Evaluation Of Northern Hemisphere Surface Wind Speeds And Wind Power Density From Multiple Sets Of Reanalysis And CMIP6 Model Data

Reanalysis products have become more and more popular for wind energy scientific community to analyze the wind speed variability and get long-term wind power estimations.The present study evaluates the biases of near-surface wind speed and wind power density in four of the most reputable reanalysis datasets,which include ERA-Interim,JRA-55,CFS and MERRA-2.The results indicate that the abilities of reanalysis products to reproduce the variabilities of wind speeds are different in different regions.JRA-55 and CFS offer the best estimates of annual and seasonal variabilities of surface wind speeds over the Northern Hemisphere.In detail,JRA-55 is the best to reproduce surface wind speeds in Asia,CFS has the best performance in Europe,and MERRA-2 just can reproduce the central part of North America.All the four datasets show decreasing tendencies in surface winds over the Northern Hemisphere during 1980-2016,although the trends are largely diverse among them.In terms of wind speed trend,JRA-55 is the best dataset,which can accurately reproduce the spatial distribution characteristics of Europe,Asia and central and eastern North America.The temporal and spatial characteristics of wind power density are similar to wind speed,but there are also obvious differences.The wind power density is proportional to the cube of the wind speed.Therefore,the fluctuation in wind power density is more drastic than that in wind speed,and the percentage of deviation between the reanalysis data and the observations in wind power density is much bigger than that in wind speed.Moreover,the biases of near-surface wind speed over the Northern Hemisphere are also evaluated in eighteen CMIP6 model datasets.The results show that each of CMIP6 model datasets can reflect the downward trend of wind speed,but the trend magnitude is much lower than the observed data.By comparing the correlation coefficient,standard deviation ratio,and bias,EC-Earth3 and CESM2 are closest to the observations.