands of kilometers. The team that will evaluate the data assimilation capabilities for Iberdrola Renewables also participates in WRF development, and thus has an intimate knowledge on how to best apply and tune the system for Iberdrola Renewables' needs.
- NCAR has developed an advanced weather prediction system for the U.S. Army Test and Evaluation Command called the WRF RTFDDA (Real–Time Four–Dimensional Data Assimilation), which calculates fine–scale features from the atmospheric boundary layer through the upper troposphere. The WRF RTFDDA system has supported mission–critical operations at seven US Army test ranges since 1996, and has been used in guiding decision–makers in Operations Enduring Freedom, Iraqi Freedom, and firefighter support during major wildfire events, among many other missions. Similarly, estimates of the wind and temperature climatology, for any location in the world are generated for the National Ground Intelligence Center (NGIC). The WRF RTFDDA is also the technology chosen by Xcel Energy for their advanced wind energy prediction system.
- At the U.S. Army's Dugway Proving Ground in Utah, personnel release aerosols into the atmosphere to validate models that calculate plume dispersion. NCAR has developed meteorological models that provide detailed wind and other weather parameters to DoD plume models. Because of the importance of the boundary layer winds at Dugway and at other ranges, a tremendous amount of effort has gone into improving the model's handling of surface winds and post–processing the model's solution, which has recently included the implementation of a 3–D gridded bias correction technique.
- NCAR is a world leader in advanced mathematics and statistics. NCAR's extensive experience in areas such as turbulence, spatial statistics, Bayesian statistics and Kalman filters can be leveraged for wind energy prediction.
- UCAR's Auto–Nowcaster system combines several data sources to generate short–term predictions of surface conditions and information about thunderstorm initiation, growth, and decay. Combined datasets include radar, satellite, rawinsones, surface observations, and weather prediction model data. Lidar data also contribute where it is available. These technologies provide an excellent framework for a wind energy nowcasting capability.
- UCAR is a leader in the development of intelligent weather prediction systems. When The Weather Channel™ needed to expand and automate its digital operations to cover the globe, it sought out NCAR to develop a system for such a purpose. The Dynamic, Integrated Weather Forecast System (DICast™) was developed by UCAR and implemented by The Weather Channel. The system provides an optimized, site–specific forecast for user–defined locations. Derivatives of this technology are now being used by other large private sector weather service providers such as Telvent/Meteorlogix.
Mathematics and statistics are key components in modern weather and climate research. Some examples of these resources and projects currently in place within NCAR/RAL include the following.
- Forecast Verification – New forecast weather models need to be evaluated under representative conditions using fair and meaningful performance measures. As forecasts have evolved from describing a specific condition at a point in space to forecasting a range of possible conditions for a large gridded space, verifying forecasts has become much more complicated. Some challenges in forecast verification include developing feature–based verification techniques that describe the shape and characteristics of forecasts, developing fair skill scores that efficiently describe and compare model performance and new performance evaluation approaches that permit the comparison of forecasts produced at different scales. NCAR/RAL has the leading forecast verification research and development program in the U.S. (and perhaps internationally), and has been at the forefront of many of the new developments in this field over the last decade.
- Statistical Post–processing – NCAR has extensive experience in applying statistical post–processing techniques to improve the skill and usefulness of weather forecasts. Methods include fuzzy logic and expert systems, in addition to statistical models such as neural networks, regression trees, and generalized linear models. These methods are used to combine model output with other sources of information to provide forecasts for particular variables at a specific location, and to create probabilistic information. Model error information from deterministic models has also been used to help create probabilistic forecasts. Various model calibration techniques have been used to relate the spread from ensemble forecasts to the distribution of observed conditions.
- Spatial Statistics – The Geophysical Statistics Project is a National Science Foundation–supported, NCAR–based group that pursues the innovative development and application of statistical methodologies to address problems faced in the Earth sciences. Research by this group includes extending statistical methodology to address spatial and space/time problems as well as the application and development of Bayesian hierarchical models – models that can combine scientific information with new data.
- Experimental Design – Research experiments can be costly in terms of resources and computational effort. To help ensure projects reach robust conclusions, the statistical process of experimental design is used to help plan projects and subsequent data analysis.
Virtually all of the research and development efforts performed at NCAR's Research Applications Laboratory (RAL) are designed to provide reliable operational solutions that meet the particular needs of the sponsor. Decision support systems have been developed for aviation safety, national security, surface transportation, defense, urban flood districts, severe weather, energy, and critical infrastructure protection. Thus, NCAR/RAL has extensive experience developing mission–oriented systems that must reliably and competently meet the operational needs of public–and private–sector organizations.
Back to top