DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: The importance of round-robin validation when assessing machine-learning-based vertical extrapolation of wind speeds

Journal Article · · Wind Energy Science (Online)

The extrapolation of wind speeds measured at a meteorological mast to wind turbine rotor heights is a key component in a bankable wind farm energy assessment and a significant source of uncertainty. Industry-standard methods for extrapolation include the power-law and logarithmic profiles. The emergence of machine-learning applications in wind energy has led to several studies demonstrating substantial improvements in vertical extrapolation accuracy in machine-learning methods over these conventional power-law and logarithmic profile methods. In all cases, these studies assess relative model performance at a measurement site where, critically, the machine-learning algorithm requires knowledge of the rotor-height wind speeds in order to train the model. This prior knowledge provides fundamental advantages to the site-specific machine-learning model over the power-law and log profiles, which, by contrast, are not highly tuned to rotor-height measurements but rather can generalize to any site. Furthermore, there is no practical benefit in applying a machine-learning model at a site where winds at the heights relevant for wind energy production are known; rather, its performance at nearby locations (i.e., across a wind farm site) without rotor-height measurements is of most practical interest. To more fairly and practically compare machine-learning-based extrapolation to standard approaches, we implemented a round-robin extrapolation model comparison, in which a random-forest machine-learning model is trained and evaluated at different sites and then compared against the power-law and logarithmic profiles. We consider 20 months of lidar and sonic anemometer data collected at four sites between 50 and 100 km apart in the central United States. We find that the random forest outperforms the standard extrapolation approaches, especially when incorporating surface measurements as inputs to include the influence of atmospheric stability. When compared at a single site (the traditional comparison approach), the machine-learning improvement in mean absolute error was 28 % and 23 % over the power-law and logarithmic profiles, respectively. Using the round-robin approach proposed here, this improvement drops to 20 % and 14 %, respectively. These latter values better represent practical model performance, and we conclude that round-robin validation should be the standard for machine-learning-based wind speed extrapolation methods.

Research Organization:
National Renewable Energy Laboratory (NREL), Golden, CO (United States)
Sponsoring Organization:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Wind Energy Technologies Office
Grant/Contract Number:
AC36-08GO28308
OSTI ID:
1677497
Report Number(s):
NREL/JA--5000-78109; MainId:32018; UUID:9f734336-6146-4cfb-ae2c-7b470e3c05f6; MainAdminID:18706
Journal Information:
Wind Energy Science (Online), Journal Name: Wind Energy Science (Online) Journal Issue: 2 Vol. 5; ISSN 2366-7451
Publisher:
European Wind Energy Association - CopernicusCopyright Statement
Country of Publication:
United States
Language:
English

References (34)

Difficulties in using power laws for wind energy assessment journal January 1983
ARM: Eddy Correlation CO2 Flux Data: 4 m samples, 30-min stats
  • Billesbach, Dave; Chan, Stephen; Biraud, Sebastien
  • Atmospheric Radiation Measurement (ARM) Archive, Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (US); https://doi.org/10.5439/1025036
dataset January 2002
Limitations and breakdown of Monin-Obukhov similarity theory for wind profile extrapolation under stable stratification: Limitations and breakdown of Monin-Obukhov similarity theory for wind profile extrapolation under stable stratification journal July 2015
Effects of precipitation on sonic anemometer measurements of turbulent fluxes in the atmospheric surface layer journal May 2016
Wind Resource Assessment book May 2012
How accurate is a machine learning-based wind speed extrapolation under a round-robin approach? journal September 2020
Observations and simulations of diurnal cycles of near-surface wind speeds over land and sea journal September 1996
Moving Beyond Monin–Obukhov Similarity Theory in Modelling Wind-Speed Profiles in the Lower Atmospheric Boundary Layer under Stable Stratification journal August 2014
Power Law Extrapolation of Wind Measurements for Predicting Wind Energy Production journal May 2009
Flux Parameterization over Land Surfaces for Atmospheric Models journal March 1991
Evaluation of Cut-Off Frequency and Correction of Filter-Induced Phase Lag and Attenuation in Eddy Covariance Analysis of Turbulence Data journal August 2003
Wind Speed Extrapolation Using Machine Learning Methods and LiDAR Measurements journal January 2018
Measurements of Boundary Layer Profiles in an Urban Environment journal June 2006
Power Law Extrapolation of Wind Measurements for Predicting Wind Energy Production journal May 2009
Calculation of Wind Speed Variation with Height over the Sea journal May 2006
A Comparison of Equilibrium and Time-Evolving Approaches to Modeling the Wind Profile under Stable Stratification journal May 2017
The MATERHORN: Unraveling the Intricacies of Mountain Weather journal November 2015
Difficulties in using power laws for wind energy assessment journal January 1983
The prediction of the wind speed at different heights by machine learning methods journal July 2016
Evaluation of Cut-Off Frequency and Correction of Filter-Induced Phase Lag and Attenuation in Eddy Covariance Analysis of Turbulence Data journal August 2003
How accurate is a machine learning-based wind speed extrapolation under a round-robin approach? journal September 2020
On the Use of Power Laws for Estimates of Wind Power Potential journal March 1978
Calculation of Wind Speed Variation with Height over the Sea journal May 2006
Flux-gradient relationships in the constant flux layer journal October 1970
Observations and simulations of diurnal cycles of near-surface wind speeds over land and sea journal September 1996
Flux Parameterization over Land Surfaces for Atmospheric Models journal March 1991
machine. journal October 2001
The elements of statistical learning: data mining, inference and prediction journal March 2005
ARM: Doppler Lidar Auxiliary Data dataset January 2014
Diurnal Cycles of Surface Winds and Temperatures as Simulated by Five Boundary Layer Parameterizations journal January 2004
The Determination of Kinematic Properties of a Wind Field Using Doppler Radar journal February 1968
Measurements of Boundary Layer Profiles in an Urban Environment journal June 2006
Extrapolating Satellite Winds to Turbine Operating Heights journal April 2016
ARM: ECOR: surface vertical fluxes of momentum, sensible heat, and latent heat, 30-min avg
  • Sullivan, Ryan; Cook, David
  • Atmospheric Radiation Measurement (ARM) Archive, Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (US); https://doi.org/10.5439/1025039
dataset January 1997

Cited By (1)