Homogenization of Monthly Precipitation Series: A Case Study from Romania

The aim of this study is twofold. First, it introduces a method that combines regression modeling with periodic autoregressive errors to identify change points (CPs) in the data series and model parameters estimated via a multi-island genetic algorithm guided by the Minimum Description Length (MDL) principle. Second, building on this framework, we propose a homogenization procedure for monthly precipitation series from Dobogea region, optimizing the MDL objective to correct inhomogeneities across the regional dataset.