考虑数据均一性和自相关的中国极端气温变化趋势研究  

Study on Trends of Extreme Temperature in China Considering Data Homogenization and Autocorrelation

作　　者：胡宜昌[1] HU Yichang(China Meteorological Administration Training Centre,Beijing 100081)

机构地区：[1]中国气象局气象干部培训学院,北京100081

出　　处：《气象科技》2025年第2期211-221,共11页Meteorological Science and Technology

摘　　要：本文利用中国目前空间覆盖度最高的站点均一化逐日气温数据集,考虑时间序列自相关对长期趋势分析的影响,研究了1961—2021年中国极端气温的趋势变化特征。结果表明,中国区域平均而言,全年暖夜(冷夜)数、暖日(冷日)数的增加(减小)趋势分别为10.3(-7.8)、5.9(-3.6)d/10a,极冷夜、极暖夜、极冷日、极暖日的增温速度分别为0.52、0.30、0.30、0.21℃/10a,与未考虑自相关的趋势差值百分比均低于5%;对于单站而言,时间序列自相关对趋势大小的影响,大部分台站都在10%以内,但也有部分台站超过了50%。极端气温与平均气温的变化存在诸多不同,例如,虽然中国区域平均最低气温、最高气温的夏季升温趋势最弱,但是暖夜数、暖日数的夏季增加趋势却最强,冬季增加趋势反而最弱。空间覆盖度高的均一化资料揭示出中国极端气温变化更多的细节特征,例如,极暖日在长江、三角洲、珠江三角洲、京津冀、成渝等城市群所在的区域内增温趋势尤其显著,是否与城市化发展有关还有待进一步研究。China frequently experiences extreme temperature events,which often have severe impacts on social production and daily life.Therefore,it is of great importance to study the long-term trends of extreme temperature changes.The homogenisation of the observation dataset is crucial for detecting temperature change trends.In the meantime,whether to consider time series autocorrelation can also affect the detection results.Failure to consider the homogenisation of the temperature dataset or the autocorrelation of the temperature time series brings about uncertainty in research conclusions.In addition,the higher the spatial coverage of observation sites,the more advantageous it is to reveal spatial differences in change characteristics.This study analyses the trends of extreme temperature changes in China during the period of 1961-2021 using a homogenised daily station temperature dataset with the most spatial coverage currently,while taking into account the impacts of time series autocorrelation.For China as a whole,the annual warm nights(days),where daily minimum(maximum)temperature is above its 90th percentile,have an increasing trend of 10.3(5.9)d/10a,while the annual cold nights(days),where daily minimum(maximum)temperature is below its 10th percentile,have a decreasing trend of-7.8(-3.6)d/10a on space average,respectively.The warming rates of the annual coldest night(TNn),warming night(TNx),coldest day(TXn),and warmest day(TXx)are 0.52,0.30,0.30,and 0.21℃/10a on space average,respectively.For the regional average time series of extreme temperature in China,the percentage differences between the original trend and the decorrelation trend are all less than 5%.For a single station,the impact of time series autocorrelation on the magnitude of long-term linear trend is less than 10%for most stations,but there are also some stations with impacts exceeding 50%.There are great differences between extreme temperature changes and average temperature changes.For example,although the summer warming trend is the weakest in te

关 键 词：极端气温 线性趋势 均一化 自相关 

分 类 号：P467[天文地球—大气科学及气象学]