BP神经网络在估算模式非系统性预报误差中的应用
|
摘要
基于数值天气预报误差在时间上的相依性,采用BP神经网络方法建立预测数值模式非系统性预报误差的模型,并利用2003-2007年T213模式分析场和24 h高度预报场资料验证了该模型的预测能力,结果表明:所建立的3层BP神经网络模型对未来24 h的非系统性预报误差有较好的预估能力,对大多数样本而言所估测的非系统性预报误差的分布特征和其真值较为一致。BP神经网络模型估测的非系统性预报误差可以在系统性预报误差订正的基础上进一步对预报做出修正,其订正效果好于仅进行系统性预报误差订正的效果。
Based on the temporal dependence of forecast errors derived from numerical weather prediction,the back-propagation(BP) neural network is used to establish the prediction model for predicting non-systematic forecast error.The effectiveness of this model is tested with the analysis and 24-hours forecast data produced by T213 model from 2003 to 2007.The results show that the established BP neural network model has a good ability on predicting non-systematic error in the next 24 hours.For most of 332 test samples,the spatial distribution of the predicted non-systematic errors is consistent with the truth.The non-systematic error estimated by BP neural network model can further correct forecasts on the basis of the systematic error correction,and its correction effectiveness is better than that of the systematic error correction only.For 332 test samples,the effective rate of systematic error correction on forecasts is 61%,but the effective rate of nonsystematic error further correction can increase to 82%.
Based on the temporal dependence of forecast errors derived from numerical weather prediction,the back-propagation(BP) neural network is used to establish the prediction model for predicting non-systematic forecast error.The effectiveness of this model is tested with the analysis and 24-hours forecast data produced by T213 model from 2003 to 2007.The results show that the established BP neural network model has a good ability on predicting non-systematic error in the next 24 hours.For most of 332 test samples,the spatial distribution of the predicted non-systematic errors is consistent with the truth.The non-systematic error estimated by BP neural network model can further correct forecasts on the basis of the systematic error correction,and its correction effectiveness is better than that of the systematic error correction only.For 332 test samples,the effective rate of systematic error correction on forecasts is 61%,but the effective rate of nonsystematic error further correction can increase to 82%.
引文
[1]Danforth C M,Kalnay E,Miyoshi T.Estimating and Correcting Global Weather Model Error[J].Mon Wea Rev,2006,135(2):281-282.
[2]Leith C E.Objective methods for weather prediction[J].Annu Rev Fluid Mech,1978,10(1):107-128.
[3]DelSole T,Hou A Y Empirical correction of a dynamical model.Part I:Fundamental issues[J].Mon Wea Rev,1999,127(11):2533-2545.
[4]Danforth C M,Kalnay E.Using Singular Value Decomposition to Parameterize State-Dependent Model Errors[J].J Atmos Sci,2008,65(4):1467-1478.
[5]D'Andrea F,Vautard R.Reducing systematic errors by empirically correcting model errors[J].Tellus A,2000,52(1):21-41.
[6]Qiu C J,Chou J F.A new approach to improve the numerical weather prediction[J].Science in China(Ser B),1987,17(8):903-909.
[7]邱崇践,丑纪范.天气预报的相似—动力方法[J].大气科学,1989,13(1):22-28.
[8]Zeng Qingcun,Zhang Banglin,Yuan Chongguang,et al.A note on some methods suitable for verifying and correcting the prediction of climate anomaly[J].Adv Atmos Sci,1994,11(2):121-127.
[9]邵爱梅,希爽,邱崇践.修正数值天气预报的非系统性误差的变分方法[J].中国科学(D辑).2009,39(2):235-244.
[10]邵月红,张万昌,刘永和,等.BP神经网络在多普勒雷达降水量的估测中的应用[J].高原气象,2009,28(4):846-853.
[11]马学款,普布次仁,唐叔乙,等,人工神经网络在西藏中短期温度预报中的应用[J].高原气象,2007,26(3):491-495.
[12]金龙,吴建生,林开平,等.基于遗传算法的神经网络短期气候预测模型[J].高原气象,2005,24(6):981-987.
[13]金龙,秦伟良.多步预测的小波神经网络预报模型[J].大气科学,2000,24(1):79-86.
[14]金龙.人工神经网络技术发展及在大气科学领域的应用[J].气象科技,2004,32(6):385-392.
[15]刘洪兰,张俊国,阙龙凯,等.基于BP神经网络的张掖国家湿地公园水域结冰厚度预报模型[J].高原气象,2014,33(3):832-837,doi:10.7522/j.issn.1000-0534.2013.00051.
[16]金龙,罗莹,王业宏,等.月降水量的神经网络混合预报模型研究[J].高原气象,2003,22(6):618-623.
[17]刘亚亚,毛节泰,刘钧,等.地基微波辐射计遥感大气廓线的BP神经网络反演方法研究[J].高原气象,2010,29(6):1514-1523.
[18]胡春梅,陈道劲,于润玲.BP神经网络和支持向量机在紫外线预报中的应用[J].高原气象,2010,29(2):539-544.
[19]Box G E,Jenkins G M.Time Scres Analysis:Forecasting and Control[M].San Francisco:Holden-Day,1976.
[2]Leith C E.Objective methods for weather prediction[J].Annu Rev Fluid Mech,1978,10(1):107-128.
[3]DelSole T,Hou A Y Empirical correction of a dynamical model.Part I:Fundamental issues[J].Mon Wea Rev,1999,127(11):2533-2545.
[4]Danforth C M,Kalnay E.Using Singular Value Decomposition to Parameterize State-Dependent Model Errors[J].J Atmos Sci,2008,65(4):1467-1478.
[5]D'Andrea F,Vautard R.Reducing systematic errors by empirically correcting model errors[J].Tellus A,2000,52(1):21-41.
[6]Qiu C J,Chou J F.A new approach to improve the numerical weather prediction[J].Science in China(Ser B),1987,17(8):903-909.
[7]邱崇践,丑纪范.天气预报的相似—动力方法[J].大气科学,1989,13(1):22-28.
[8]Zeng Qingcun,Zhang Banglin,Yuan Chongguang,et al.A note on some methods suitable for verifying and correcting the prediction of climate anomaly[J].Adv Atmos Sci,1994,11(2):121-127.
[9]邵爱梅,希爽,邱崇践.修正数值天气预报的非系统性误差的变分方法[J].中国科学(D辑).2009,39(2):235-244.
[10]邵月红,张万昌,刘永和,等.BP神经网络在多普勒雷达降水量的估测中的应用[J].高原气象,2009,28(4):846-853.
[11]马学款,普布次仁,唐叔乙,等,人工神经网络在西藏中短期温度预报中的应用[J].高原气象,2007,26(3):491-495.
[12]金龙,吴建生,林开平,等.基于遗传算法的神经网络短期气候预测模型[J].高原气象,2005,24(6):981-987.
[13]金龙,秦伟良.多步预测的小波神经网络预报模型[J].大气科学,2000,24(1):79-86.
[14]金龙.人工神经网络技术发展及在大气科学领域的应用[J].气象科技,2004,32(6):385-392.
[15]刘洪兰,张俊国,阙龙凯,等.基于BP神经网络的张掖国家湿地公园水域结冰厚度预报模型[J].高原气象,2014,33(3):832-837,doi:10.7522/j.issn.1000-0534.2013.00051.
[16]金龙,罗莹,王业宏,等.月降水量的神经网络混合预报模型研究[J].高原气象,2003,22(6):618-623.
[17]刘亚亚,毛节泰,刘钧,等.地基微波辐射计遥感大气廓线的BP神经网络反演方法研究[J].高原气象,2010,29(6):1514-1523.
[18]胡春梅,陈道劲,于润玲.BP神经网络和支持向量机在紫外线预报中的应用[J].高原气象,2010,29(2):539-544.
[19]Box G E,Jenkins G M.Time Scres Analysis:Forecasting and Control[M].San Francisco:Holden-Day,1976.