First results in the attempt to adept a data mining strategy for distinguishing global, regional and local factors reflecting variation patterns in river water chemistry