A method for the reconstruction and temporal extension of climatological time series is provided. This method was focused on a combination of methods, including harmonic analysis, seasonal weights, and the Durbin-Watson (DW) regression method. The DW method has been modified in this paper and is described in detail because it represents a novel use of the original DW method.
The method is applied to monthly means of daily wind-run data sets recorded in two historical observatories (M series and A series) within the Parque del Retiro in Madrid (Spain) and covering different time periods with an overlapping period (1901)(1902)(1903)(1904)(1905)(1906)(1907)(1908)(1909)(1910)(1911)(1912)(1913)(1914)(1915)(1916)(1917)(1918)(1919). The aim of the present study is to fill up to and to construct a historical time series ranging from 1867 to 1992. The proposed model is developed for the 19061919 calibration period and validated over the 1901-1905 verification period, which includes the hypothesis of constant ratio of variances. The verification results are almost as good as those for the calibration period. Hence, the M series was extended back to 1867, which results in the longest climatological wind-run data-set in Spain. Also, the reconstruction is shown to be reliable.