This paper addresses the problem of integrating and down scaling the effects of atmospheric-oceanic global phenomena and solar variability, to enhance regional hydrologic forecasting. Our approach uses a rule set induction technique based on second-order tables (database relations in which tuples have sets of atomic values as components). In this theoretical framework, learning can be viewed as a table compression problem in which a table consisting of training set data is transformed into a second-order table with fewer rows by merging rows in consistency preserving ways. Given a data set of global climate conditions, solar variability, and inflows for Lake Okeechobee, we apply table compression to produce rules for predicting future Lake Okeechobee inflows, which can then be incorporated in an operational schedule for managing the lake. The paper presents preliminary results.