摘要
为了进一步提高WRF模式对风速预报的准确性,以我国某风电场01#测风塔2007年5月和11月的数据为例,通过线性回归方法并结合滚动和极值处理技术,对WRF模式模拟风速进行了订正。结果表明,直接使用线性回归方法对于模拟风速的订正无明显效果;采用滚动技术的线性回归订正效果与步长有关,与线性回归订正相比总体上有较大改进,其中3 h步长改进更明显;相同步长,线性滚动极值处理订正较线性滚动订正相比有进一步改进,其中1 h步长线性滚动极值处理效果最优,如5月和11月订正前模拟风速的相对均方根误差(rRMSE)分别为29.274%、33.583%,订正后下降为14.714%、14.493%。订正后精度明显提高,更接近实况风速,线性滚动极值处理订正方法能够较好订正模式模拟风速,有效提高风速预报准确率。
To find out the way of further improving the ability and accuracy of wind speed prediction, two sets of 10 minute data in May and November 2007 were used to revise the WRF-forecast wind speed with a linear regression method, rolling and extremum technology in this paper. One is the observed mast data on a wind farm in China and the other is the simulated wind speed by the WRF model. The results are shown as follows: (a) The simple linear regression method has no significant effect on revising WRF wind speed. (b) Generally, linear regression with the rolling technology is better than the simple linear regression, and its revised results depend on step, and 3 h step is improved more than other steps. (c) Under the same step conditions, linear rolling regression with rolling and extremum technology is better than the linear regression only with rolling. For all steps, the 1 h revision is the best, and the relative root mean square error (rRMSE) of wind speed forecast drops from 29.274% to 14.714% in May and from 33.583% to 14.493% m/s in November, after the revision. The prediction precision was significantly improved, resulting in a wind speed that is closer to reality. It proves that linear regression can better correct the wind speed of WRF forecasting forecasting. combined with the rolling and extremum technology and effectively enhance the accuracy of wind speed forecasting.
出处
《热带气象学报》
CSCD
北大核心
2013年第4期681-686,共6页
Journal of Tropical Meteorology
基金
公益性行业(气象)科研专项(GYHY201006035)资助
关键词
线性滚动极值处理
模拟风速
风速订正
linear rolling and extremum regression
simulated wind speed
wind speed correction