CLDAS气温实况融合产品在兰州和武威的检验评估及偏差订正

doi:10.11755/j.issn.1006-7639(2024)-01-0146

摘要/Abstract

摘要：

为更好地理解格点融合实况数据与观测数据的差异和代表性，利用甘肃兰州和武威两地站点的观测数据对中国气象局陆面数据同化系统（CMA Land Data Assimilation System，CLDAS）地面2 m气温融合产品进行检验评估及偏差订正。结果表明：（1）逐小时气温和日最低气温融合产品的平均误差总体为负值，较实际气温偏低，且在2 500 m以下误差随海拔上升而减小；日最高气温融合产品平均误差在海拔1 500 m附近为负值，1 500 m以上误差变为正值且随海拔升高而增大；日最高和最低气温误差较逐小时气温误差偏大，但平均误差均在2 ℃以内。（2）通过近网格点检验，发现逐小时CLDAS气温产品白天与实况相近，夜间较实况偏低0.2 ℃；日平均气温CLDAS融合产品总体较实况偏低1 ℃，兰州城区产品偏差相对较小；30 ℃以上高温天数融合产品与实况分布基本一致，但在兰州城区，CLDAS融合产品的高温天数较观测天数偏少。（3）线性回归法和递减平均法对CLDAS气温融合产品都有一定的订正效果，递减平均法订正效果更优且在高海拔地区订正效果更明显。CLDAS气温实况融合产品在兰州和武威两地能较好地反映气温变化特征，但日最高、最低气温误差较逐小时气温大，且在复杂地形下误差相对较大。

关键词: 2 m气温融合产品, 独立站点检验, 近网格点检验, 偏差订正

Abstract:

In order to make a further understand of the difference and representativeness between gridded merging real-time product and observed data, the paper evaluated and corrected the CLDAS 2 m temperature merging product based on the observational data of automatic stations in Lanzhou and Wuwei region. The results are as follows: (1) The hourly temperature and daily minimum temperature products are lower than observations, and these errors decrease with the altitude going up below 2 500 m. The mean error of the daily maximum temperature product is negative around the altitude of 1 500 m, and changes to positive values above 1 500 m, then the positive mean error increases with the increase of altitude. The errors of daily maximum and minimum temperature are larger than those of hourly temperature, but their mean errors are all within 2 ℃. (2) The near gridding validation further shows that the diurnal change of CLDAS hourly temperature is generally similar to observations in the daytime, while it is relatively 0.2 ℃ lower than observations at night. The daily average temperature of CLDAS merging product is generally lower about 1 ℃, and the negative deviation in Lanzhou urban area is relatively small. Spatial distribution of high temperature days above 30 ℃ of merging products is basically consistent with observations, but there are more actual high temperature days in Lanzhou urban area. (3) Both the linear regression and the decaying averaging method have a certain correction effect on CLDAS temperature merging products, and the latter has a better correction effect. The correction effect becomes better with the altitude increasing. To sum up, the CDLAS temperature merging products can better reflect the characteristics of actual temperature change in Lanzhou and Wuwei region, but its ability to reflect the daily maximum and minimum temperature is not as good as the hourly temperature, and the error is relatively large in complex terrain.

Key words: 2 m temperature merging product, independent sites validation, near-gridding sites validation, bias correction

中图分类号:

P456
P413

郭润霞, 刘新伟, 王一丞, 刘娜, 周子涵. CLDAS气温实况融合产品在兰州和武威的检验评估及偏差订正[J]. 干旱气象, 2024, 42(1): 146-155.

GUO Runxia, LIU Xinwei, WANG Yicheng, LIU Na, ZHOU Zihan. Verification and correction of 2 m temperature merging product of CLDAS in Lanzhou and Wuwei, Gansu Province[J]. Journal of Arid Meteorology, 2024, 42(1): 146-155.

图/表 13

图1 兰州（a）和武威（b）地区新建站分布

Fig.1 Distribution of newly-built sites in Lanzhou (a) and Wuwei (b) region

图2 兰州地区原有区域站点分布（红色点为用于近网格点检验的站点，虚线为0.05°×0.05°网格）

Fig.2 Distribution of original regional weather stations in Lanzhou region （The red dots represent the sites used for near grid verification，the dotted line represents 0.05°×0.05° grid）

图3 2021年3月至2022年4月逐小时气温误差的日变化（a）和月变化（b）

Fig.3 Diurnal (a) and monthly (b) variations of hourly temperature errors from March 2021 to April 2022

图4 2021年3月至2022年4月日最高（a）、最低气温（b）误差的逐月变化

Fig.4 Monthly variation of errors of daily maximum（a） and minimum（b） temperature from March 2021 to April 2022

图5 兰州（a、b、c）和武威（d、e、f）逐小时气温的平均误差（a、d）、均方根误差（b、e）及平均绝对误差（c、f）空间分布

Fig.5 Spatial distributions of ME (a, d), RMSE (b, e) and MAE (c, f) of hourly temperature in Lanzhou (a, b, c) and Wuwei (d, e, f) region

图6 不同地区日最高气温（a、b、c）与日最低气温（d、e、f）误差对比

Fig.6 Error comparison of daily maximum temperature（a，b，c） and daily minimum temperature（d，e，f） in different regions

图7 逐小时气温误差（a）和日最高气温误差（b）与海拔的关系

Fig.7 Relationship between the errors of hourly temperature（a），daily maximum temperature（b） and altitude

图8 兰州地区日最高气温（a）和日最低气温（b）融合产品准确率的空间分布

Fig.8 Spatial distribution of the accuracy of daily maximum temperature (a) and daily minimum temperature (b) in Lanzhou region

图9 日平均气温平均误差空间分布

Fig.9 Spatial distribution of ME of the daily average temperature

图10 观测数据（a）和CLDAS（b）日平均气温30 ℃以上日数空间分布

Fig.10 The spatial distribution of days with daily average temperature above 30 ℃ of observation（a） and CLDAS（b）

图11 不同地区日平均气温偏差对比

Fig.11 Comparison of daily average temperature’s ME in different regions

图12 逐小时气温产品订正前后误差随时间变化

Fig.12 The temporal change of errors of hourly temperature product before and after correction

图13 武威（a、b）和兰州（c、d）地区逐小时气温误差线性回归算法（a、c）和递减平均算法（b、d）订正改善率

Fig.13 Improvement rate after correcting with the linear regression（a，c） and the decaying averaging methods（b，d） in Wuwei（a，b） and Lanzhou（c，d） region

参考文献 26

[1]	陈玉玮, 黄晓莹, 2021. 广东区域智能网格实况分析产品的检验评估[J]. 广东气象, 43(1): 42-44.
[2]	何泽能, 张德军, 叶勤玉, 等, 2022. 近40 a重庆城市热岛特征及其与天气状况的关系[J]. 干旱气象, 40(4): 683-689. DOI
[3]	蒋雨荷, 黄晓龙, 杜冰, 2022. 两种高分辨率气温实况融合产品在四川省的质量评估[J]. 高原山地气象研究, 42(1): 109-116.
[4]	龙柯吉, 师春香, 韩帅, 等, 2019. 中国区域高分辨率温度实况融合格点分析产品质量评估[J]. 高原山地气象研究, 39(3): 67-74.
[5]	马清, 2008. 中尺度集合预报的偏差订正与多模式集成研究[D]. 南京: 南京信息工程大学.
[6]	马旭林, 时洋, 和杰, 等, 2015. 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, 73(5): 952-964.
[7]	齐铎, 刘松涛, 张天华, 等, 2020. 基于格点的中国东北中北部2 m温度数值预报检验及偏差订正[J]. 干旱气象, 38(1): 81-88.
[8]	尚建设, 李本亮, 孙小丽, 等, 2018. 济南市夏季城市热岛效应特征分析[J]. 干旱气象, 36(1): 70-74. DOI
[9]	师春香, 潘旸, 谷军霞, 等, 2019. 多源气象数据融合格点实况产品研制进展[J]. 气象学报, 77(4): 774-783.
[10]	师春香. 2018. 全国智能网格实况融合分析产品进展[C]// 中国气象学会. 第35届中国气象学会年会论文集: S20 深度信息化:应用支持与智能发展. 北京: 气象出版社.
[11]	宋超辉, 孙安健, 1995. 非均一性气温气候序列订正方法的研究[J]. 高原气象, 14(2): 88-93.
[12]	孙靖, 程光光, 黄小玉, 2021. 中国地面气象要素格点融合业务产品检验[J]. 高原气象, 40(1): 178-188. DOI
[13]	屠其璞, 1979. 平均气温序列延长订正的讨论[J]. 南京气象学院学报, (2): 193-200.
[14]	王丹, 戴昌明, 娄盼星, 等, 2021. 陕西ECMWF、GRAPES_Meso和SCMOC气温预报的对比检验及订正[J]. 干旱气象, 39(4): 697-708.
[15]	王丹, 黄少妮, 高红燕, 等, 2016. 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, 34(3): 575-583. DOI
[16]	王丹, 王建鹏, 白庆梅, 等, 2019. 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, 45(9): 1 310-1 321.
[17]	王利亚, 2021. 内蒙古自治区智能网格实况产品基于三种地形的评估研究[J]. 电子技术与软件工程, (1): 192-193.
[18]	韦青, 代刊, 林建, 等, 2020. 2016—2018年全国智能网格降水及温度预报检验评估[J]. 气象, 46(10): 1 272-1 285.
[19]	姚帅, 刘柏鑫, 慕秀香, 等, 2022. 基于CMA-Meso的吉林省2 m温度预报检验及订正[J]. 气象灾害防御, 29(3): 34-38.
[20]	俞剑蔚, 李聪, 蔡凝昊, 等, 2019. 国家级格点实况分析产品在江苏地区的适用性评估分析[J]. 气象, 45(9): 1 288-1 298.
[21]	ALBERGEL C, DORIGO W, BALSAMO G, et al, 2013. Monitoring multi-decadal satellite earth observation of soil moisture products through land surface reanalyses[J]. Remote Sensing of Environment, 138: 77-89.
[22]	CUI B, TOTH Z, ZHU Y J, et al, 2012. Bias correction for global ensemble forecast[J]. Weather and Forecasting, 27(2): 396-410.
[23]	DU J, ZHOU B B, 2011. A dynamical performance-ranking method for predicting individual ensemble member performance and its application to ensemble averaging[J]. Monthly Weather Review, 139(10): 3 284-3 303.
[24]	RODELL M, HOUSER P R, JAMBOR U, et al, 2004. The global land data assimilation system[J]. Bulletin of the American Meteorological Society, 85(3): 381-394.
[25]	WANG J Z, CHEN J, DU J, et al, 2018. Sensitivity of ensemble forecast verification to model bias[J]. Monthly Weather Review, 146(3): 781-796.
[26]	XIA Y L, MITCHELL K, EK M, et al, 2012. Continental-scale water and energy flux analysis and validation for the North American Land Data Assimilation System project phase 2 (NLDAS-2): 2. Validation of model-simulated streamflow[J]. Journal of Geophysical Research: Atmospheres, 117(D3): D03110, DOI: 10.1029/2011JD016051