本文利用1979-2019年ERA5再分析资料和ERSSTv5海表温度数据集，选择绝对阈值（35℃、30℃）和相对阈值（90百分位数和95百分位数）两类定义下的高温，通过线性趋势法、经验正交函数分解等方法分析了北半球夏季近41年来高温事件（包括高温日数、高温强度、高温影响范围的变化）的气候特征及其长期变化，并进一步探讨了高温日数变化与气候变暖、海表温度等气候因子的可能联系。主要结果如下：

(1) 统计了日最高气温超过35℃（以及30℃）的高温事件(后文分别简称35℃高温和30℃高温)，结果显示北半球夏季35℃高温主要分布在10°N~48°N之间，18°N~28.5°N是高温最多的纬度带，年平均高温日数超过16天。在70.5°N以北整个资料时段没有出现35℃高温。区域上看，非洲北部和中东地区是高温最多的地区，其年平均高温日数超过65天。在过去41年间，35℃高温日数以2.0days/10a的速率在增加，非洲北部、中东地区、美国西南部是高温日数增长最快的区域，增长幅度超过4days/10a。此外，35℃高温日数有2年左右的年际变化周期。30℃高温基本特征显示，北半球30℃高温主要分布在55°N以南，16.5°~33°N是高温最多的纬度带，在76.5°N以北没有出现。在过去41年间，30℃高温日数以1.9days/10a的速率在增加，功率谱分析显示有4年左右的年际尺度周期。

(2) 逐个格点统计1979-2019年时段日最高气温的90和95百分位数，进而逐年计算了超过该阈值的高温事件（后文分别简称90百分位高温和95百分位高温）。气候态看，90和95百分位数对应的气温阈值从低纬度到高纬度呈现逐渐下降的趋势，非洲北部、中东地区、印度是北半球高温相对阈值最高的区域，对应的气温在40℃以上，而在高纬度地区，如格陵兰岛是相对阈值最低的地方，对应的气温在0℃以下。过去41年间，90百分位高温日数以3.0days/10a的速率在增加，95百分位高温日数以1.9days/10a的速率在增加。其趋势有明显的区域差异，其中亚洲西部、欧洲东南部、非洲低纬度地区、北美洲东北部都是高温日数增长最快的区域，90百分位和95百分位高温日数的增长幅度均超过2days/10a。

(3) 对高温强度和覆盖范围的统计显示，1979-2019年夏季北半球高温强度逐渐增强，覆盖范围扩大。分析北半球格点每年高温日数总数和高温日日最高气温总和等指标，发现高温强度逐渐增大，其中2010为高温强度最强的年份，35℃和30℃高温日数总数分别为68256天和150192天。通过统计研究范围内出现高温的格点数（换算成面积），发现高温的影响范围在逐渐扩大。35℃高温事件在2010年影响范围最大，达4.29×10¹³m²，30℃高温事件在1998年的影响范围最大，达8.43×10¹³m²。

(4)北半球35℃ 高温日数经验正交函数分析表明， 第一模态能解释总方差的29.7%，其空间分布特征是北美中南部、非洲北部、中东地区一致型正值分布，高温日数有明显的上升趋势。分析表明，第一模态的时间系数(PC1)与北半球平均温度有高度相关（r=0.96），这可能说明伴随全球变暖，副热带高压向北扩张和哈德莱环流的增强，有利于北半球高温日数的增加。进一步统计相关的平均温度和日气温标准差的变化，发现北半球平均最高气温和非洲北部、北美洲南部以及中东地区平均最高气温呈正相关关系，和平均最高气温的标准差呈负相关关系。说明非洲北部、北美洲南部以及中东地区高温日数的增多，最高气温平均值变化的贡献要大于标准差变化的贡献。

(5)北半球35℃ 高温日数经验正交函数分析表明， 第二模态能解释总方差的9.2％，其空间分布反映出北美洲南部、中东地区和非洲北部、印度反向变化特征。分析表明，第二模态的时间系数（PC2）与同期SST 呈显著相关性。在大西洋，三个主要异常中心分别位于北大西洋高纬度、中纬度和低纬度海域，在太平洋，异常区域主要位于赤道中东太平洋，说明北大西洋三极型SST 异常、ENSO、Mega-ENSO可能是影响第二模态的因素之一。当NINO3.4、Mega-ENSO 指数增高（减小）时，印度高温日数增多（减少），非洲西北部高温日数减少（增多）。NATI 指数（北大西洋三极型海温指数）增高（减小）时，非洲西北部、北美中部偏南地区高温日数增多（减少），印度高温日数减少（增多）。在以上变化中，最高气温平均值变化的贡献要大于标准差变化的贡献。

In this paper, based on the ERA5 reanalysis data and ERSSTV5 SST dataset from 1979 to 2019，we analysis the climate characteristics and long-term changes of high temperature events including the changes of high temperature days, high temperature intensity and high temperature influence range, under two definitions of absolute threshold (35 ℃, 30 ℃) and relative threshold (90%,95%) in the northern hemisphere summer in recent 41 years. Furthermore, we also discussed the influence mechanism of climate factors such as climate warming and sea surface temperature on high temperature. The main conclusions are as follows: 

(1) The basic characteristics of high temperature at 35℃ show that high temperature is mainly distributed between 10°N and 48°N in the northern hemisphere, and there is no high temperature event in the north of 70.5°N.Northern Africa and the Middle East are the regions with the most high temperature events, with the average annual high temperature days exceeding 65 days. The latitudes with the most high temperature events are 18°N~28.5°N, with the average annual high temperature days exceeding 16 days.Over the past 41 years, the number of high temperature days increased by 2.0days/10a. Northern Africa, the Middle East, and the Southwest of the United States were the regions with the fastest increases, with an increase of more than 4days/10a.In addition to the long-term trend, the number of 35℃ high temperature days has an annual cycle of 2 years.The basic characteristics of high temperature at 30 ℃ show that the high temperature is mainly distributed in the south of 55°N in the northern hemisphere, and there is no high temperature event in the north of 76°N. The latitude zone with the most high temperature events is 16.5°~33°N.Over the past 41 years, the number of high temperature days increased at a rate of 1.9days/10a with a 4-year interannual cycle.

(2) The temperature corresponding to the relative threshold of high temperature gradually decreases from low latitude to high latitude. Northern Africa, the Middle East and India have the highest relative threshold in the Northern Hemisphere and the corresponding temperature is above 40℃, while in high latitude, Greenland has the lowest relative threshold and the corresponding temperature is below 0℃.Over the past 41 years, the 90th percentile threshold has increased at a rate of 3.0days/10a, while the 95th percentile threshold has increased at a rate of 1.9days/10a.Under these two relative thresholds, Western Asia, Southeast Europe, low latitudes of Africa, and Northeastern North America all showed the fastest increase in the number of high temperature days, with an increase of more than 2days per 10 years.

(3) During the summer period from 1979 to 2019, the intensity of high temperature in the Northern Hemisphere increased and the coverage area expanded.  Analysing the total number of high temperature days and the total daily maximum temperature of high temperature days, it was found that the high temperature intensity was gradually increasing, and 2010 was the year with the strongest high temperature intensity. The total number of high temperature days at 35℃ and 30℃ were 68,256 and 150,192 days. According to the number of grids where high temperatures occur each year and the area covered by these grids, we find that the impact of high temperatures is gradually expanding. For 35℃ high temperature in 2010, the influence range of high temperature was the largest(4.29×10¹³㎡); for 30℃ high temperature, the influence range of high temperature in 1998 was the largest(8.43×10¹³㎡).

(4) The first mode of EOF is closely related to climate warming. From 1979 to 2019, the 500hPa subtropical high expands northward, while the Hadley circulation is strengthened, which means that the tropical zone has a trend of northward expansion, which has a direct impact on the number of high temperature days in the Northern Hemisphere. The mean maximum temperature in the northern hemisphere is positively correlated with the mean maximum temperature in the northern Africa, the southern North America and the Middle East, and negatively correlated with the standard deviation of the mean maximum temperature. This shows that the increase in the number of high temperature days in northern Africa, southern North America and the Middle East is mainly caused by the increase in average maximum temperature.

(5) There were significant correlations between the EOF second mode of high temperature days and SST of the same period. In the Atlantic, three major anomaly centers are located in the high, middle and low latitudes of the North Atlantic. In the Pacific, the anomalous region is mainly located in the equatorial Middle East Pacific. It is found that the North Atlantic tripolar SST anomaly 、 ENSO、Mega-ENSO may be the factors affecting the second mode.When the NANO3.4 and Mega-ENSO indices increased (decreased), the number of high temperature days in India increased (decreased) and the number of high temperature days decreased (increased) in north-west Africa. As the NATI index increases (decreases), the number of hot days increases (decreases) in north-west Africa and south-central North America, and the number of high temperature days decreases (decreases) in India. In the above changes, the contribution of the change of the average maximum temperature is greater than the contribution of the change of variance.