SEARCH

SEARCH BY CITATION

Keywords:

  • change-point detection;
  • inhomogeneities;
  • time series homogenization;
  • precipitation extremes;
  • trend analysis;
  • field significance testing

Abstract

The historical daily precipitation observations for the Netherlands were fully digitized recently. The homogeneity of the precipitation series was tested by pairwise comparisons of the monthly totals using an algorithm of Menne and Williams. Forty percent of the precipitation series was indicated as inhomogeneous if the algorithm was applied to the untransformed monthly totals. The use of a square-root transformation to reduce the skewness was only successful in an application to the data for the 1951–2009 period. Changes in the annual precipitation amounts, the precipitation amounts in the winter and summer halves of the year, the number of days per year with a precipitation amount greater than 20 or 30 mm, and the 5-d annual maximum precipitation amount were determined both for the period 1951–2009 using the data from 240 stations and the period 1910–2009 with the data from 102 stations. Significant increases were found for all six indices. The centennial increases in mean annual, winter and summer precipitation are 25, 35, and 16%, respectively. The exceedance frequency of the 30 mm threshold almost doubled during the 1910–2009 period. Much attention is given to the field significance of trends, the statistical significance of regional differences in trends and nonlinearity of trends. In contrast to the increase in mean winter precipitation, which is statistically significant for the majority of the stations, the increase in mean summer precipitation is mainly restricted to coastal regions. The mean summer precipitation and the exceedance frequencies of the 20 and 30 mm thresholds show a relatively strong increase from the beginning of the 1980s. Copyright © 2012 Royal Meteorological Society