太行山及以东复杂地形下多模式降水预报日变化偏差特征研究

doi:10.11755/j.issn.1006-7639-2025-02-0308

摘要/Abstract

摘要： 评估数值模式在降水日变化方面的偏差特征，有助于深入了解模式的预报能力。本文利用太行山脉、伏牛山及以东复杂地形区域的夏季降水观测资料，结合模式12~36 h预报时效的逐3 h降水量数据，计算了降水频次、降水强度及其峰值、振幅等关键参数。采用偏差分析和空间相关性分析等方法，将欧洲中期天气预报中心（European Centre for Medium-Range Weather Forecasts，ECMWF）数值预报模式、美国国家环境预报中心全球预报系统（National Centers for Environmental Prediction-Global Forecast System，NCEP-GFS）、中国气象局区域台风数值预报系统（CMA-TYM）、中国气象局上海数值预报模式系统（CMA-SH9）等业务常用模式的预报特征与实际观测结果进行对比，并对产生偏差的原因开展了初步分析。结果表明，区域模式CMA-TYM和CMA-SH9对降水的预报具有“低频次、高强度”的特征，全球模式ECMWF和NCEP-GFS则为“高频次、低强度”，且各模式预报结果与观测的空间相关性均较弱。区域模式中，CMA-TYM降水量预报偏差较小，在0附近波动，降水频次的峰值时间和日变化振幅预报效果较好；CMA-SH9对大部分时次的降水量预报正偏差最大，预报的降水强度峰值时间和日变化振幅效果最好，但预报降水频次偏大且其峰值时间基本为前半夜，而观测的华北平原、山东省中东部及黄淮平原北部的峰值时间主要出现在清晨，相差近12 h。全球模式中，NCEP-GFS的降水量预报大部分时次小于观测且夜间到清晨和下午的负偏差绝对值最大，预报的降水频次大部分时次为正偏差最大，降水强度为负偏差绝对值最大；ECMWF低估傍晚到夜间的降水量，高估清晨到中午的降水量，降水频次各时次均为正偏差，降水强度均为负偏差。综合分析有（无）降水时的气温场及风场，模式降水量迅速增大主要是CMA-SH9对晚上和ECMWF对中午预报的气温偏高配合东南暖湿气流，容易在近地层形成不稳定层结，从而触发对流天气，产生较大的降水量尤其是较大的降水频次。

关键词: 降水日变化, 业务数值模式, 偏差特征

Abstract:

Evaluating the bias characteristics of numerical models in the diurnal variation of precipitation is helpful to understand the forecasting ability of the model. The key parameters such as precipitation frequency, precipitation intensity, its peak value and amplitude are calculated by using the 3 h observation data of summer precipitation, combined with the 3 h precipitation of the 12-36 h forecast time of the model in the complex terrain of the Taihang Mountains, Funiu Mountains and the areas to their east. Using the methods of deviation analysis and spatial correlation analysis, the forecast characteristics of the ECMWF (European Centre for Medium-Range Weather Forecasts) numerical forecast model, NCEP-GFS (National Centers for Environmental Prediction-Global Forecast System), CMA-TYM (the China Meteorological Administration Regional Typhoon Numerical Forecast System) and CMA-SH9 (the China Meteorological Administration Shanghai Numerical Forecast Model System) are compared with the observations, and the causes of the deviations are analyzed preliminarily. The results show that the regional models CMA-TYM and CMA-SH9 have the characteristics of “low frequency and high intensity” in precipitation forecast, while the global models ECMWF and NCEP-GFS feature “high frequency and low intensity”. Moreover, the spatial correlation between each model and the observational data is weak. In the regional models, the bias of precipitation forecast about CMA-TYM is relatively small and it fluctuates near 0, and the peak time and diurnal variation amplitude of precipitation frequency are better. The CMA-SH9 has the largest positive deviation of precipitation most of the time, and the predicted peak time and diurnal variation amplitude of precipitation intensity are the best, but the frequency of precipitation is larger and its peak time is basically in the early night, while the peak time of observations in the North China Plain, central and eastern Shandong Province and northern Huanghuai Plain mainly occurs in the early morning with a difference of nearly 12 h. In the global model, the precipitation forecast of NCEP-GFS is smaller than observation in most of the time, and the absolute value of negative deviation from night to early morning and afternoon is the largest. The precipitation frequency of NCEP-GFS has the largest positive deviation most of the time, while the absolute value of the negative deviation of precipitation intensity is the largest. The ECMWF underestimates the precipitation from evening to night and overestimates the precipitation from early morning to noon, where the precipitation frequency is positive deviation, and the precipitation intensity of each time is negative deviation. According to the comprehensive analysis of the temperature and wind field with or without precipitation, the rapid increase of precipitation in the model is mainly due to the high temperature forecasted by CMA-SH9 at night and ECMWF at noon, combined with the southeast warm and humid airflow, which is easy to form unstable stratification in the near surface layer, thus triggering convective weather and producing large precipitation, especially large precipitation frequency.

Key words: daily variation of precipitation, operational numerical prediction, deviation characteristics

中图分类号:

P467.6

曲巧娜, 吴炜. 太行山及以东复杂地形下多模式降水预报日变化偏差特征研究[J]. 干旱气象, 2025, 43(2): 308-320.

QU Qiaona, WU Wei. Diurnal variation bias characteristics of precipitation forecast about various models under complex terrain in Taihang mountain and its east area[J]. Journal of Arid Meteorology, 2025, 43(2): 308-320.

图/表 9

图1 研究区域地形（填色，单位：m）及气象站点空间分布注:基于国家测绘地理信息局标准地图服务网站下载的审图号为GS(2016)1552号的标准地图制作，底图边界无修改，下同。

Fig.1 The spatial distribution of topography （the gray shaded， Unit： m） and meteorological stations in the study area

图2 2020—2023年6—8月不同预报模式与观测的降水量（a）、降水频次（c）、降水强度（d）的偏差及模式预报与观测的降水量空间相关系数（b）日内逐3 h变化

Fig.2 The 3 hours variation in a day of the deviation of precipitation (a), precipitation frequency (c), precipitation intensity (d) between different forecast models and observations, and the spatial correlation coefficient of precipitation between model forecast and observation (b) from June to August from 2020 to 2023

图3 2020—2023年6—8月观测（a）与ECMWF（b）、NCEP-GFS（c）、CMA-TYM（d）、CMA-SH9（e）模式预报的20：00降水量第95百分位数空间分布（单位：mm）

Fig.3 The spatial distribution of the 95th percentile of precipitation at 20:00 of the observation (a) and forecasted by ECMWF (b), NCEP-GFS (c), CMA-TYM (d), CMA-SH9 (e) model from June to August during 2020-2023 (Unit: mm)

图4 2020—2023年6—8月观测（a）与ECMWF（b）、NCEP-GFS（c）、CMA-TYM（d）、CMA-SH9（e）模式预报的降水频次峰值时间分布

Fig.4 The distribution of occurrence time of precipitation frequency peaks of the observation (a) and forecasted by ECMWF (b), NCEP-GFS (c), CMA-TYM (d), CMA-SH9 (e) model from June to August during 2020-2023

图5 2020—2023年6—8月观测（a）与ECMWF（b）、NCEP-GFS（c）、CMA-TYM（d）、CMA-SH9（e）模式预报的降水频次日变化振幅分布

Fig.5 The distribution of diurnal variation amplitude of precipitation frequency of the observation (a) and forecasted by ECMWF (b), NCEP-GFS (c), CMA-TYM (d), CMA-SH9 (e) model from June to August during 2020-2023

图6 2020—2023年6—8月有降水（a、b）和无降水（c、d）时ECMWF模式在14：00（a、c）和20：00（b、d）的气温预报偏差（单位：℃）分布

Fig.6 The distribution of the forecast deviation of air temperatures (Unit: ℃) by the ECMWF model at 14:00 (a, c) and 20:00 (b, d) from June to August during 2020-2023 under the conditions of precipitation (a, b) and no-precipitation (c, d)

图7 2020—2023年6—8月有降水（a、b）和无降水（c、d）时ECMWF模式在14：00（a、c）和20：00（b、d）的风场预报偏差（单位：m·s-1）分布

Fig.7 The distribution of the forecast deviation of wind field (Unit: m·s-1) by the ECMWF model at 14:00 (a, c) and 20:00 (b, d) from June to August during 2020-2023 under the conditions of precipitation (a, b) and no-precipitation (c, d)

图8 2020—2023年6—8月有降水（a、b）和无降水（c、d）时CMA-SH9模式在05：00（a、c）和20：00（b、d）的气温预报偏差（单位：℃）分布

Fig.8 The distribution of the forecast deviation of air temperature (Unit: ℃) by the CMA-SH9 model at 05:00 (a, c) and 20:00 (b, d) from June to August during 2020-2023 under the conditions of precipitation (a, b) and no-precipitation (c, d)

图9 2020—2023年6—8月有降水（a、b）和无降水（c、d）时CMA-SH9模式在05：00（a、c）和20：00（b、d）的风场预报偏差（单位：m·s-1）分布

Fig.9 The distribution of the forecast deviation of wind field (Unit: m·s-1) by the CMA-SH9 model at 05:00 (a, c) and 20:00 (b, d) from June to August during 2020-2023 under the conditions of precipitation (a, b) and no-precipitation (c, d)

参考文献 36

[1]	艾克代·沙拉木, 何清, 阿力木·阿巴斯, 等, 2022. 帕米尔高原东部地区4—9月降水日变化特征研究[J]. 沙漠与绿洲气象, 16(6):42-50.
[2]	陈春艳, 王建捷, 唐冶, 等, 2017. 新疆夏季降水日变化特征[J]. 应用气象学报, 28(1):72-85.
[3]	戴泽军, 宇如聪, 陈昊明, 2009. 湖南夏季降水日变化特征[J]. 高原气象, 28(6):1463-1 470.
[4]	董新宁, 吴遥, 黄安宁, 等, 2023. 川渝地区夏季极端降水日变化特征分析[J]. 大气科学, 47(2):259-272.
[5]	段婧, 王新, 陈勇, 等, 2024. 祁连山中段夏季降雨与对流特征分析[J]. 干旱气象, 42(3):405-414. DOI
[6]	郭军, 熊明明, 黄鹤, 2019. 京津冀地区暖季降水日变化特征分析[J]. 海洋气象学报, 39(2):58-67.
[7]	郭玉琳, 赵勇, 张同文, 等, 2023. 新疆天山山区夏季极端降水的日变化特征[J]. 沙漠与绿洲气象, 17(2):44-51.
[8]	郭玉琳, 赵勇, 周雅蔓, 等, 2022. 新疆天山山区夏季降水日变化特征及其与海拔高度关系[J]. 干旱区地理, 45(1):57-65. DOI
[9]	韩函, 吴昊旻, 黄安宁, 2017. 华北地区夏季降水日变化的时空分布特征[J]. 大气科学, 41(2):263-274.
[10]	黄安宁, 张耀存, 朱坚, 2008. 物理过程参数化方案对中国夏季降水日变化模拟的影响[J]. 地球科学进展, 23(11):1174-1 184.
[11]	刘煦, 凌艺, 张凯静, 2020. 山东半岛地区夏半年降水日变化特征[J]. 干旱气象, 38(2):234-241.
[12]	刘亚楠, 王东海, 李国平, 等, 2019. 南海夏季风爆发前后华南前汛期降水日变化对比分析[J]. 热带气象学报, 35(3):365-378.
[13]	卢冰, 孙继松, 仲跻芹, 等, 2017. 区域数值预报系统在北京地区的降水日变化预报偏差特征及成因分析[J]. 气象学报, 75(2):248-259.
[14]	全国气象防灾减灾标准化技术委员会, 2012. 降水量等级:GB/T 28592—2012[S]. 北京: 中国标准出版社.
[15]	唐菁, 陈生, 李峙, 等, 2023. 基于雷达观测的华南夏季降水类型日变化的时空分布特征[J]. 气象, 49(3):279-290.
[16]	唐永兰, 徐桂荣, 王晓芳, 等, 2023. 1992—2021年三峡库区夏季小时强降水时空分布特征[J]. 干旱气象, 41(4):589-598. DOI
[17]	王新伟, 杜明哲, 肖瑶, 2018. 河南省汛期降水的日变化特征[J]. 干旱气象, 36(3):383-389. DOI
[18]	王颖, 刘丹妮, 张玮玮, 等, 2019. 2004—2016年浙江省夏季降水的日变化特征[J]. 干旱气象, 37(1):1-9. DOI
[19]	魏凤英, 2007. 现代气候统计诊断与预测技术[M]. 北京: 气象出版社.
[20]	肖安, 尹小飞, 刘献耀, 2022. 江西省降水日变化时空分布特征[J]. 干旱气象, 40(5):840-848. DOI
[21]	徐瑷, 喻琴昆, 刘成刚, 等, 2024. 川西高原甘孜州汛期短时强降水的时空分布特征[J]. 高原山地气象研究, 44(2):50-57.
[22]	徐明, 王晓芳, 高琦, 等, 2016. 基于TRMM卫星资料揭示的长江流域梅雨季节降水日变化[J]. 长江流域资源与环境, 25(12):1934-1 944.
[23]	闫荞荞, 王必强, 刘园园, 等, 2025. 一种自动气象站小时降水数据非实时质量控制方法及应用[J]. 暴雨灾害, 44(1):102-111.
[24]	杨霞, 张俊兰, 华烨, 等, 2021. 新疆伊犁河谷不同季节降水的日变化特征[J]. 干旱气象, 39(3):394-405.
[25]	叶茂, 吴钲, 游婷, 等, 2022. CMA-SH9在川渝地区的降水日变化预报效果评估[J]. 高原山地气象研究, 42(2):56-62.
[26]	宇如聪, 李建, 2016. 中国大陆日降水峰值时间位相的区域特征分析[J]. 气象学报, 74(1):18-30.
[27]	宇如聪, 李建, 陈昊明, 等, 2014. 中国大陆降水日变化研究进展[J]. 气象学报, 72(5):948-968.
[28]	张宏芳, 潘留杰, 陈昊明, 等, 2020. 秦岭及周边地区暖季降水日变化及其成因分析[J]. 高原气象, 39(5):935-946. DOI
[29]	张武龙, 杨康权, 康岚, 等, 2024. 川西高原和攀西地区不同级别短时强降水环境参量特征对比分析[J]. 高原山地气象研究, 44(2): 75-82.
[30]	钟琦, 孙卓, 陈昊明, 等, 2022. 京津冀强降水日变化的多模式预报偏差及成因分析[J]. 中国科学:地球科学, 52(9):1831-1 848.
[31]	BECHTOLD P, SEMANE N, LOPEZ P, et al, 2014. Representing equilibrium and nonequilibrium convection in large-scale models[J]. Journal of the Atmospheric Sciences, 71(2), 734-753.
[32]	CUI Z Q, PU Z X, 2023. The use of regional data assimilation to improve numerical simulations of diurnal characteristics of precipitation during an active madeen-Julian oscillation event over the maritime continent[J]. Remote Sensing, 15(9): 2405. DOI: 10.3390/rs15092405.
[33]	KOO M S, HONG S Y, 2010. Diurnal variations of simulated precipitation over East Asia in two regional climate models[J]. Journal of Geophysical Research: Atmospheres, 1155(D5): D05105. DOI: 1029/2009JD012574.
[34]	LIU X M, ZHANG M J, WANG S J, et al, 2017. Assessment of diurnal variation of summer precipitation over the Qilian Mountains based on an hourly merged dataset from 2008 to 2014[J]. Journal of Geographical Sciences, 27(3): 326-336. DOI
[35]	YU R C, ZHOU T J, XIONG A Y, et al, 2007. Diurnal variations of summer precipitation over contiguous China[J]. Geophysical Research Letters, 34(1): L01704. DOI: 10.1029/2006GL028129.
[36]	YUAN W H, YU R C, ZHANG M H, et al, 2013. Diurnal cycle of summer precipitation over subtropical east Asia in CAM5[J]. Journal of Climate, 26(10): 3 159-3 172.