A study on surface temperature and wind speed forecast method at Winter Olympic stations in complex terrain area based on the CMA-BJ numerical weather prediction model products
-
摘要: 北京冬奥服务对站点气象要素预报提出了明确需求，2 m气温预报偏差在±2℃以内，10 m风速预报平均偏差小于观测的30%，文中提出一种基于相似集合嵌套一元线性回归的预报方法—嵌套相似集合（AnEn-Ne），该方法基于相似集合思路，在满足一定条件时，启动其嵌套一元线性回归提供订正预报。冬奥赛期（2021年11月1日—2022年3月15日）实时业务预报表明，嵌套相似集合具有较好的预报效果，相对业务数值模式（CMA-BJ）预报，预报精度显著提高，相对相似集合预报和一元线性回归预报精度明显提高，其预报结果满足冬奥服务需求。复杂地形下的要素预报检验表明，CMA-BJ模式预报2 m气温虽然存在较明显的系统偏差，但与观测相关较强，对观测的表征意义明显，订正后能有效消除复杂地形影响，10 m风速模式预报偏差振荡明显，模式预报与观测相关较弱，表征意义差，订正后站间差异明显；改进CMA-BJ模式复杂地形区近地面风速预报对观测的表征意义，可进一步提高该预报方法对10 m风速订正预报的精度。Abstract: Beijing Winter Olympic service has made a clear request for prediction of weather elements at individual sites. The 2 m temperature forecast bias is less than ±2℃, and 10 m wind speed forecast bias is less than 30% of observation average. This paper proposes a forecast method based on analog ensemble (AnEn) nested linear regression (LR)—analog ensemble nested linear regression (AnEn-Ne). When certain conditions are met, the nested linear regression is activated to provide revised forecasts. The real-time operational forecasts during the Winter Olympic period (1 November 2021 to 15 March 2022) show that the AnEn-Ne method has a better forecast effect. Compared with that of the CMA-BJ, the forecast accuracy of the AnEn-Ne is improved significantly; compared with that of AnEn and LR, the forecast accuracy of the AnEn-Ne is obviously improved, and the forecasts meet the service demand at the Winter Olympic stations. The verification of forecasted elements in complex terrain area shows that, despite an obvious systematic deviation of 2 m temperature in the forecast by the CMA-BJ, the forecasts are strongly correlated with observations and show an obvious representation of observations, and the influence of complex terrain can be effectively eliminated after correction. The CMA-BJ forecast Bias of 10 m wind speed demonstrates an oscillation feature, and the correlation between the model forecasts and observations is weak, indicating a poor model representation of surface wind. The differences in 10 m wind speed between stations are obvious after correction. Improving the representation of the CMA-BJ model on surface wind speed prediction in complex terrain areas can further improve the accuracy of the method for the correction of 10 m wind speed prediction.
-
Key words:
- AnEn-Ne /
- AnEn /
- LR /
- Elements prediction /
- CMA-BJ
-
图 1 延庆和崇礼赛区冬奥自动站分布及分类 （a.延庆，b.崇礼）
Figure 1. Distribution and classification of Winter Olympic Automatic Weather Stations （AWS） in Yanqing and Chongli competition areas （a. Yanqing，b. Chongli）
图 2 2020年11月1日—2021年3月15日03时起始1—72 h逐时AnEn、LR、AnEn-Ne订正预报偏差相对历史样本集 （2018年7月21日—预报前1天） 观测平均百分率及10 m风速观测变化百分率 （OBS_c） 的51个冬奥测站分站统计
Figure 2. Percentages of hourly biases corrected by AnEn，LR and AnEn-Ne to historical （ 21 July 2018—1 day before the current forecast） observation averages during 1—72 h forecast started at 03:00 UTC from 1 November 2020 to 15 March 2021，and percentage of 10 m wind speed observation changes （OBS_c） at 51 Winter Olympic stations
图 3 10 m风速预报分站检验均方根误差 （RMSE）、预报偏差 （Bias） 箱线和模式与实际地形高度差 （a. CMA-BJ，b. AnEn，c. LR，d. AnEn-Ne） 及按预报时效统计检验10 m风速预报均方根误差 （e） 和预报偏差 （f）
Figure 3. Boxplots of RMSE and bias for 10 m wind speed predictions and terrain height difference between model and observation （model terrain minus observation） at individual stations （a. CMA-BJ，b. AnEn，c. LR，d. AnEn-Ne），and changes in RMSE （e） and Bias （f） of 10 m wind speed predictions with forecast time
图 4 2020年11月1日—2021年3月15日03时起始38个冬奥测试站1—72 h逐时2 m气温预报均方根误差 （a） 和预报偏差 （b） 按起报时间统计
Figure 4. Hourly RMSE （a） and Bias （b） of 2 m temperature predictions at 38 Winter Olympic stations during 1—72 h forecast starting at 03:00 UTC from 1 November 2020 to 15 March 2021
图 5 2 m气温预报分站检验均方根误差 （RMSE）、预报偏差 （bias） 箱线 （单位：℃） 和模式与实际地形高度差 （单位：m）（a. CMA-BJ，b. AnEn，c. LR，d. AnEn-Ne） 及按预报时效统计检验2 m气温预报均方根误差 （e） 和预报偏差 （f） （单位：℃）
Figure 5. Boxplots of RMSE and bias （unit：℃） for 2 m temperature predictions and terrain height difference （model terrain minus observation） at individual stations （unit：m） （a. CMA-BJ， b. AnEn，c. LR，d. AnEn-Ne）， and changes in RMSE （e） and bias （f） of 2 m temperature predictions with forecast time （unit：℃）
图 6 CMA-BJ、AnEn、LR和AnEn-Ne预报2 m气温分站点统计高 （a）、低 （b） 温预报准确率
Figure 6. High （a） and low （b） prediction accuracy of 2 m temperature at individual stations for CMA-BJ，AnEn，LR and AnEn-Ne forecast
图 7 10 m极大风速预报分站检验均方根误差 （RMSE）、预报偏差 （bias） 箱线 （单位：m/s） 和模式与实际地形高度差 （单位：m）（a. CMA-BJ，b. AnEn，c. LR，d. AnEn-Ne） 及按预报时效统计检验10 m极大风速预报均方根误差 （e） 和预报偏差 （f） （单位：m/s）
Figure 7. Boxplots of RMSE and bias （unit：m/s） of 10 m maximum wind speed predictions and terrain height difference （model terrain minus observation） at individual stations （unit：m） （a. CMA-BJ，b. AnEn，c. LR，d. AnEn-Ne），and changes in RMSE （e） and bias （f） of 10 m maximum wind speed predictions with forecast time （unit：m/s）
图 8 冬奥赛期分站统计2 m气温预报偏差 （a） 和10 m风速预报偏差 （b） 占观测的平均百分比
Figure 8. Bias of 2 m temperature predictions （a） and percentage of 10 m wind speed predictions Bias to observation averages （b） during 2022 Beijing Winter Olympics period at individual stations
图 9 北京20个国家级气象站分站统计2021年11月1日—2022年3月15日03时起始1—72 h逐时2 m气温和10 m风速预报均方根误差和预报偏差 （a. 2 m气温预报均方根误差，b. 2 m气温预报偏差，c.10 m风速预报均方根误差，d.10 m风速预报偏差）
Figure 9. Hourly RMSEs and biases of 2 m temperature and 10 m wind speed during 1—72 h forecast period started at 03:00 UTC From 1 November 2021 to 15 March 2022 at 20 national weather stations （a. RMSE of 2 m temperature predictions，b. Bias of 2 m temperature predictions，c. RMSE of 10 m wind speed predictions，d. Bias of 10 m wind speed predictions）
-
[1] 丁士晟. 1985. 中国MOS预报的进展. 气象学报,43(3):332-338. Ding S S. 1985. The advance of model output statistics method in China. Acta Meteor Sinica,43(3):332-338 (in Chinese Ding S S .1985 . The advance of model output statistics method in China. Acta Meteor Sinica,43 （3 ）:332 -338 （in Chinese）[2] 郝翠,张迎新,王在文等. 2019. 最优集合预报订正方法在客观温度预报中的应用. 气象,45(8):1085-1092. Hao C,Zhang Y X,Wang Z W,et al. 2019. Application of analog ensemble rectifying method in objective temperature prediction. Meteor Mon,45(8):1085-1092 (in Chinese Hao C, Zhang Y X, Wang Z W, et al .2019 . Application of analog ensemble rectifying method in objective temperature prediction. Meteor Mon,45 （8 ）:1085 -1092 （in Chinese）[3] 胡艺,符娇兰,陶亦为等. 2022. 冬奥会延庆赛区气象要素分布特征分析. 气象,48(2):177-189. Hu Y,Fu J L,Tao Y W,et al. 2022. Characteristics of meteorological elements over Yanqing Area during Winter Olympic Games. Meteor Mon,48(2):177-189 (in Chinese Hu Y, Fu J L, Tao Y W, et al .2022 . Characteristics of meteorological elements over Yanqing Area during Winter Olympic Games. Meteor Mon,48 （2 ）:177 -189 （in Chinese）[4] 李嘉睿,符娇兰,陶亦为等. 2022. 冬奥会张家口赛区气温与风的特征分析. 气象,48(2):149-161. Li J R,Fu J L,Tao Y W,et al. 2022. Temperature and wind characteristic analysis in Zhangjiakou Olympic Area for the Winter Olympic Games. Meteor Mon,48(2):149-161 (in Chinese Li J R, Fu J L, Tao Y W, et al .2022 . Temperature and wind characteristic analysis in Zhangjiakou Olympic Area for the Winter Olympic Games. Meteor Mon,48 （2 ）:149 -161 （in Chinese）[5] 全继萍,李青春,仲跻芹等. 2022. “CMA 北京模式”中三种不同阵风诊断方案在北京地区大风预报中的评估. 气象学报,80(1):108-123. Quan J P,Li Q C,Zhong J Q,et al. 2022. Evaluation of three different gust diagnostic schemes in the CMA-BJ for gale forecasting over Beijing. Acta Meteor Sinica,80(1):108-123 (in Chinese Quan J P, Li Q C, Zhong J Q, et al .2022 . Evaluation of three different gust diagnostic schemes in the CMA-BJ for gale forecasting over Beijing. Acta Meteor Sinica,80 （1 ）:108 -123 （in Chinese）[6] 任萍,陈明轩,曹伟华等. 2020. 基于机器学习的复杂地形下短期数值天气预报误差分析与订正. 气象学报,78(6):1002-1020. Ren P,Chen M X,Cao W H,et al. 2020. Error analysis and correction of shortterm numerical weather prediction under complex terrain based on machine learning. Acta Meteor Sinica,78(6):1002-1020 (in Chinese Ren P, Chen M X, Cao W H, et al .2020 . Error analysis and correction of shortterm numerical weather prediction under complex terrain based on machine learning. Acta Meteor Sinica,78 （6 ）:1002 -1020 （in Chinese）[7] 沈学顺,王建捷,李泽椿等. 2020. 中国数值天气预报的自主创新发展. 气象学报,78(3):451-476. Shen X S,Wang J J,Li Z C,et al. 2020. China's independent and innovative development of numerical weather prediction. Acta Meteor Sinica,78(3):451-476 (in Chinese Shen X S, Wang J J, Li Z C, et al .2020 . China's independent and innovative development of numerical weather prediction. Acta Meteor Sinica,78 （3 ）:451 -476 （in Chinese）[8] 王丹,王建鹏,白庆梅等. 2019. 递减平均法与一元线性回归法对ECMWF温度预报订正能力对比. 气象,45(9):1310-1321. Wang D,Wang J P,Bai Q M,et al. 2019. Comparative correction of air temperature forecast from ECMWF Model by the decaying averaging and the simple linear regression methods. Meteor Mon,45(9):1310-1321 (in Chinese Wang D, Wang J P, Bai Q M, et al .2019 . Comparative correction of air temperature forecast from ECMWF Model by the decaying averaging and the simple linear regression methods. Meteor Mon,45 （9 ）:1310 -1321 （in Chinese）[9] 王倩倩,权建农,程志刚等. 2022. 2019年冬季北京海陀山局地环流特征及机理分析. 气象学报,80(1):93-107. Wang Q Q,Quan J N,Cheng Z G,et al. 2022. Local circulation characteristics and mechanism analysis of Haituo mountain in Beijing during winter 2019. Acta Meteor Sinica,80(1):93-107 (in Chinese Wang Q Q, Quan J N, Cheng Z G, et al .2022 . Local circulation characteristics and mechanism analysis of Haituo mountain in Beijing during winter 2019. Acta Meteor Sinica,80 （1 ）:93 -107 （in Chinese）[10] 王在文,郑祚芳,陈敏等. 2012. 支持向量机非线性回归方法的气象要素预报. 应用气象学报,23(5):562-570. Wang Z W,Zheng Z F,Chen M,et al. 2012. Prediction of meteorological elements based on nonlinear support vector machine regression method. J Appl Meteor Sci,23(5):562-570 (in Chinese Wang Z W, Zheng Z F, Chen M, et al .2012 . Prediction of meteorological elements based on nonlinear support vector machine regression method. J Appl Meteor Sci,23 （5 ）:562 -570 （in Chinese）[11] 王在文,陈敏,Monache L D等. 2019. 相似集合预报方法在北京区域地面气温和风速预报中的应用. 气象学报,77(5):869-884. Wang Z W,Chen M,Monache L D,et al. 2019. Application of analog ensemble method to surface temperature and wind speed prediction in Beijing area. Acta Meteor Sinica,77(5):869-884 (in Chinese Wang Z W, Chen M, Monache L D, et al .2019 . Application of analog ensemble method to surface temperature and wind speed prediction in Beijing area. Acta Meteor Sinica,77 （5 ）:869 -884 （in Chinese）[12] 杨璐,宋林烨,荆浩等. 2022. 复杂地形下高精度风场融合预报订正技术在冬奥会赛区风速预报中的应用研究. 气象,48(2):162-176. Yang L,Song L Y,Jing H,et al. 2022. Fusion prediction and correction technique for high-resolution wind field in Winter Olympic Games area under complex terrain. Meteor Mon,48(2):162-176 (in Chinese doi: 10.7519/j.issn.1000-0526.2021.092902 Yang L, Song L Y, Jing H, et al .2022 . Fusion prediction and correction technique for high-resolution wind field in Winter Olympic Games area under complex terrain. Meteor Mon,48 （2 ）:162 -176 （in Chinese） doi: 10.7519/j.issn.1000-0526.2021.092902[13] 曾庆存. 2013. 天气预报—由经验到物理数学理论和超级计算. 物理,42(5):300-314. Zeng Q C. 2013. Weather forecast:From empirical to physicomathematical theory and super-computing system engineering. Physics,42(5):300-314 (in Chinese Zeng Q C .2013 . Weather forecast: From empirical to physicomathematical theory and super-computing system engineering. Physics,42 （5 ）:300 -314 （in Chinese）[14] 曾晓青,薛峰,赵瑞霞等. 2019. 几种格点化温度滚动订正预报方案对比研究. 气象,45(7):1009-1018. Zeng X Q,Xue F,Zhao R X,et al. 2019. Comparison study on several grid temperature rolling correction forecasting schemes. Meteor Mon,45(7):1009-1018 (in Chinese Zeng X Q, Xue F, Zhao R X, et al .2019 . Comparison study on several grid temperature rolling correction forecasting schemes. Meteor Mon,45 （7 ）:1009 -1018 （in Chinese）[15] 张延彪,陈明轩,韩雷等. 2022. 数值天气预报多要素深度学习融合订正方法. 气象学报,80(1):153-167. Zhang Y B,Chen M X,Han L,et al. 2022. Multi-element deep learning fusion correction method for numerical weather prediction. Acta Meteor Sinica,80(1):153-167 (in Chinese Zhang Y B, Chen M X, Han L, et al .2022 . Multi-element deep learning fusion correction method for numerical weather prediction. Acta Meteor Sinica,80 （1 ）:153 -167 （in Chinese）[16] Barker D M,Huang W,Guo Y R,et al. 2004. A three-dimensional variational data assimilation system for MM5:Implementation and initial results. Mon Wea Rev,132(4):897-914 doi: 10.1175/1520-0493(2004)132<0897:ATVDAS>2.0.CO;2 [17] Bauer P,Thorpe A,Brunet G. 2015. The quiet revolution of numerical weather prediction. Nature,525(7567):47-55 doi: 10.1038/nature14956 [18] Benjamin S G,Brown J M,Brunet G,et al. 2019. 100 years of progress in forecasting and NWP applications. Meteor Monogr,59(1):13.1-13.67 [19] Frogner I L,Nipen T,Singleton A,et al. 2016. Ensemble prediction with different spatial resolutions for the 2014 Sochi Winter Olympic Games:The effects of calibration and multimodel approaches. Wea Forecasting,31(6):1833-1851, doi: 10.1175/WAF-D-16-0048.1 [20] Glahn H R,Lowry D A. 1972. The use of model output statistics (MOS) in objective weather forecasting. J Appl Meteor,11(8):1203-1211 doi: 10.1175/1520-0450(1972)011<1203:TUOMOS>2.0.CO;2 [21] Han K,Choi J,Kim C. 2018. Comparison of statistical post-processing methods for probabilistic wind speed forecasting. Asia-Pac J Atmos Sci,54(1):91-101 doi: 10.1007/s13143-017-0062-z [22] Huang X Y,Xiao Q N,Barker D M,et al. 2009. Four-dimensional variational data assimilation for WRF:Formulation and preliminary results. Mon Wea Rev,137(1):299-314 doi: 10.1175/2008MWR2577.1 [23] Isaac G A,Joe P I,Mailhot J,et al. 2014. Science of nowcasting olympic weather for vancouver 2010 (SNOW-V10):A world weather research programme project. Pure Appl Geophys,171(1):1-24 [24] Kalnay E. 2003. Atmospheric Modeling,Data Assimilation and Predictability. Cambridge:Cambridge University Press,341pp [25] Klein W H,Lewis B M,Enger I. 1959. Objective prediction of five-day mean temperatures during winter. J Meteor,16(6):672-682 doi: 10.1175/1520-0469(1959)016<0672:OPOFDM>2.0.CO;2 [26] Li Y P,Lang J W,Ji L,et al. 2021. Weather forecasting using ensemble of spatial-temporal attention network and multi-layer perceptron. Asia-Pac J Atmos Sci,57(3):533-546 doi: 10.1007/s13143-020-00212-3 [27] Liu Y B,Warner T T,Astling E G,et al. 2008. The operational mesogamma-scale analysis and forecast system of the U. S. Army test and evaluation command. Part Ⅱ:Interrange comparison of the accuracy of model analyses and forecasts. J Appl Meteor Climatol,47(4):1093-1104 doi: 10.1175/2007JAMC1654.1 [28] Marzban C. 2003. Neural networks for postprocessing model output:ARPS. Mon Wea Rev,131(6):1103-1111 doi: 10.1175/1520-0493(2003)131<1103:NNFPMO>2.0.CO;2 [29] Monache L D,Eckel F A,Rife D L,et al. 2013. Probabilistic weather prediction with an analog ensemble. Mon Wea Rev,141(10):3498-3516. doi: 10.1175/MWR-D-12-00281.1 [30] Xia J J,Li H C,Kang Y Y,et al. 2020. Machine learning—Based weather support for the 2022 winter Olympics. Adv Atmos Sci,37(9):927-932 doi: 10.1007/s00376-020-0043-5 [31] Zhang H L,Pu Z X,Zhang X B. 2013. Examination of errors in near-surface temperature and wind from WRF numerical simulations in regions of complex Terrain. Wea Forecasting,28(3):893-914 doi: 10.1175/WAF-D-12-00109.1 -
下载:
点击查看大图
计量
- 文章访问数: 66
- HTML全文浏览量: 20
- PDF下载量: 16
- 被引次数: 0
