采用中尺度气象模式WRF(Weather Research Forecast)对北京地区的太阳辐射进行了4个典型月的逐时预报试验, 用南郊观象台的辐射观测数据对预报结果进行了对比分析和初步订正试验。结果表明: 在现有模式条件下, 5 km分辨率的短波辐射预报结果和1 km分辨率预报结果无明显差别; WRF模式对太阳辐射的预报性能在晴天较好, 多云天次之, 在满云或阴雨天最差; 通过误差分解发现, 位相偏差、系统偏差及振幅偏差在各月对均方根误差的贡献有明显差异; 针对模式预报结果的系统偏差和振幅偏差.经过简单的线性订正可以较明显地改进模式预报结果; 双偏订正(DBC)法比线性回归(LR)法对预报误差的改进效果略明显; 仅通过简单的线性订正, 位相差很难消除, 需要针对位相差研究新的订正方法。
Weather Research Forecast (WRF) model was used in experimental forecast of hourly solar radiation in four typical months in Beijing area. The solar radiation observational data of Nanjiao weather observatory were used to evaluate and correct the forecasting result. Conclusions were as follows: the forecasting result of solar radiation with 5 km horizontal resolution was very similar to that with 1 km horizontal resolution in condition of the existing model. The forecast effect of solar radiation in sunny days was the best with WRF model, the secondary was in cloudy days, and the worst was in fully cloudy days or rainy days. Through analysis of error decomposition, the contribution of phase bias, system bias and amplitude bias was different obviously in each month. The ordinary linear correction method was useful to remove most of system bias and amplitude bias obviously, and in general, the correcting effect of double bias correction method was slightly better than that by linear regression method. The phase bias was difficult to eliminate only by ordinary linear correction method, so a new correction method should be adopted.
[1]Cote S, Tatnall A R L. A neural network-based method for tracking features from satellite sensor images[J]. Int J Remote Sens, 1995, 16 (18): 3695-3701.
[2]Hammer A, Heinemann D, Hoyer C, et al. Solar energy assessment using remote sensing technologies[J]. Remote Sens Environ, 2003, 86(3): 423-432.
[3]Kaifel A K, Jesemann P. An adaptive filtering algorithm for very short-range forecast of cloudiness applied to Meteosat data[R]. Proc. 9<sup>th</sup> Meteosat Users Meeting, Locarno, 1992.
[4]Armstrong M A. Comparison of MM5 forecast shortwave radiation with data obtained from the atmospheric radiation measurement program[D]. Maryland: University of Maryland, 2000.
[5]Lorenz E, Remund J, Muller S, et al. Benchmarking of different approaches to forecast solar irradiance[R]. 24<sup>th</sup> European Photovoltaic Solar Energy Conference, Hamburg, 2009: 1-10.
[6]马金玉, 罗勇, 申彦波, 等. 太阳能预报方法及其应用和问题[J]. 资源科学, 2011, 33(5): 829-837.
[7]李芬, 陈正洪, 陈驰, 等. 太阳能光伏发电量预报方法的发展[J]. 气候变化研究进展, 2011, 7(2): 136-142
[8]彭继达, 程兴宏, 孙治安, 等. 两种不同初始场对太阳辐射模拟效果的影响[J]. 高原气象, 2014, 33(5): 1352-1362, doi: 10.7522/j.issn.1000-0534.2013.00098.
[9]符传博, 丹利, 吴涧, 等. 近46年西南地区晴天日照时数变化特征及其原因初探[J]. 高原气象, 2013, 32(6): 1729-1738, doi: 10.7522/j.issn.1000-0534.2012.00162.
[10]辛渝, 赵逸舟, 毛炜峄, 等. 新疆太阳总辐射资料的均一性检验与气候学估算式的再探讨[J]. 高原气象, 2011, 30(4): 878-889.
[11]刘清惓, 戴伟, 杨荣康, 等. 探空温度测量太阳辐射误差的流体动力学分析[J]. 高原气象, 2013, 32(4): 1157-1164, doi: 10.7522/j.issn.1000-0534.2012.00109.
[12]刘俊峰, 陈仁升, 阳勇, 等.实际地形下30min太阳辐射模拟及误差分析—以祁连山马粪沟流域为例[J]. 高原气象, 2011, 30(6): 1647-1652.
[13]白永清, 陈正洪, 王明欢, 等, 基于WRF模式输出统计的逐时太阳总辐射预报初探[J]. 大气科学学报, 2011, 34(3): 363-369.
[14]Weather Research and Forecast ARW Version 3 Modeling Systems User's Guide[EB/OL]. http://www.mmm.ucar.edu/wrf/docs/user_guide_V3/SRWUsers-GuideV3.pdf, 2010.
[15]吴其重, 王自发, 崔应杰. 我国近20年太阳辐射时空分布状况模式评估[J]. 应用气象学报, 2010, 21(3): 343-351.
[16]黄海波, 陈春艳, 朱雯娜. WRF模式不同云微物理参数化方案及水平分辨率对降水预报效果的影响[J]. 气象科技, 2011, 39(5): 529-536.
[17]闫之辉, 邓莲堂. WRF模式中的微物理过程及其预报对比试验[J]. 沙漠与绿洲气象, 2007, 1(6): 1-6.
[18]沈元芳, 黄丽萍, 徐国强, 等. 长波辐射对大气变化的敏感性和在WRF模式中的应用检验[J]. 气象学报, 2004, 62(2): 213-227.
[19]Lange M. On the uncertainty of wind power predictions-analysis of the forecast accuracy and statistical distribution of errors[J]. Journal of Solar Energy Engineering, 2005, 127(2): 177-184, doi: 10.1115/1.1862266.
[20]Lange M. Analysis of the uncertainty of wind power predictions[D]. Oldenburg: University of Oldenburg, http://docserver.bis.unioldenburg.de/publikationen/dissertation/2003/lanana03/lanana03.html, 2003: 28-29.
