【摘要】 本文对早期发表的《北京 2 5 0年降水》资料进行复算 ,并讨论其复原方法。在此基础上进一步探讨清代“晴雨录”降水资料复原方法的改进问题。试作了两种多因子回归复原方案 ,经对比这些方案及其试算结果 ,认为以 8因子的逐步回归方案效果最佳。将此结果与《北京 2 5 0年降水》1) 及同期的器测降水量记录等进行比较 ,给出新的北京 1 72 4～ 2 0 0 0年降水量复原序列。更多还原

【Abstract】 The 1724～1904 Records of Clear and Rain Weather by ex Qing Dynasty royal officers contain the beginning and end of rainfall/snowfall events and dry weather recorded on a daily basis and is thus a dataset of paramount importance to climate research. As early as 1974, the Chinese meteorologists made efforts to reconstruct daily weather series from the records for that period, arriving at sequences of the numbers of rainfall (snowfall) days, hours and amounts for each of the years on a monthly basis with the inclusion of detailed description of the method utilized.\;In checking the results from our colleagues achieved in 1974, recalculating the records and examining their scheme for reconstruction, we found the questions statistical deviation in their work and some questions of their technique. In dealing with the reconstruction and improvement, we proposed a 8 factor stepwise regression scheme for which the daily precipitation intensity is denoted by use of a 7 level categorization, followed by making a stepwise regression formula with the numbers of days of different rainfall intensities in a month and the monthly total of rainy days as the variables to re establish monthly precipitation total by means of the information on rainfall given in the "Clear and Rain Records", followed by use of device measurements after 1904 to constitute 1724～2000 monthly, seasonal and annual sequences for Beijing. A study is undertaken of the findings presented in this work vs the evidence of the book 250 year Precipitation in Beijing compiled in 1974 and against discontinuous rainfall measurements in 1841～1904, showing good utility of our scheme.\;\;更多还原