Evaluation of long-term trends in spring onset in the Northern Europe using Singular Spectral Analysis

We examine the long-term trends in spring onset in three locations at the Northern Europe (Stockholm, Postdam and Schwerin) using daily mean temperature series. These series have been chosen by their length and homogeneity (ECA database, http://eca.knmi.nl/). Stockholm stands out by its length (from 1756 to 2009). We use Singular Spectral Analysis (SSA) to filter the high frequency variability of the series, and the onset of spring is selected as the date when the mean daily temperature filtered series exceeds the 5 ºC threshold. We also calculate the onset of spring as the date when at least 5 days have temperature above 5 ºC. SSA is also used, in a second step to obtain the long-term variability of these two indices.

Results show few differences between the use of these two criteria for Stockholm series. Essentially, we find a first period, until 1860, with a significant trend to retard the spring onset, and a second period (from 1860 until the end of the records) with a significant tendency to advance the beginning of spring. Both periods show a similar absolute value (~0.7 days/decade). Our results are also in agreement with those obtained by other authors that use a different methodology.

The behaviour of spring onset in the Northern Europe seems to be related with the long-term variability of temperature. It is interesting to stand out that the variability observed in the onset of spring using these two indices is similar to that found using a very simple index, which can be computed as the first day of the year when the mean temperature exceeds on 5º C. Currently, we are studying the possibility of using this simpler index a precursor of the spring onset and their thermal characteristics.