Back to the programme printer.gif Print



Wednesday, 12 March 2014
11:15 - 12:45 Wind speed predictions: Are we at the limit of our knowledge or can we improve?
Resource Assessment  


Room: Ponent
Session description

No new long-term correction methods have appeared for years and it is possible that current techniques are optimal. There are several issues which affect any long-term correction analysis:

… from the mundane: the optimum reference period? how do we measure success? re-analysis data or ground based stations? non-integer years of data?

… to the exotic: atmospheric stability, climate change decadal variations, sun spots activity and solar cycles.

It is likely that long-term correction techniques which consider these may provide more reliable predictions than has previously been possible.

This session describes new long-term correction methodologies and compares the results with those of conventional methods. Innovative techniques to improve the representativeness of long-term data series are discussed, different long-term data series are compared and conclusions on the decadal-scale variability of the wind speed are presented. The overall objective of this session is to give insight on how these developments contribute to a greater certainty in future wind speed predictions.

Learning objectives

  • Evaluate innovative methods to improve the representativeness of long-term data series and the overall accuracy of long-term extrapolations
  • Compare new long-term correction methods to traditional methods
  • Understand how a more accurate description of the decadal-scale variability of the wind speed contributes to the reduction of the uncertainty in the long-term corrected wind speed
Lead Session Chair:
Sónia Liléo, Kjeller Vindteknikk AS, Sweden

Co-chair(s):
Steve Ross, 3Tier
Abel Tortosa-Andreu Vortex, S.L., Spain
Co-authors:
Abel Tortosa-Andreu (1) F P Pau Casso (1) Patrícia Puig (1) Javier Viscarret (2) Ricardo Martínez (4) Nuria Sal de Rellán (5) Luís Prieto (6) Günter Vahlkmapf (7) Inma Murillo (3)
(1) Vortex, S.L., Barcelon, Spain (2) Acciona Energi�a, S.A., Sarriguren, Spain (3) Alstom Renovables Espa�a, S.L., Barcelona, Spain (4) Barlovento Recursos Naturales, S.L., Logroño, Spain (5) E.ON Renovables S.L.U., Madrid, Spain (6) Iberdrola Renovables Energía, S.A., Madrid, Spain (7) Sowitec development GmbH, Sonnenbühl, Germany

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Abel Tortosa has been working in the wind industry for more than five years. He is currently working at Vortex as a wind meteorologist, leading the forecasting research. Prior to joining Vortex, Abel worked as a project manager, head of wind and solar forecasting department, senior scientific programmer at AWS Truepower. He holds a Bsc. in physics and a Msc. in computational sciences and astrophysics. He has also undertaken PhD research into numerical simulation of the emergence and evolution of the magnetic field into the solar atmosphere at Institute of Astrophysics of Canary Island.

Abstract

On the benefit of a multivariate description of wind for a better long-term extrapolation

Introduction

Wind behavior is linked to other atmospheric quantities in a highly-dynamic and non-linear fashion over a broad spectrum of physical processes with different characteristic temporal and length scales. Current long-term extrapolation procedures obviate this inherent complexity; linear adjustment with measures based solely on the wind from a modeled series is an oversimplification, limiting the ability to accurately reproduce the features of the long-term resource. Modeling shortcomings, e.g. lack of synchronicity, regime-dependent bias, deficit of calms and peak winds among others, reminiscent of the different nature of observed and modeled data, currently preclude some traditional MCP techniques.

Approach

This paper proposes extending the range of application of the information provided by mesoscale modeling in favor of improving the representativeness, consistency and overall accuracy of long-term extrapolations through a non-linear multivariate statistical approach (named Re-modeling). The methodology presented aims to provide the best approximation to real measurement from long-term modeled data, offering a viable alternative or complement to traditional series and extrapolation methods, particularly in situations where these provide poor performance.

Main body of abstract

The Re-modeling makes use of a long-term reference high resolution WRF series of a heterogeneous (variables and levels) dataset and a year's worth of measurement data at the location of interest. In the proposed approach: (1) a multivariate analysis allows to understand the relation between the model variables and discriminate between useful and redundant information; (2) a state-of-the-art non-linear statistical procedure is applied to the model-related variables together with site observations for the coincident period; and (3) the physical-statistical model thus obtained is then used to predict the wind (speed and direction) profile over the original long-term series (of +20 years).

The paper presents a comprehensive analysis and validation of the method for more than 140 sites worldwide. A study of the out-of-training sensitivity of results was conducted for +10 sites using 5 years’ of wind certified measurements. A third-party validation of the proposed approach has been carried out by a number of sector-leading companies (Acciona, Alstom, Barlovento, E-On, Iberdrola, and Sowitec). Finally, the Re-modeling was compared with various MCP methods extensively used in the industry.

Conclusion

In 90 % to 100 % of the cases, the proposed methodology improves most of the attributes of the original series (hourly to monthly R², RMSE, BIAS, Wind Rose, etc.), and in a 70 % of cases, the Weibull shape, regardless of the reference year for training and for both the in- and out-of-training periods. The results obtained overall outperform those of traditional MCP methods available on the market.

The method proves to be a solid proposal for improvement of the current long-term series. It is not intended to exclude traditional MCP methods, but to cohabit with them for the sake of more reliable long-term wind representation.



Learning objectives
Present an innovative method to improve long-term series from measured data.
Understand that linear extrapolation methods based solely on the wind provide an incomplete description of the long-term resource.
Promote the use of a greater amount of information from modeling as a way to improve knowledge of local characteristics and evolution of the wind resource.