Future Wind Power Projections In China Based On The CMIP6 Multi-model Ensembl

To achieve the national target of carbon neutrality by the year 2060,clean energy such as wind power should be developed substantially in China.However,climate change causes large uncertainties to the future usage of wind resources.In recent years,the phenomenon of decreasing wind speed on land surface has received widespread attention,which further affects the use of wind energy,in order to quantify this phenomenon,this study used wind speed data and wind power data from 29 observed stations,combined with the CN05.1 homogenized observation data to calculate the wind power density distribution characteristics in China in2010s,and then using the machine learning method to modify the coupling mode to compare the simulation results of wind speed under SSP1-2.6 and SSP5-8.5 emission scenarios of the sixth phase of the plan（CMIP6）mode,the next three time periods in China（2025-2039,2050-2064,2085-2099）,relative to the temporal and spatial variation characteristics of the reference period（2000-2014）,the long-term changes of wind energy resources affected by wind speed were studied,and the main conclusions are as follows:（1）Wind power density calculated using the CN05.1 data resembles the observed wind speed.In this paper,the truncated wind speed is determined to be 10 m s^-1 with the minimum root-mean-square-error（RMSE）between observations and simulations.Accordingly,the wind power density calculated from the CN05.1 data shows an annual average value of 44.30 W m^-2during 2000-2014,with the maximum over northern China,Tibet,and the eastern coastal areas.（2）This study uses the Artificial Neural Network（ANN）method to calibrate the simulated surface wind speed from 16 CMIP6 climate models.Results show that the ANN on average reduces the RMSE of the CMIP6 climate models by 39.93±9.57%and increases the correlation coefficient from 0.56 to 0.83 between the multi-model ensemble mean simulations and observations,suggesting that the ANN-based calibration can better reproduces the spatial distribution and seasonal variation of surface wind in China at the present day.（3）The calibrated CMIP6 model data were used to project future wind power density,which showed lower values compared to the present day.Under the SSP1-2.6 scenario,the national wind power density is reduced by 0.52±0.83 W m^-2,1.01±0.94 W m^-2 and 0.98±1.17W m^-2 for the near-term（2025-2039）,carbon-neutral（2050-2064）and long-term（2085-2099）,respectively.Under the SSP5-8.5 scenario,it is reduced by 0.54±1.02 W m^-2,1.11±1.45 W m^-2,and 1.83±1.17 W m^-2.As the season with the highest wind power density,spring shows a decreasing trend in both emission scenarios,reduced by 1.16±1.14 W m^-2and 1.43±1.58 W m^-2respectively.And for spatial variation,the southeastern part of China shows an increasing trend in the SSP5-8.5 scenario,while all others show a decreasing trend.The correction of CMIP6 data using the e Xtreme Gradient Boosting（XGB）method shows smaller projection magnitude as those calibrated with the ANN method but similar tendencies,suggesting that the decreasing trend of wind power density in China becomes more significant with the strengthened global warming.