Trend analysis of atmospheric deposition data: A comparison of statistical approaches

Numerical simulation was used to compare the most used trend analysis techniques on data series of ionic concentrations in atmospheric deposition. The Seasonal Kendall Test (SKT) showed the highest power, which increased in particular when using original weekly data instead of pooling together the samples in monthly or yearly volume-weighted averages. The simulation also showed that differences in power among tests and pooling intervals would be negligible for data series longer than about 12 years.We tested these results using data from a network of bulk deposition samplers at nine forest sites in Italy, for which data have been available since 1998. These sites were selected in different forests, ranging from arid Mediterranean evergreen oak forest to rainy Alpine beech or spruce forests. The results showed relevant differences as regards the number of significant trends detected using different techniques and different data pooling, even for 13-year data series.The use of minimum–maximum autocorrelation factor analysis allowed a better interpretation of the data, showing the main trend shapes among stations and variables.

► We used numerical simulation and deposition data to compare trend analysis techniques.
► The Seasonal Kendall Test showed the highest power.
► Its power was increased when using weekly data instead of pooling the samples.