Fine resolution 30-year climatic wind energy dataset over China for renewable energy assessment and operation via Weather Research and Forecasting model hindcast

Abstract

One of the greatest obstacles in the exploitation of wind and solar energy is their intermittency and fluctuation. Long-term dataset of wind energy resource, which is built on basis of atmospheric numerical models, has been proved to be the most effective approach for wind energy resource assessment in wide area, on grid level and in fine resolution, as well as renewable energy electric energy capability forecast, site planning and dispatch/ operation schedule making of power system. To improve the accuracy of wind energy resource dataset by numerical models, the Weather Research and Forecasting model (WRF) and the Climate Four-Dimensional Data Assimilation (CFDDA) are adopted to conduct a sensitivity experiment, in 9 km spatial resolution, to distinguish a reliable model configuration in this study. Based on the statistics between WRF-CFDDA model output in hub-height wind speed and in-situ observations (sited on typical wind farms in Gansu, Xinjiang and Jilin province, 2010-2013), it has been confirmed that the model configuration with MERRA2 background field, Topowind topographic correction method and smoothed VASSO (VAriance of Sub-grid Scale Orography) terrain data is the most reasonable one, with a correlation coefficient 0.79 and RMSE 1.62 for 10-m height wind speed. Following this configuration, with assimilating all 30-year (1987-2016) ground-based meteorological observation from Chinese international exchange sites, the wind energy resource from 1987 to 2016 around China has been hindcasted and assessed in a resolution of 9 km, 15 min. The hindcast is capable to reproduce the characteristics in temporal and spatial distribution. This model system can be a reliable tool for reproducing decades of reanalyzed climatology and finer resolution assessment on hub-height wind energy, moreover, for reconstruction of typical wind power output curve.