A new methodology to define homogeneous regions through an entropy based clustering method

One of the most crucial steps in flow frequency studies is the definition of Homogenous Regions (HRs), i.e. areas with similar hydrological behavior. This is essential in ungauged catchments, as HR allows information to be transferred from a neighboring river basin. This study proposes a new, entropy-based approach to define HRs, in which regions are defined as homogeneous if their hydrometric stations capture redundant information. The problem is handled through the definition of the Information Transferred Index (ITI) as the ratio between redundant information and the total information provided by pairs of stations. The methodology is compared with a traditional, distance-based clustering method through a Monte Carlo experiment and a jack-knife procedure. Results indicate that the ITI-based method performs well, adding value to current methodologies to define HRs.