利用CMORPH卫星与中国30000余个自动站逐时降水融合资料,基于面向对象模式诊断分析MODE方法研究了降水对象的客观表现以及日本细网格模式对2012-2013年中国34个降水个例的预报能力,并与传统TS、ETS检验方法进行了对比分析。结果表明:模式整体上能较好地捕获降水对象,但不同卷积半径下对象的个数及特征差异明显;模式对局地性或弱降水事件预报偏多(偏大),对区域性或强降水事件预报偏少(偏弱);模式较好地反映不同卷积半径下,降水对象在10 mm左右识别对象的面积、数量的变化特性,不足之处在于模式的固定阈值更为集中,识别对象的数量减少幅度偏快;采用较大半径进行卷积,有助于强降水区域的形态判定,在不同量级降水情况下存在提高收益函数的合适的卷积半径;中、低纬降水预报的总收益整体优于高纬,但不同属性呈现较大差异,具体而言,高纬地区降水强度预报更为合理,而低纬降水面积预报优于高纬;与传统TS、ETS检验相比,MODE方法受气候概率影响较小,其检验结果更加客观、多样。
Using CMORPH (NOAA Climate Prediction Center Morphing Method) and fusion datas of hourly precipitation from more than 30 thousand automatic stations, based on MODE (Method for Object-Based Diagnostic Evaluation) object matching technology, the forecasting ability of Japan high resolution model of 34 precipitation cases in China from 2012 to 2013 was studied and compared with the traditional TS and ETS skill scores. The result shows that:first, the model can generally forecast the rainfall, but the number and characteristics of objects significant differences in different convolution radius. Second, the model gives much forecast of local or weak precipitation cases and limited forecast of regional or heavy precipitation. Third, the model can better tell variation characteristics of the area and the quantity when different convolution radii exist and the rainfall is about 10 mm. The disadvantage of the model is that the fixed threshold is much more centralized and the number of identified cases decrease quickly. If bigger radius are used to convolution, the model can identify the pattern better in heavy rainfall regions. There are possibilities to exist a proper convolution radius to improve total interest in different cases of different magnitude of rainfall. Fourth, the forecasting reward generally is bigger in the low-and mid-latitude areas than in high-latitude regions. However, the model shows no stable abilities to forecast different aspects of the rainfall. Specifically speaking, forecasting rainfall intensity in high-latitudes regions is more reasonable, while in low latitudes forecasting the rainfall area is superior to that in high latitudes regions. Compared with the traditional TS, ETS verification, MODE method is less affected by climate probability, the verify results are more objective and diversity.
[1]Baldwin M E,Kain J S. 2006. Sensitivity of several performance measures to displacement error,bias,and event frequency[J].Wea Forecasting,21(4):636-648.
[2]Casati B,Ross D,Stephenson B. 2004. A new intensity-scale approach for the verification of spatial precipitation forecasts[J].Met Apps,11(2):141-154.
[3]Chakraborty A. 2010. The skill of ECMWF medium-range forecasts during the year of tropical convection[J]. Mon Wea Rev,138(10):3787-3805.
[4]Davis C A,Brown B,Bullock R G. 2006a. Object-based verification of precipitation forecasts. Part I:Methods and application to mesoscale rain areas[J]. Mon Wea Rev,134(7):1772-1784.
[5]Davis C A,Brown B G,Bullock R G. 2006b. Object-based verification of precipitation forecasts,Part II:Application to convective rain systems[J]. Mon Wea Rev,134(7):1785-1795.
[6]Gilleland E,David A,Brown G. 2009. Intercomparison of spatial forecast verification methods[J]. Wea Forecasting,24(5):1416-1429.
[7]Hoffman R N,Liu Z,Louis J F,et al. 1995. Distortion representation of forecast errors[J]. Mon Wea Rev,123(9):2758-2770.
[8]Johnson A,Wang X,Kong F. 2013. Object-based evaluation of the impact of horizontal grid spacing on convection-allowing forecasts[J]. Mon Wea Rev,141(10):3413-3425.
[9]Mass C F,Ovens D,Westrick K,et al. 2002. Does increasing horizontal resolution produce more skillful forecasts?[J] Bull Amer Meteor Soc,83(3):407-430.
[10]Tustison B,Foufoula E,Harris D. 2003. Scale-recursive estimation for multisensor quantitative precipitation forecast verification:A preliminary assessment[J]. J Geophys Res,108(8):8377-8391. DOI:10.1029/2001JD001073.
[11]陈海山,孙照渤. 2005. 陆面模式CLSM的设计及性能检验II:模式检验[J]. 大气科学,29(2):272-282.Chen Haishan,Sun Zhaobo. 2005. Design of a comprehensive land surface model and its validation part II. Model validation[J]. Chinese J Atmos Sci,29(2):272-282.
[12]公颖,王叶红,赖安伟. 2010. 三个模式对2008年夏半年西南区降水预报效果的检验[J]. 高原气象,29(6):1441-1451. Gong Ying,Wang Yehong,Lai Anwei. 2010. Verification of three model precipitation forecast performances in Southwest China in summer half year,2008[J]. Plateau Meteor,29(6):1441-1451.
[13]郝民,徐枝芳,陶士伟,等. 2011. GRAPES RUC系统模拟研究及应用试验[J]. 高原气象,30(6):1573-1583. Hao Ming,Xu Zhifang,Tao Shi Wei,et al. 2011. Simulation study and application experiment of GRAPES RUC system[J]. Plateau Meteor,30(6):1573-1583.
[14]何由,阳坤,姚檀栋,等. 2012. 基于WRF模式对青藏高原一次强降水的模拟[J]. 高原气象,31(5):1183-1191. He You,Yang Kun,Yao Chandong,et al. 2012. Numerical simulation of a heavy precipitation in Qinghai-Xizang Plateau based on WRF model[J]. Plateau Meteor,31(5):1183-1191.
[15]李虎超,邵爱梅,何邓新,等. 2015. BP神经网络在估算模式非系统性预报误差中的应用[J]. 高原气象,34(6):1751-1757. Li Huchao,Shao Aimei,He Dengxin,et al. 2015. Application of back-propagation neural network in predicting non-systematic error in numerical prediction model[J]. Plateau Meteor,34(6):1751-1757. DOI:10.7522/j.issn. 1000-0534.2014.00120.
[16]刘佳,徐金霞,马振峰,等. 2014. 第二代月动力延伸预报产品对西南汛期降水的预报检验[J]. 高原气象,33(6):1468-1479. Liu Jia,Xu Jin Xia,Ma Zhengfeng,et al. 2014. Evaluation of flood season prediction in Southwest Region by the second-generation monthly dynamic extended-range forecast[J]. Plateau Meteor,33(6):1468-1479. DOI:10.7522/j.issn. 1000-0534.2013.00168.
[17]潘留杰,张宏芳,王建鹏,等. 2014a. 日本高分辨率模式对中国降水预报能力的客观检验[J]. 高原气象,33(2):483-494.Pan Liujie,Zhang Hongfang,Wang Jianpeng,et al. 2014a.An objective verification of forecasting ability of Japan high-resolution model precipitation in China[J]. Plateau Meteor,33(2):483-494. DOI:10.7522/j.issn. 1000-0534.2012.00188.
[18]潘留杰,张宏芳,王建鹏. 2014b. 数值天气预报检验方法研究进展[J]. 地球科学进展,29(3):327-335. Pan Liujie,Zhang Hongfang,Wang Jianpeng. 2014b. Progress on verification methods of numerical weather prediction[J]. Adv Earth Sci,29(3):327-335.
[19]彭新东,常燕,王式功. 2010. GRAPES模式对2008年两次强降水过程的数值预报检验[J]. 高原气象,29(2):321-330. Peng Xindong,Chang Yan,Wang Shigong. 2010. Numerical validation of GRAPES model with two severe precipitation processes in 2008[J]. Plateau Meteor,2010,29(2):321-330.
[20]石彦军,任余龙,王式功,等. 2012. BCC_CSM气候模式对中国区域气候变化模拟能力的检验[J]. 高原气象,31(5):1257-1267. Shi Yanjun,Reng Yulong,Wang Shigong,et al. 2012. Verification of simulation ability of BCC_CSM climate model in regional climate change in China[J]. Plateau Meteor,31(5):1257-1267.
[21]王叶红,赵玉春,李红莉,等. 2011. AREM模式的热启动数值模拟-以2007年7月13日暴雨过程为例[J]. 高原气象,30(6):1488-1504. Wang Yehong,Zhao Yuchun,Li Hongli,et al. 2011. A numerical study on hot start AREM Model-taking rainstorm process on 13 July 2007 as a case[J]. Plateau Meteor,30(6):1488-1504.
[22]王颖,张镭,胡菊,等. 2010. WRF模式对山谷城市边界层模拟能力的检验及地面气象特征分析[J]. 高原气象,29(6):1397-1407. Wang Ying,Zhang Lei,Hu Ju,et al. 2010. Verification of WRF simulation capacity on PBL characteristic and analysis of surface meteorological characteristic over complex terrain[J]. Plateau Meteor,29(6):1397-1407.
[23]许建伟,高艳红. 2014. WRF模式对夏季黑河流域气温和降水的模拟及检验[J]. 高原气象,33(4):937-946. Xu JianWei,Gao Yan Hong. 2014. Validation of summer surface air temperature and precipitation simulation over Heihe river basin[J]. Plateau Meteor,33(4):937-946. DOI:10.7522/j.issn. 1000-0534.2013.00149.
[24]张宏芳,潘留杰,杨新. 2014. ECMWF、日本高分辨率模式降水预报能力的对比分析[J]. 气象,40(4):424-432. Zhang Hongfang,Pan Liujie,Yang Xin. 2014. Comparative analysis of precipitation forecasting capabilities of ECMWF and Japan high-resolution models[J]. Meteor Mon,40(4):424-432.
[25]张利红,何光碧. 2014. GRAPES_Meso模式对2011年夏季青藏高原东部及周边区域的预报检验[J]. 高原气象,33(1):14-25. Zhang Lihong,He Guangbi. 2014. Verification for GRAPE _Meso model in Eastern Qinghai-Xizang Plateau and its surroundings in summer of 2011[J]. Plateau Meteor,33(1):14-25. DOI:10.7522/j.issn. 1000-0534.2012.00175.
