Zie paper: https://rmets.onlinelibrary.wiley.com/doi/pdf/10.1002/joc.5792
The method opted for homogenization of the monthly airtemperature time series in Belgium relies on the use of the HOMER software (Mestreet al.,2013) developed with sup-port of the European Union, through the COST ActionES0601—Advances in Homogenization Methods of Climateseries: an Integrated Approach (HOME)—for the detection of a multiple number of break-points and the calculation of adjustments. This interactive semi-automatic softwareincludes the best features of some state-of-the-art methods, namely PRODIGE (Caussinus and Mestre, 2004), ACMANT (Domonkos, 2011), and cghseg a joint segmentation method that was developed originally by bio-statisticians in the context of DNA segmentation (Picardet al.,2011). These methods identify change-points in a timeseries by computing an optimal segmentation of this timeseries in homogeneous subseries. A dynamic programming approach enables to determine the change-points by minimizing the internal variance of the subseries (Caussinus and Lyazrhi, 1997). Based on the identified change-points, these methods apply a network-wide unified correction model (ANOVA, Caussinus and Mestre (2004)).
Quote selectie