Future hub-height wind speed distributions from statistically downscaled CMIP5 simulations
Abstract
In order to realistically estimate wind-power yields, we need to know the hub-height wind speed under future climate conditions. Climate conditions of the upper atmosphere are commonly simulated using general circulation models (GCMs). However their typical resolutions are too coarse to assess the climate at the height of a wind turbine. This study simulates the hub-height wind speed probability distributions (PDFs) over Europe under future climate conditions. The analysis is based on the simulations of the CMIP5 earth system models, which are the latest development of GCMs. They include more components and feedbacks and their runs are performed at higher resolutions. In a first step, the ensemble of GCMs is evaluated on their representation of the wind speed PDFs in the lower atmosphere using ERA-Interim data. The evaluation indicates that GCMs are skillful down to their lowest model levels apart for the south of Europe, which is affected by a large scale winter bias and for certain coastal and orographical regions. Secondly, a statistical approach is developed which downscales the GCM output to the wind speed PDF at the height of the wind turbine hub. Since the evaluation analysis shows that GCMs are also skillful at the lower model levels, the statistical downscaling uses GCM variables describing the lower atmosphere, instead of the commonly used large scale circulation variables of the upper atmosphere. By doing so less uncertainty will be added trough the downscaling implementation. The downscaling methodology is developed for an observational site in the Netherlands, using hub-height wind speed observations and ERA-Interim data for the period 1989-2009. The statistical approach is based on a regression analysis of the parameters of the PDFs. Results indicate that the predictor selection is very much defined by the stability conditions of the atmospheric boundary layer. During convective summer-day conditions, the observed hub-height wind speed can skillfully be modeled using the GCM wind speed as the only predictor. In contrast, during summer-nights the stable boundary layer is much shallower and the statistical model indicates that the simulation of hub-height wind speed PDFs is substantially improved when temperature information is included in the statistical downscaling model. In a final step the statistical downscaling model will be applied on simulations of the ensemble of CMIP5 GCMs for present and future climate conditions. The downscaled hub-height wind speed PDFs will be used to analyze potential wind-power curves.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.A11L..02D
- Keywords:
-
- 1637 GLOBAL CHANGE Regional climate change;
- 9335 GEOGRAPHIC LOCATION Europe;
- 3337 ATMOSPHERIC PROCESSES Global climate models;
- 3307 ATMOSPHERIC PROCESSES Boundary layer processes