基于CLDAS的格点温度预报偏差订正方法

doi:10.11755/j.issn.1006-7639(2021)-05-0847

摘要/Abstract

摘要：

利用国家气象信息中心CLDAS格点温度实况、中央气象台SCMOC格点温度预报以及山西省站点观测温度,采用非独立性检验综合评估CLDAS在山西区域的适用性。在此基础上,采用滑动训练期订正方案,基于格点实况开展SCMOC温度预报场的客观订正。结果表明:(1)复杂地形对山西CLDAS格点温度实况的精度有一定影响,但最高气温的分析精度优于最低气温,表明地形对最低气温的偏差影响更显著,高海拔地区CLDAS最低气温一般对应为负偏差,低海拔地区一般对应为正偏差。(2)CLDAS格点温度实况的偏差空间分布具有时间延续性,进行简单的系统偏差订正后,最高、最低气温格点实况的精度分别提升1.1%、9.7%,与站点观测更为吻合。(3)基于改进后的CLDAS格点温度实况,采用滑动偏差订正方案,显著改善了山西省SCMOC温度预报的准确率。2019年,滑动偏差订正后的24 h时效最高、最低气温预报准确率较SCMOC温度预报分别提升2.7%、4.7%,订正后的短期温度预报质量有较大提高,优于预报员主观预报。

关键词: 温度, CLDAS, SCMOC, 偏差订正, 滑动训练期

Abstract:

Based on CLDAS grid temperature data from National Meteorological Information Center of China, SCMOC grid temperature forecast data from Central Meteorological Observatory of China and temperature observation data at weather stations of Shanxi Province, the applicability of CLDAS temperature in Shanxi Province was evaluated comprehensively by using non-independence test method. And on this basis, based on CLDAS grid temperature data, the objective correction of SCMOC temperature forecast field was studied by using the sliding training period scheme. The results are as follows: (1) The complex terrain in Shanxi Province had a certain influence on the accuracy of CLDAS temperature, and the maximum temperature of CLDAS exhibited a better accuracy than the minimum temperature of CLDAS, which indicated that the influence of terrain on deviation of the minimum temperature was more significant, and the deviation of the minimum temperature in high altitude areas was negative generally, while that in low altitude areas was positive. (2) The deviation of CLDAS grid temperature had a continuity of time in space. After the simple deviation correction, the accuracy of the maximum and minimum temperature of CLDAS promoted by 1.1% and 9.7%, respectively, the revised temperatures were more consistent with observation. (3) Based on improved CLDAS grid temperature, the accuracy rate of SCMOC temperature forecast improved significantly by using the sliding deviation correction scheme. Compared to SCMOC, the accuracy rate of the 24-hour maximum and minimum temperature forecast in Shanxi Province in 2019 respectively increased by 2.7% and 4.7% after the sliding deviation correction. The quality of short-term temperature forecast after the sliding deviation correction had greatly improved, and it was superior to the subjective forecast of forecasters.

Key words: temperature, CLDAS, SCMOC, deviation correction, sliding training period

中图分类号:

P456.7

董春卿,郭媛媛,张磊,胡嘉缨. 基于CLDAS的格点温度预报偏差订正方法[J]. 干旱气象, 2021, 39(5): 847-856.

DONG Chunqing,GUO Yuanyuan,ZHANG Lei,HU Jiaying. Deviation Correction Method of Grid Temperature Prediction Based on CLDAS Data[J]. Journal of Arid Meteorology, 2021, 39(5): 847-856.

图/表 12

图1 山西省地形高度(阴影,单位:m)和气象站点(点和十字)分布

Fig.1 The distribution of terrain height (shadows, Unit: m) and meteorological stations (dots and crosses) in Shanxi Province

图2 滑动训练期订正方案示意图

Fig.2 Schematic diagram of the sliding training period correction scheme

图3 技术路线示意图

Fig.3 Schematic diagram of technical route

图4 山西省CLDAS最高、最低气温的平均误差(a)、均方根误差(b)和准确率(c)月际分布

Fig.4 The monthly distribution of ME (a), RMSE (b) and AR (c) of the maximum and minimum temperature of CLDAS in Shanxi Province

图5 2019年山西省CLDAS日最低气温平均偏差逐月分布

Fig.5 The monthly distribution of mean error of the daily minimum temperature based on CLDAS data in Shanxi Province in 2019

图6 2019年山西省CLDAS日最高气温平均偏差逐月分布

Fig.6 The monthly distribution of mean error of the daily maximum temperature based on CLDAS data in Shanxi Province in 2019

表1 山西省CLDAS各月及年Tmax和Tmin平均偏差与测站海拔高度的相关系数

Tab.1 The correlation coefficients between ME of CLDAS Tmax, Tmin and altitude of stations in each month and the whole year in Shanxi Province

时间	T_min	T_max
1月	-0.31^*	0.19
2月	-0.30^*	0.24
3月	-0.35^**	0.13
4月	-0.31^*	0.06
5月	-0.35^**	0.15
6月	-0.32^**	0.15
7月	-0.33^**	-0.02
8月	-0.31^*	-0.08
9月	-0.30^*	-0.11
10月	-0.30^*	-0.04
11月	-0.34^**	-0.11
12月	-0.30^*	0.18
年平均	-0.32^**	0.06

图7 系统偏差订正前后山西省各月CLDAS最高、最低气温格点实况的平均误差(a)、均方根误差(b)和准确率(c)对比

Fig.7 The comparison of ME (a), RMSE (b) and AR (c) of the CLDAS maximum and minimum temperature in each month before and after system deviation correction in Shanxi Province

图8 系统偏差订正前(a、c)、后(b、d)山西省CLDAS的1月Tmin(a、b)和7月Tmax(c、d)平均误差空间分布

Fig.8 The spatial distribution of ME of Tmin in January (a, b) and Tmax in July (c, d) of CLDAS data before (a, c) and after (b, d) system deviation correction in Shanxi Province

图9 不同滑动训练期对应的山西省2019年各月SCMOC 24 h时效Tmax(a)、Tmin(b)预报的客观订正 (各月条形图从左至右依次为原始值及其前1~18 d的订正值,下同)

Fig.9 The objective correction of 24-hour Tmax (a) and Tmin (b) forecast of SCMOC under different sliding training periods in each month of 2019 in Shanxi Province (The bars from left to right in each month are original value and its corrected value on previous 1 to 18 days, respectively, the same as below)

图10 改进后的不同滑动训练期下山西省2019年各月SCMOC 24 h时效Tmax(a)、Tmin(b)预报的客观订正

Fig.10 The objective correction of 24-hour Tmax (a) and Tmin (b) forecast of SCMOC after the improvement under different sliding training periods in each month of 2019 in Shanxi Province

表2 SCMOC预报及两种订正方案在山西省2019年日最高、最低气温的预报效果检验

Tab.2 The effect test of the daily maximum and minimum temperature forecast of SCMOC and two revised schemes in 2019 in Shanxi Province

要素	时效/h	预报准确率/%			均方根误差/℃
要素	时效/h	SCMOC	预报员	客观订正	SCMOC	预报员	客观订正
T_max	24	68.5	73.2	71.2	2.23	2.03	1.9
	48	61.7	67.0	66.0	2.55	2.38	2.17
	72	56.6	61.3	62.3	2.83	2.67	2.41
T_min	24	65.1	67.2	69.8	2.35	2.07	1.81
	48	62.3	65.8	68.3	2.44	2.23	1.87
	72	59.5	62.1	65.5	2.59	2.43	2.01

参考文献 26

[1]	潘留杰, 薛春芳, 王建鹏, 等. 一个简单的格点温度预报订正方法[J]. 气象, 2017, 43(12):1584-1593.
[2]	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.
[3]	杨关盈, 邓学良, 翟菁, 等. 安徽地区多模式空气污染预报效果检验与订正[J]. 气象与环境科学, 2021, 44(5):17-25.
[4]	吴振玲, 潘璇, 董昊, 等. 天津市多模式气温集成预报方法[J]. 应用气象学报, 2014, 25(3):293-301.
[5]	吴乃庚, 曾沁, 刘段灵, 等. 日极端气温的主客观预报能力评估及多模式集成网格释用[J]. 气象, 2017, 43(5):581-590.
[6]	尹姗, 马杰, 张恒德, 等. ECMWF模式的延伸期日最高气温预报偏差估计及订正分析[J]. 沙漠与绿洲气象, 2020, 14(6):77-84.
[7]	曹润东, 陈军明, 赵平. 地面观测资料地形高度订正对我国东部暴雨数值模拟的影响[J]. 气象与环境科学, 2021, 44(1):46-55.
[8]	罗菊英, 周建山, 闫永财. 基于数值预报及上级指导产品的本地气温MOS预报方法[J]. 气象科技, 2014, 42(3):443-450.
[9]	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.
[10]	刘还珠, 赵声蓉, 陆志善, 等. 国家气象中心客观要素预报——MOS系统[J]. 新疆气象, 2004, 27(3):4-7.
[11]	王敏, 李晓莉, 范广洲, 等. 区域集合预报系统2 m温度预报的校准技术[J]. 应用气象学报, 2012, 23(4):395-401.
[12]	马清, 龚建东, 李莉, 等. 超级集合预报的误差订正与集成研究[J]. 气象, 2008, 34(3):42-48.
[13]	马旭林, 时洋, 和杰, 等. 基于卡尔曼滤波递减平均算法的集合预报综合偏差订正[J]. 气象学报, 2015, 73(5):952-964.
[14]	王婧, 徐枝芳, 范广洲, 等. GRAPES_RAFS系统2 m温度偏差订正方法研究[J]. 气象, 2015, 41(6):719-726.
[15]	王丹, 黄少妮, 高红燕, 等. 递减平均法对陕西SCMOC精细化温度预报的订正效果[J]. 干旱气象, 2016, 34(3):575-583.
[16]	STENSRUD D J, YUSSOUF N. Bias-corrected short-range ensemble forecasts of near surface variables[J]. Meteorological Applications, 2010, 12(3):217-230. DOI URL
[17]	YUSSOUF N, STENSRUD D J. Bias-corrected short-range ensemble forecasts of near-surface variables during the 2005/06 cool season[J]. Weather & Forecasting, 2007, 22(6):1274-1286.
[18]	ZHU J S, KONG F Y, LEI H C. Bias-corrected short-range ensemble forecasts for near-surface variables during the summer season of 2010 in northern China[J]. Atmospheric and Oceanic Science Letters, 2014, 7(4):334-339. DOI URL
[19]	师春香, 潘旸, 谷军霞, 等. 多源气象数据融合格点实况产品研制进展[J]. 气象学报, 2019, 77(4):774-783.
[20]	刘军建, 师春香, 贾炳浩, 等. FY-2E地面太阳辐射反演及数据集评估[J]. 遥感信息, 2018, 33(1):104-110.
[21]	韩帅, 师春香, 姜志伟, 等. CMA高分辨率陆面数据同化系统(HRCLDAS-V1.0)研发及进展[J]. 气象科技进展, 2018, 8(1):102-108.
[22]	朱智, 师春香. 中国气象局陆面同化系统和全球陆面同化系统对中国区域土壤湿度的模拟与评估[J]. 科学技术与工程, 2014, 14(32):138-144.
[23]	崔园园, 覃军, 敬文琪, 等. GLDAS和CLDAS融合土壤水分产品在青藏高原地区的适用性评估[J]. 高原气象, 2018, 37(1):123-136.
[24]	韩帅, 师春香, 姜立鹏, 等. CLDAS土壤湿度模拟结果及评估[J]. 应用气象学报, 2017, 28(3):369-378.
[25]	孙帅, 师春香, 梁晓, 等. 不同陆面模式对我国地表温度模拟的适用性评估[J]. 应用气象学报, 2017, 28(3):369-378.
[26]	师春香, 张帅, 孙帅, 等. 改进的CLDAS降水驱动对中国区域积雪模拟的影响评估[J]. 气象, 2018, 44(8):985-997.