Diurnal variation characteristics of warm season precipitation in Xinjiang based on K-means clustering method

doi:10.11755/j.issn.1006-7639-2025-01-0104

Abstract

Abstract:

An in-depth understanding of the daily variation characteristics of precipitation at classified stations is essential for optimizing and improving accurate forecasting methods. Based on the hourly precipitation observation data from 105 national meteorological stations in Xinjiang during the warm season (May to September) from 2010 to 2019, the stations were classified using the K-means clustering method, and the precipitation characteristics of each category of stations were analyzed according to hourly average precipitation amount, precipitation frequency, and precipitation intensity. The results show that the stations in Xinjiang can be classified into four categories: southern Xinjiang and desert areas (Class I), Tianshan Mountains (Class II), northern Xinjiang and the southern slope of the West Tianshan Mountains (Class III), and valley areas (Class IV). The clustering result is similar to that of classifications based on geographic location and topographic height but it is more detailed and scientific. The distribution of cumulative precipitation and precipitation hours of four types of stations is relatively concentrated, with annual average cumulative precipitation of 54, 354, 110, and 217 mm, corresponding to 67, 311, 118, and 213 hours, respectively. The diurnal variation of precipitation frequency in the warm season at most stations in Xinjiang follows a ‘single-peak’ pattern, but the times of the peaks and valleys vary depending on altitude differences. Precipitation and precipitation intensity generally follow a ‘multi-peak’ structure, and hourly precipitation intensity greater than 1 mm mainly concentrated at Class II stations, with the peaks occurring at 16:00—17:00. Precipitation and precipitation frequency are highest in June and lowest in September. The monthly distribution and the month-to-month diurnal variation characteristics of precipitation intensity differ significantly, but the peak intensity does not show a notable difference. In 2016, all three precipitation indicators characteristics during the warm season were significantly higher than those of other years. In 2010, Class I stations had the highest precipitation intensity, and the precipitation amount and frequency reached a secondary peak, while the changes at the other categories of stations were relatively small.

Key words: K-means clustering algorithm, site classification, warm season precipitation, diurnal variability

摘要： 深入理解分类站点的降水日变化特征，对优化和提升精准预报方法至关重要。基于2010—2019年新疆105个国家站暖季（5—9月）逐时降水观测数据，采用K-means聚类方法对站点进行分类，根据逐时平均降水量、降水频率和降水强度3个指标分析各类站点的降水特征。结果表明：新疆站点可分为南疆及荒漠区（I类）、天山山区（II类）、北疆及西天山南坡区（III类）和山谷区（IV类），聚类结果与按地理位置和地形高度划分的结果相近，但更具精细性和科学性。各类站点的累计降水量和降水时数分布相对集中，年均累计降水量分别为54、354、110和217 mm，对应的降水时数分别为67、311、118和213 h。新疆大部站点暖季降水频率的日变化呈“单峰”型，但峰值和谷值出现的时间因海拔差异有所不同；降水量和降水强度多呈“多峰”结构，其中小时降水强度大于1 mm的主要集中在II类站点，峰值出现在16：00—17：00。各类站点的降水量和降水频率均在6月最高、9月最少；降水强度的月分布和逐月日变化特征差异显著，但峰值强度无明显差异。2016年暖季的3个降水特征指标均明显高于其他年份；2010年I类站点的降水强度最大，降水量和降水频率达到次高峰，而其他类别的站点变化相对较小。

关键词: K-means聚类算法, 站点分类, 暖季降水, 日变化

CLC Number:

P426.61+4

LI Anbei, ZHANG Meng, LI Ruqi, MU Huan, WAN Yu. Diurnal variation characteristics of warm season precipitation in Xinjiang based on K-means clustering method[J]. Journal of Arid Meteorology, 2025, 43(1): 104-113.

李桉孛, 张萌, 李如琦, 牟欢, 万瑜. 基于K-means聚类的新疆暖季降水日变化特征[J]. 干旱气象, 2025, 43(1): 104-113.

Figures/Tables 9

Tab.1 Test and evaluation of clustering results of different initial protocols

聚类评价指标	样本初始类数k				指标说明
聚类评价指标	k=3	k=4	k=5	k=6	指标说明
轮廓系数	0.403	0.371	0.325	0.314	接近1的值表示分离良好的簇
方差比准则	160.191	171.244	160.884	140.499	数值越大表示簇之间内聚性越高
Davies-Bouldin指数	0.818	0.815	0.929	1.068	数值越低表示簇越紧凑且分离良好

Fig.1 Distribution of national meteorological stations in Xinjiang and the classification of stations after clustering

Tab.2 Statistical results of different station classifications in Xinjiang from 2010 to 2019

类别	基本信息		5—9月
类别	站点数/个	平均海拔高度/m	降水量/mm	降水量平均/mm	降水时数/h	降水时数平均/h
Ⅰ类（南疆及荒漠区）	47	987	6~82	54	16~116	67
Ⅱ类（天山山区）	6	2 057	298~400	354	274~357	311
Ⅲ类（北疆及西天山南坡区）	40	824	84~141	110	85~148	118
Ⅳ类（山谷区）	12	1 672	165~278	217	164~322	216

Fig.2 Box plots of the average precipitation （black box，Unit： mm） and precipitation hours （red box，Unit： h） in the warm season for four classifications of stations in Xinjiang from 2010 to 2019

Fig.3 Hour-by-hour distribution of mean precipitation amount （a），precipitation frequency （b），and precipitation intensity （c） in Xinjiang in the warm season from 2010 to 2019

Fig.4 The hourly distributions of average precipitation amount （the left），precipitation frequency （the middle），and precipitation intensity （the right） for four types of stations in Xinjiang in warm season from 2010 to 2019

Fig.5 The hourly distributions of monthly average precipitation amount （the top，Unit： mm），precipitation frequency （the middle，Unit： %），and precipitation intensity （the bottom，Unit： mm·h-1） for average of the entire Xinjiang and four types of stations in worm season from 2010 to 2019 （The top line represents the monthly cumulative amount）

Fig.6 The hourly distribution of annual average precipitation amount （the top，Unit： mm），precipitation frequency （the middle，Unit： %），and precipitation intensity （the bottom，Unit： mm·h-1） for the regional average of the entire Xinjiang and four types of stations from 2010 to 2019 （The top line represents the year-by-year accumulation）

Fig.7 The average 500 hPa wind field (contours) and its anomaly (the color shaded) at 11:00 from May to September during 2010—2019, and average 500 hPa wind speed (wind vectors) in 2016 (a), and 850 hPa average wind speed anomaly (contours) at 11:00 from May to September during 2010—2019, and average wind speed (arrow vectors) in 2016 (b) (Unit: m·s-1)

References 35

[1]	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.
[2]	陈春艳, 赵克明, 阿不力米提江·阿布力克木, 等, 2015. 暖湿背景下新疆逐时降水变化特征研究[J]. 干旱区地理, 38(4): 692-702.
[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): 1 463-1 470.
[4]	段赛男, 焦瑞莉, 吴成来, 2024. 一种基于K-means聚类算法的沙尘天气客观识别方法[J]. 气候与环境研究, 29(2): 178-192.
[5]	高旭旭, 于长文, 朱卫浩, 2018. 河北省不同区域降水日变化特征差异[J]. 河北农业科学, 22(4): 63-67.
[6]	苟浩锋, 王遂缠, 杨银, 等, 2020. 基于聚类分析的兰州地区自动站降水特征分析[J]. 沙漠与绿洲气象, 14(1): 108-114.
[7]	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
[8]	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.
[9]	黄艳, 杨莲梅, 唐鹏, 等, 2024. 新疆伊犁河谷暖季短历时强降水变化特征[J]. 沙漠与绿洲气象, 18(6): 32-37.
[10]	李晓婷, 刘利民, 李雪洋, 等, 2021. 基于逐小时数据的辽宁省夏季降水日变化特征分析[J]. 沈阳农业大学学报, 52(2): 129-136.
[11]	林春泽, 刘琳, 林文才, 等, 2016. 湖北省夏季降水日变化特征[J]. 大气科学学报, 39(4): 490-500.
[12]	刘伟东, 尤焕苓, 任国玉, 等, 2014. 北京地区自动站降水特征的聚类分析[J]. 气象, 40(7): 844-851.
[13]	秦贺, 陈春艳, 阿不力米提江·阿布力克木, 等, 2019. 新疆暖季短时强降水特征[J]. 干旱区研究, 36(6): 1 440-1 449.
[14]	孙倩倩, 刘晶淼, 2014. 基于聚类分析的中国东北地区气温和降水时空变化特征[J]. 气象与环境学报, 30(3): 59-65.
[15]	唐红玉, 顾建峰, 俞胜宾, 等, 2011. 西南地区降水日变化特征分析[J]. 高原气象, 30(2): 376-384.
[16]	王丽伟, 黄晓龙, 冯晓莉, 等, 2019. 2006—2015年吉林省夏季降水日变化特征[J]. 中国农学通报, 35(3): 122-130. DOI
[17]	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
[18]	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
[19]	谢泽明, 周玉淑, 杨莲梅, 2018. 新疆降水研究进展综述[J]. 暴雨灾害, 37(3): 204-212.
[20]	杨森, 周晓珊, 高杰, 2011. 辽宁省夏季降水的日变化特征[J]. 气象, 37(8): 943-949.
[21]	杨涛, 杨莲梅, 2003. 新疆强对流暴雨的气候特征和概率分布模式研究[J]. 灾害学, 18(1): 49-54.
[22]	杨涛, 杨莲梅, 张云惠, 等, 2021. 新疆短时强降水天气系统环流配置及雷达回波特征[J]. 干旱气象, 39(4): 631-640.
[23]	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.
[24]	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
[25]	张家宝, 邓子风, 1987. 新疆降水概论[M]. 北京: 气象出版社.
[26]	张俊兰, 杨霞, 肖俊安, 等, 2023. 昆仑山北部夏季降水多尺度时空变化特征[J]. 高原山地气象研究, 43(3): 1-10.
[27]	赵勇, 黄丹青, 古丽格娜, 等, 2010. 新疆北部夏季强降水分析[J]. 干旱区研究, 27(5): 773-779.
[28]	赵玉春, 王叶红, 2020. 我国东南沿岸及复杂山地后汛期降水日变化的数值研究[J]. 大气科学, 44(2): 371-389.
[29]	中国气象局, 国家气候委员会, 2017. 2016年中国气候公报[R]. 北京: 中国气象局.
[30]	朱小凡, 张明军, 王圣杰, 等, 2016. 2008—2013年新疆夏季降水的日变化特征[J]. 生态学杂志, 35(2): 478-488.
[31]	ANDERSSON T, 1969. The diurnal variation of precipitation in Sweden[J]. Geografiska Annaler: Series A, Physical Geography, 51(3): 176-184.
[32]	BOWMAN K P, COLLIER J C, NORTH G R, et al, 2005. Diurnal cycle of tropical precipitation in Tropical Rainfall Measuring Mission (TRMM) satellite and ocean buoy rain gauge data[J]. Journal of Geophysical Research: Atmospheres, 110(D21): D21104. DOI:10.1029/2005JD005763.
[33]	YU R C, ZHOU T J, XIONG A Y, et al, 2007a. Diurnal variations of summer precipitation over contiguous China[J]. Geophysical Research Letters, 34(1): L01704. DOI:10.1029/2006GL028129.
[34]	YU R C, XU Y P, ZHOU T J, et al, 2007b. Relation between rainfall duration and diurnal variation in the warm season precipitation over central Eastern China[J]. Geophysical Research Letters, 34(13): L13703. DOI:10.1029/2007GL030315.
[35]	ZHOU T J, YU R C, CHEN H M, et al, 2008. Summer precipitation frequency, intensity, and diurnal cycle over China: A comparison of satellite data with rain gauge observations[J]. Journal of Climate, 21(16): 3 997-4 010.