Back to the programme printer.gif Print




Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'Wind speed predictions: Are we at the limit of our knowledge or can we improve?' taking place on Wednesday, 12 March 2014 at 11:15-12:45. The meet-the-authors will take place in the poster area.

Dario Patane EREDA, Spain
Co-authors:
Javier Peña (1) F P Dario Patane (1) Fernando De La Blanca (1)
(1) EREDA, madrid, Spain

Printer friendly version: printer.gif Print

Abstract

Evaluating the sensitivity of mesoscale modelling to reanalysis data

Introduction

Mesoscale modeling has been proven to be an effective tool for preliminary resource assessment of sites where no measurement is available. Mesoscale models process low resolution global reanalysis data in order to obtain a fine-grained description of the local atmospheric flow. The final result is thus a complex combination of the synoptic scale features ingested from the reanalysis inputs plus local effects such as topographic or thermally driven phenomena. Here we investigate to what extent the mesoscale wind resource assessment is affected by using different reanalysis data.

Approach

Specifically we use the Weather Research and Forecasting (WRF) model which has been extensively validated in wind energy assessment studies during the last years.
We considered 4 reanalysis databases and compared the results obtained for a set of 30 high quality meteorological masts located worldwide in sites with a varied range of characteristics, ranging from complex and flat terrains, coastal zones and offshore sites.

Main body of abstract

The wind conditions at each meteorological mast were simulated for a cumulated period of 30 non consecutive days in order to properly take into account seasonal variations. Simulated wind speed and turbulence were compared to the measured values and statistical tests were evaluated.
Avarage wind speed mean absolute errors ranged from 7% to 10%, while for wind turbulence all reanalysis datasets showed similar errors (~18%) mostly due to an underestimation bias.
Biases and mean errors were found to differ about 3% on average, however for some masts a spread of more than 10% was found, which is comparable with the modelling error. In these specific cases a strong sensitivity of the results to the choice of the reanalysis data is found.
On the other hand correlation coefficients between the simulated and the measured time series show a statistically significant difference between the NCAR 2.5º reanalysis and the others, the former having substantially lower correlation on average. c

Conclusion

The results of the study suggest that if we are interested in the correlation between the simulated and measured wind speed time series, as in the case of MCP analysis, than it is important to properly choose the reanalysis data. On the other hand statistical metrics such as bias and mean absolute error, which are more relevant for preliminary wind resource assessment, show similar results with no significant differences for the analyzed datasets.


Learning objectives
The aim of this presentation is to shed some light on the relation between reanalysis data and mesoscale modelling both for long term and preliminary resource assessment.