Conference programme

Back to the programme printer.gif Print




Delegates are invited to meet and discuss with the poster presenters during the poster presentation sessions between 10:30-11:30 and 16:00-17:00 on Thursday, 19 November 2015.

Lead Session Chair:
Stephan Barth, ForWind - Center for Wind Energy Research, Germany
Llorenç Lledó AWS Truepower, Spain
Co-authors:
Oriol Lacave (1) F Llorenç Lledó (1) Jose Vidal (1)
(1) AWS Truepower, Barcelona, Spain

Share this poster on:

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2015 and are published here unedited

Mr. Lledó holds a Bachelor of Mathematics and a Master in Meteorology. Mr. Lledó has been involved in the renewable energy field for 7 years. He has focused in the numerical simulation of wind conditions for operative forecasting and reanalysis purposes. He has worked in the development of several wind atlases for governments, institutions and private companies. Mr. Lledó has also been involved in the development of databases of global and regional reanalysis. He has 10 years of experience in the field of numerical weather prediction, configuring, running and post processing data from models as WRF, MASS, MM5 & WW3.


Poster

Poster Download poster (16.80 MB)

Abstract

MCP studies with re-analysis models: get a forecast of the uncertainty and complexity without observational data.

Introduction

Re-analysis models are widely used in the wind industry when facing projects that need long term data correction adjustments. The available data from monitoring towers is also a required source in such studies. Re-analysis limitations arise only after comparison between both are performed.
We present several maps with the purpose to give quantitative information of correlations, directional patterns and tendencies between the different models. This information should lead to prior knowledge of the complexity of the long term adjustment process before comparing to tower data.

Approach

We want to provide a resource that is available for the whole globe. This is not feasible with the use of observational data. A comparison for all the models pair to pair was done. It first required a model post-processing to obtain a comparable common data format for all the different sources. A new methodology was developed in order to compare different models with different grids, which involves nodes located at different places. Because additional quantitative information to the correlation is highly valuable, more metrics were applied in order to obtain tendencies and directional wind patterns.

Main body of abstract

The models used in the study are MERRA from NASA, NNRP from NCEP/NCAR, CFSR from NCEP and ERA-Interim from ECMWF. Each model has its own vertical level coordinates. Obtaining data at a fixed height above ground requires understanding of the different systems. Different vertical interpolation methods were applied for each model focusing in minimizing the steps and thus the variable degradation.
Horizontal comparison is also a problem to overcome. To allow comparison between data located at the different grid points for each unique model grid, a method using a virtual grid was developed. This method captures a reliable model representation at each virtual grid node. The comparison is then made on the new grid node by node.
Due to the importance of correlation parameter in the measure correlate compare (MCP) method, this has been first used to compare the models. Analysis of wind speed distribution differences has added information in the following step, specially for those places where correlations where high, differentiating in most cases the models behaviour. One problem when obtaining long term data with MCP method using re-analysis as reference data is the tendency inconsistencies between the two datasets. If tendency for reference data changes in the concurrent period respect to long term period, the long term adjustment increases the uncertainty. That explains why detailed spatial information on the tendencies for the different models were compared for recent years and long term periods. This has also revealed spatial differences in the tendencies when analysing models separately.

Conclusion

In addition to correlation, more information is needed to quantify reliability and uncertainty of the results and to choose the models to use as reference data. With the described model-to-model comparison method it emerges new information to the classical tower-models comparison.
It has been observed when analysing models separately that tendencies are not spatially homogeneous. Any model behaviour found in local studies should not then be generalized. A deeper analysis with the temporal changes in the sources of data assimilation should be performed to complement this study and extract conclusions on this issue if possible.


Learning objectives
The study intends to allow a better understanding of how representative and reliable long term data will be prior to any case study. Anticipating complexity of an MCP study would be of great value for an estimate of the resulting derived uncertainty. We’ve found cases with different tendencies when comparing last years and long term periods. It would then be recommended to use more meteorological data with longer periods.