中文摘要：

依据清代《翁同龢日记》（以下简称《日记》）中逐目的冷暖感知记录和同时期的器测月均温资料,采用相关分析和线性回归分析等统计方法,以月为基本统计单元,对《日记》中冷暖感知记录反演气候变化的能力进行了分析.结果说明,《日记》中的冷暖感知记录可以用于气候变化研究,但是最佳代用指标因气候变化的时间尺度和季节而异.总体而言,极热、偏热、偏凉、极冷日数对年内月到季时间尺度气候变化的代表性较好,其中尤以偏凉和极冷日数最佳.对于年际尺度上的气候波动,从季节对比来看,冷暖感知日数反演夏季（6-8月份）月均温的能力最差;从冷暖感知类型对比来看,极冷日数是多个月份月均温的最佳代用指标,1、3、5、9和12月份的最佳代用指标均是极冷日数.并且,还可以依据极冷日数的多寡识别极端冷、极端热年.由此可见,古代私人日记中的冷暖感知记录可以用于反演历史时期气候的冷暖变化.

英文摘要：

Based on the perception records on cold/warm in Weng Tonghe Diary and the contemporary monthly average instrumental temperatures in Beijing and Shanghai. The correlation between the perception records on cold/warm and temperature variation is investigated with correlation analysis and regression analysis. The result shows that they are significantly correlated. It means that the monthly days of colder （CD）, extremely cold （EC）, hotter （HT）, and extremely hot （EH） could be used as a proxy index to indicate the past temperature change. However, the validity of the proxy index is various on the time scale of climate change and the season. Firstly, the monthly days of EH, HT, CD, EC could represent the month-quarter variation of temperature change, with CD and EC being the best. Secondly, on the time scale of inter-annual, the perception days on cold/warm can not represent the temperature variation for the summer months （Jun., Jul. and Aug. ） , the monthly days of EC are the best proxy indexes for Jan., Mar., May, Sep. and Dec.. Besides, the monthly days of EC can also be used to identify the extremely cold/ warm year. Thus, it can be concluded that the perception records on cold/warm in historical private diaries can be used to study climate changes of historical period.