Study on Surface Temperature Error Correction Method Based on the Terrain Height Deviation of CMA-TRAMS Model
-
摘要: 采用一元线性方法建立南海台风模式CMA-TRAMS地形高度偏差和地面气温预报误差的回归关系,分别开展不分级、高度偏差分级和地面气温误差分级的三种订正方法的研究,并进行订正效果评估。结果表明,模式地面气温预报误差与地形高度偏差总体呈负的线性相关关系,地面气温预报绝对误差随地形高度偏差绝对值增大而增大(对模式地形高度偏低站点尤为明显),但不同时刻地面气温预报误差特征表现不同,模式对地形高度偏高(即模式地形高于测站高度)和地形高度偏差小于50 m的站点,06时地面气温(世界时,下同)预报总体偏低,对地形高度偏低大于50 m的站点(即模式地形低于测站高度),06时地面气温预报总体偏高;而无论站点地形高度偏差如何,模式对18时地面气温预报总体偏高。三种订正方法中地面气温误差分级法能有效地减小地面气温预报误差,该方法订正后的分析场准确率可达96%~99%,12~48小时时效预报场准确率总体可提升至90% 以上,该方法具有回归关系稳定、效果显著、适用性广、简单易行等特点。
-
关键词:
- 南海台风模式CMA-TRAMS /
- 模式地形高度偏差 /
- 地面气温 /
- 误差分级订正
Abstract: Based on the regression relationship between the terrain height deviation of CMA-TRAMS (China Meteorological Administration - Tropical Region Atmospheric Modeling for South China Sea) and the surface temperature error, a unitary linear regression method is utilized to evaluate three correction methods in this study, including non-grading, height deviation grading and temperature error grading, respectively, and the correction effect was evaluated. It is demonstrated that a negative linear correlation can be found between the surface temperature prediction error and the terrain height deviation, the absolute error of surface temperature forecast increases with the increase of the absolute value of terrain height deviation (especially obvious for sites with lower terrain height). but the error characteristics of surface temperature at different times are different, the stations with higher terrain height and the terrain height deviation is less than 50 m in the model usually have lower surface temperature forecast at 06:00, and vice versa. However, the prediction of surface temperature at 18:00 is generally higher, regardless of the terrain height deviation. The temperature error grading correction method of the three correction methods has the best performance in effectively reducing the error of surface temperature forecast, the accuracy rate of the temperature prediction of analysis field is 96%—99%, and 12—48 h forecast field is generally improved to above 90%. The temperature error grading correction method has the characteristics of stable regression relationship, obvious effect, wide applicability and easy operation. -
表 1 三种订正方法的模式地形高度偏差与2020年12月06时、18时不同预报时效地面气温误差的线性回归关系
时效 不分级订正法 高度偏差分级订正法 地面气温误差分级订正法 (-∞,-50) [-50, 50] (50,+∞) (-∞,-2) [-2, 2] (2,+∞) 06时 回归系数 00 -0.004 -0.005 -0.004 -0.003 0.000 -0.001 -0.002 12 -0.005 -0.006 -0.000 -0.004 -0.001 -0.001 -0.002 24 -0.005 -0.005 -0.003 -0.003 -0.001 -0.001 -0.002 36 -0.004 -0.006 0.002 -0.003 -0.001 -0.001 -0.002 48 -0.005 -0.006 -0.001 -0.004 -0.001 -0.001 -0.002 常数项 00 0.776 1.216 0.320 0.537 -3.149 0.141 3.498 12 0.603 1.013 -0.119 0.517 -3.098 0.049 3.681 24 0.974 1.366 0.435 0.785 -3.184 0.159 3.736 36 1.578 1.912 0.734 1.544 -3.070 0.235 4.140 48 1.249 1.517 0.703 1.197 -3.019 0.174 3.924 18时 回归系数 00 -0.003 -0.003 -0.002 -0.002 0.000 -0.001 -0.002 12 -0.003 -0.003 -0.007 -0.002 -0.001 -0.001 -0.001 24 -0.003 -0.003 -0.003 -0.002 -0.000 -0.001 -0.001 36 -0.003 -0.003 -0.009 -0.002 -0.000 -0.001 -0.001 48 -0.003 -0.003 -0.003 -0.002 -0.000 -0.001 -0.001 常数项 00 1.085 1.662 0.573 0.873 -3.030 0.256 3.386 12 0.645 1.213 0.409 0.347 -2.837 0.152 3.236 24 1.019 1.596 0.604 0.757 -2.903 0.192 3.529 36 0.719 1.339 0.543 0.377 -3.023 0.153 3.438 48 1.537 2.102 1.023 1.331 -2.805 0.281 3.821 -
[1] 任宏利, 丑纪范. 数值模式的预报策略和方法研究进展[J]. 地球科学进展, 2007, 22(4): 376-385. [2] SURANJANA S. Response of the NMC MRF model to systematic-error correction within integration[J]. Mon Wea Rev, 1992, 120(2): 345-360. [3] 吴启树, 韩美, 郭弘, 等. MOS温度预报中最优训练期方案[J]. 应用气象学报, 2016, 27(4): 426-434. [4] 李佰平, 智协飞. ECMWF模式地面气温预报的四种误差订正方法的比较研究[J]. 气象, 2012, 38(8): 897-902. [5] 王丹, 王建鹏, 白庆梅, 等. 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[J]. 气象, 2019, 45(9): 1 310-1 321. [6] 吴国雄, 王军, 刘新, 等. 欧亚地形对不同季节大气环流影响的数值模拟研究[J]. 气象学报, 2005, 63(5): 603-612. [7] 陶诗言. 中国之暴雨[M]. 北京: 科学出版社, 1980, 13. [8] 赵滨, 李子良, 张博, 等. 三维插值方法在2m温度评估中的应用[J]. 南京信息工程大学学报(自然科学版), 2016, 8(4): 343-355. [9] SUN R H, ZHANG B P. Topographic effects on spatial pattern of surface air temperature in complex mountain environment[J]. Environmental Earth Sciences, 2016, 75(7): 1-12. [10] 智协飞, 吴佩, 俞剑蔚, 等. GFS模式地形高度偏差对地面2 m气温预报的影响[J]. 大气科学学报, 2019, 42(5): 652-659. [11] 陈德辉, 沈学顺. 新一代数值预报系统GRAPES研究进展[J]. 应用气象学报, 2006, 17(6): 773-777. [12] 徐道生, 陈子通, 钟水新, 等. 对流参数化与微物理过程的耦合及其对台风预报的影响研究[J]. 气象学报, 2014, 72(2): 337-349. [13] 徐道生, 陈子通, 戴光丰, 等. 对流参数化方案的改进对GRAPES模式台风预报的影响研究[J]. 热带气象学报, 2014, 30(2): 831-840. [14] 陈子通, 戴光丰, 钟水新, 等. 中国南海台风模式(TRAMS-V2.0)技术特点及其预报性能[J]. 热带气象学报, 2016, 32(6): 210-218. [15] 林晓霞, 冯业荣, 陈子通, 等. 华南区域高分辨率数值模式前汛期预报初步评估[J]. 热带气象学报, 2021, 37(4): 656-668. [16] 中国气象局. 全国智能网格气象预报业务规定(试行)[S]. 北京: 中国气象局, 2017. -