Proceedings of the sixth seminar for homogenization and quality control in climatological databases
The aim of the homogenization procedures is to detect the inhomogeneities and to correct the series. In practice there are absolute and relative methods applied for this purpose. However the application of absolute methods is very problematic and hazardous since the separation of climate change signal and the inhomogeneity signal is essentially impossible. Relative methods can be applied if there are more station series given, which can be compared mutually. The methodology of comparison is related to the following questions: reference series creation, difference series constitution, multiple comparisons of series etc. These topics are very important for detection as well as for correction, because the efficient comparison of series can increase both the significance and the power. The development of efficient comparison methods can be based on the examination of the spatial covariance structure of data series. Consequently the statistical spatiotemporal modelling is also a key question of data series homogenization. The adequate comparison, break point detection and correction procedures are depending on the statistical model.