An open-database of Grape Harvest dates for climate research: Data description and quality assessment

Abstract

We present a dataset of grape harvest dates (GHD) series that has been compiled from international and non-translated French and Spanish literature and from unpublished documentary sources from public organizations and from wine-growers. As of June 2011, this GHD dataset comprises 378 series mainly from France (93% of the data) as well as series from Switzerland, Italy, Spain and Luxembourg. The series have variable length and contain gaps of variable sizes. The longest and most complete ones are from Burgundy, Switzerland, Southern Rhône valley, Jura and Ile-de-France. The GHD series were grouped into 27 regions according to their location, to geomorphological and geological criteria, and to past and present grape varieties. The GHD regional composite series (GHD-RCS) were calculated and compared pairwise to assess the quality of the series. Significant (p-value < 0.001) and strong correlations exist between most of them. As expected, the correlations tended to be higher when the vineyards are closer, the highest correlation (R = 0.91) being obtained between the High Loire Valley and the Ile-de-France GHD-RCS. The strong dependence of vine cycle on temperature and, therefore, the strong link between GHD and the temperature of the growing season was also used to test the quality of the GHD series. The strongest correlations are obtained between the GHD-RCS and the temperature series of the nearest weather stations. Moreover, the GHD-RCS/temperature correlation maps show spatial patterns similar to temperature correlation maps. The stability of the correlations over time is explored. The most striking feature is their generalized deterioration at the late 19th-early 20th turning point. The possible effects on the GHD of the phylloxera crisis, which took place at this time, are discussed. The median of the standardized GHD-RCS was calculated. The distribution of the extreme years of this general synthetic series is not homogenous. Extremely late years all occur during a two-century long time-window from the early 17th to the early 19th century, while extremely early years are frequent during the 16th and since the mid-19th century. The dataset is made accessible for climate research through the Internet. It should allow a variety of climate studies, including reconstructions of atmospheric circulation over Western Europe. © 2011 Author(s).