The development and application of automated baseflow separation algorithms for describing individual discharge components still cover a major part of modern hydrology. A key problem is the applicability of baseflow separation methods in ungauged or anthropogenically impacted catchments due to their complex conceptuali-ty. With increasing anthropogenic impact, the dependence of calculated baseflow rates on measured total runoff also increases. In this study, we suggest statistical approaches for testing the suitability of different hydrograph separation algorithms. For our main study site, the mining-impacted Geisel catchment in the Central Ger-many, we calculated a mean baseflow rate of 0.28 m³/s for the period from 1981 to 2017. First, 14 different algorithms (graphical and statistical methods, digital-filter-approaches, and one physically based algorithm) were tested in seven catchments. The calculated baseflow rates for the Geisel catchment showed questionable curves, in particular a high amplitude. Thus, similarities between measured dis-charge and calculated baseflow were demonstrated (quasi-parallelism), which were quantified with correlation analyses on different time scales. Following this, a pro-found analysis of the baseflow index (BFI) was carried out. Finally, we applied a sta-tistical regionalisation approach to derive validated baseflow information for the Geisel catchment using the calculated baseflow indices and numerical catchment descriptors. As a result, the questionable baseflow hydrographs of the Geisel could be corrected. This promising method enables improved estimations of environmental flow components, improved analyses of the hydrological processes to foster the un-derstanding of anthropogenic impacts, and provides essential information for water management in the Geisel catchment. Furthermore, characteristic properties of long-term BFI values were revealed, which can be used to develop new physically based hydrograph separation procedures by including spatially distributed physical catchment descriptors.