Error Analysis of the Numerical Value Forecast for Tropical Cyclone Paths with the T213L31 Model
-
摘要: 应用国家气象中心的T213L31模式对2006—2009年热带气旋路径数值预报的结果从平均距离误差、距离误差分布、模式系统性偏差、平均全移速误差、模式对背景场的预报偏差进行全面检验分析表明:对于不同走向的热带气旋 (简称TC) 路径预报而言,预报误差最小的是西行路径TC,预报误差最大的是东北行路径TC。按照TC强度进行分类统计发现,TC强度强比强度弱的路径预报平均距离误差小。T213对所有样本的平均预报而言,存在西北偏西向的系统性偏差,不同类型路径TC存在不同的系统性偏差。通过计算平均全移速可以看出,对于西北行、西行、北上类型路径在整个预报时效中移速偏差较小,而对于所有转向类路径移速偏差较大,即偏慢较明显。对T213模式预报2009年TC路径误差较大的个例 (0904,0906,0907号热带气旋) 检验模式对背景场的预报偏差,个例分析表明:模式对中高纬度西风槽预报偏强、对TC环境风场垂直切变预报偏大、对两个相近的TC预报更容易发生藤原效应,这些因素是导致TC路径偏差的主要原因。Abstract: The numerical forecast result for tropical cyclone paths from 2006 to 2009 is fully checked from mean distance error, distance error distribution, system error of the model, mean full-shift velocity error and the mode forecast error to circumstance fields by using T213L31 model of China National Meteorological Center. The analysis shows that the model forecast distance errors of 24, 48, 72, 96 hours forecasts are 135.8, 250.1, 376.3, 515.9 and 695.1 km. For the TC path of different alignment, the forecast for west path type is the best, followed by excrescent path and path of going up north. The biggest forecast error path is the path going up northeast, with the path east to change direction in second. Classifying the tropical cyclone strength indicates that the forecast error to tropical cyclone decreases as the strength increases, and the forecast is most efficient when the strength attains super typhoon. The forecast error to strong tropical storm is significant when forecast time increases, and the error to the tropical storm is also notable. The T213 produces a system deviation of northwest to west path among all average forecast samples. For northwest line and path to west cyclones the deviation is northeast to east and northeast to north. For cyclones changing direction to the northwest, middle, west and northeast, the deviation is northwest path, northwest to west and west system deviation respectively. But for the east changing direction and path going up north, the deviation is southwest to west and southwest respectively. Calculating the mean moving velocity, it's found that forecast for northeast path cyclones generates great error on initial velocity, reching up to 1.11 m/s, which leads to forecast error of distance. For the northwest, west, going up north type path cyclones the error of velocity is smaller, but for all changing direction type ones the velocity deviation is bigger. From the cases of T213 model which forecasts TC tracks with larger error in 2009 (such as No.0904, No.0906, No.0907), by examining the model prediction error of the background field, it shows that the model overestimates westerly trough in middle and higher latitude regions. For forecasts of two close cyclones, the Fujiwhara Effect can happen easily, which will cause forecast deviation for TC tracks.
-
Key words:
- T213 model;
- tropical cyclone path;
- the error analysis
-
图 5 所有TC及按走向分类的TC路径与对应的平均系统性偏差
(a) 全部,(b) 西北向,(c) 西北行登陆转向,(d) 西行, (e) 西转向, (f) 中转向, (g) 东转向, (h) 东北向, (i) 北上类
Fig. 5 The TC paths of all and those classified by walking direction and their corresponding average systematic deviation (a) all TC, (b) TC towards northwest, (c) TC walking up to northwest and then landing and returning, (d) TC walking towards west, (e) TC returning to west, (f) TC returning to middle, (g) TC returning to east, (h) TC towards northeast, (i) TC towards to north
图 6 ECMWF 500 hPa高度场2009年6月25日12:00分析场 (a) 和2009年6月23日12:00 T213模式48 h预报场 (b),ECMWF 500 hPa高度场2009年6月26日12:00分析场 (c) 和2009年6月23日12:00 T213模式72 h预报场 (d),0904号TC各时次路径预报 (e) 以及环境场预报偏差模型 (f)
Fig. 6 The 500 hPa height analysis field of ECMWF at 12:00 25 June 2009(a), T213 48 h forecast field at 12:00 23 June 2009(b), the 500 hPa height analysis field of ECMWF at 12:00 26 June 2009(c), T213 72 h forecast field at 12:00 23 June 2009(d), path forecast of No.0904 TC (e) and the deviation model of environment fields forecast (f)
图 7 2009年7月17日12:00分析场 (a) 和2009年7月16日12:00 T213 24 h预报场 (b),ECMWF 2009年7月18日12:00分析场 (c) 和2009年7月16日12:00 T213 48h预报场 (d),0906号TC各时次路径预报 (e) 以及环境场预报偏差模型 (f)
(填色部分为海平面气压场;等值线为200 hPa与850 hPa的径向风垂直切变,单位:m/s)
Fig. 7 The analysis fields of ECMWF at 12:00 17 July 2009(a), T213 24 h forecast fields at 12:00 16 July 2009(b), the analysis field of ECMWF at 12:00 18 July 2009(c), T213 48 h forecast fields at 12:00 16 July 2009(d), path forecast of No.0906 TC (e), the deviation model of environment fields forecast (f)
(shaded areas are sea level pressure field; the isopleths are the radial wind vertical shear of 200 hPa to 850 hPa, unit: m/s)
图 8 ECMWF 500 hPa高度场2009年8月7日12:00分析场 (a) 和2009年8月5日12:00 T213模式48 h预报场 (b),ECMWF 500 hPa高度场2009年8月8日12:00分析场 (c) 和2009年8月5日12:00 T213模式72 h预报场 (d),0907号TC各时次路径预报 (e) 以及环境场预报偏差模型 (f)
Fig. 8 The 500 hPa height analysis field of ECMWF at 12:00 7 August 2009(a), T213 48 h forecast field at 12:00 5 August 2009(b), the 500 hPa height analysis field of ECMWF at 12:00 8 August 2009(c), T213 72 h forecast field at 12:00 5 August 2009(d), path forecast of No. 0907 TC (e), the deviation model of environment fields forecast (f)
表 1 热带气旋不同类型路径的平均全移速偏差 (单位:m/s)
Table 1 The average full moving speed deviation of different types of TC paths (unit: m/s)
预报时效/h 西北行 西北行
登陆转向西行 西转向 中转向 东转向 东北行 北上 异常路径 总计 0~6 -0.21 0.62 0.01 0.04 0.04 -0.09 -1.11 0.31 -0.11 -0.19 6~12 0.00 0.38 -0.01 0.13 -0.11 -0.16 -1.23 0.00 -0.02 -0.23 12~18 0.33 0.35 0.10 -0.28 -0.18 -0.29 -1.19 0.05 0.11 -0.14 18~24 0.18 0.29 0.09 -0.27 -0.17 -0.49 -1.15 0.08 0.18 -0.10 24~30 0.04 0.44 -0.03 -0.57 -0.30 -0.54 -1.21 -0.10 0.03 -0.23 30~36 0.00 0.64 -0.01 -0.90 -0.34 -0.71 -1.43 0.12 -0.13 -0.30 36~42 -0.04 0.02 0.07 -0.87 -0.36 -0.69 -1.67 0.09 -0.14 -0.33 42~48 -0.05 -0.58 0.04 -0.99 -0.46 -0.85 -1.61 0.35 -0.27 -0.40 48~54 0.05 -0.79 0.11 -0.89 -0.45 -1.32 -1.21 0.24 -0.30 -0.40 54~60 0.13 -0.78 0.07 -0.81 -0.54 -1.49 -1.17 0.13 -0.22 -0.42 60~66 0.09 -0.80 0.10 -0.78 -0.29 -1.56 -1.06 0.48 -0.32 -0.36 66~72 0.15 -0.54 0.01 -0.75 -0.24 -1.75 -0.65 0.25 -0.28 -0.36 72~78 -0.22 -1.32 0.26 -0.67 -0.36 -1.90 -0.46 0.00 -0.13 -0.44 78~84 -0.14 -1.41 0.44 -1.04 -0.39 -1.75 -0.65 0.21 0.01 -0.43 84~90 -0.18 -1.53 0.14 -1.00 -0.46 -2.15 -1.41 -0.08 -0.10 -0.59 90~96 -0.26 -2.39 0.09 -1.02 -0.21 -2.15 -0.86 0.41 -0.12 -0.54 96~102 -0.22 -2.66 0.23 -1.39 0.25 -2.64 -1.68 0.66 -0.12 -0.57 102~108 -0.32 -3.39 -0.11 -1.50 0.16 -2.66 -1.85 0.37 0.08 -0.68 108~114 -0.50 -3.59 0.39 -1.41 0.46 -3.34 -2.71 -0.13 0.21 -0.70 114~120 -0.74 -3.01 0.20 -1.17 0.69 -3.95 -3.34 -0.22 0.13 -0.68 -
[1] 王国民, 王诗文, 李建军.一个人造台风方案及其在移动套网格模式中的应用.热带气象学报, 1996, 12(1):9-17. http://www.cnki.com.cn/Article/CJFDTOTAL-RDQX601.001.htm [2] 王诗文.国家气象中心台风数值模式的改进及其应用试验.应用气象学报, 1999, 10(3):347-353. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=19990384&flag=1 [3] 陈德辉, 王诗文, 汪厚君.改进的物理过程参数化对台风路径数值预报的影响.应用气象学报, 1996, 7(1):1-8. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=19960101&flag=1 [4] 瞿安祥, 麻素红, Liu Qingfu, 等.全球数值模式中的台风初始化Ⅰ:方案设计.气象学报, 2009, 67(5):716-726. doi: 10.11676/qxxb2009.072 [5] 瞿安祥, 麻素红, 李娟, 等.全球数值模式中的台风初始化Ⅱ:业务应用.气象学报, 2009, 67(5):727-735. doi: 10.11676/qxxb2009.073 [6] 瞿安祥, 麻素红.非对称台风bougs方案设计和初步试验.应用气象学报, 2007, 18(3):380-387. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20070361&flag=1 [7] 陈起英, 姚明明, 王雨.国家气象中心新一代业务中期预报模式T213L31的主要特点.气象, 2004, 30(10):16-21. doi: 10.3969/j.issn.1000-0526.2004.10.004 [8] Jerry D J. An analysis of Western North Pacific tropical cyclone forecast errors. Mon Wea Rev, 1978, 7:925-937. http://adsabs.harvard.edu/abs/1978MWRv..106..925J [9] 陈联寿, 丁一汇.西太平洋台风概论.北京:科学出版社, 1979. [10] 麻素红, 瞿安祥, 张眙.台风路径数值预报模式的并行化及路径预报误差分析.应用气象学报, 2004, 15(3):322-328. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20040341&flag=1 [11] Lester E Carr Ⅲ, Russell L Elsberry. Dynamical tropical cyclone track forecast errors. Part Ⅱ: Mid-latitude circulation influences. Weather and Forecasting, 2000, 15: 662-681. doi: 10.1175/1520-0434(2000)015<0662:DTCTFE>2.0.CO;2 [12] Tracy J D. Accuracy of Atlantic tropical cyclone forecasts. Mon Wea Rev, 1966, 94: 407-418. doi: 10.1175/1520-0493(1966)094<0407:AOATCF>2.3.CO;2 [13] 朱振铎, 端义宏, 陈德辉.GRAPES2TCM业务试验结果分析.气象, 2001, 33(7):44-54. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200707006.htm [14] 丑纪范, 任宏利.数值天气预报——另类途径的必要性和可行性.应用气象学报, 2006, 17(2):240-244. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20060240&flag=1 [15] Chan J C L. On the Physical Processes Responsible for Tropical Cyclone Motion. Atmos Sci Paper No.358, Dept of Atmos Sci Colorado State Univ, 1982. [16] Carr L E Ⅲ, Elsberry R L. Dynamical tropical cyclone track forecast errors. Part Ⅰ: Tropical region errors. Wea Forecasting, 2000, 15:641-661. doi: 10.1175/1520-0434(2000)015<0641:DTCTFE>2.0.CO;2 [17] Carr L E Ⅲ, Elsberry R L. Dynamical tropical cyclone track forecast errors. Part Ⅱ: Midlatitude circulation influences. Wea Forecasting, 2000, 15: 662-681. doi: 10.1175/1520-0434(2000)015<0662:DTCTFE>2.0.CO;2 [18] 马红云, 马镜娴, 罗哲贤.切向风速水平廓线对台风路径和强度的影响.南京气象学院学报, 2003, 26(6):780-787. http://cdmd.cnki.com.cn/Article/CDMD-10300-2003063784.htm [19] 周嘉陵, 罗哲贤.台风最大风速增强的数值研究.南京气象学院学报, 2002, 25(2):159-165. http://www.cnki.com.cn/Article/CJFDTOTAL-NJQX200202008.htm