Hydrologic Information Transfer among Rainfall Stations of a Selected Quaternary Catchment of South Africa

Abstract

Hydrological data such as rainfall, streamflows, etc play an important role in water resources development and management, especially in a semi-arid country like South Africa. The current paper preliminarily evaluates, using entropy approach, the information transfer among 7 rainfall stations of the quaternary catchment (C52A) of the Upper Orange River system of South Africa. These rainfall stations were used in a previous study to model streamflow hydrographs for the infield rainwater harvesting in C52A, but no reference to the amount information transferred among the different stations was made. It was observed that the frequency distributions of the monthly rainfall data followed approximately a strong similar exponential decay as well as a strong similar trend. This could yield to highly correlated rainfall. The rainfall stations were found to be highly correlated when station pairs are considered and the mutual information or the information transfer could be evaluated. As a dimensionless quantity, the notion of minimum relative transinformation (RMT) has been introduced for the first time in this paper and it was shown that RMT values vary between 10.3 % and 45.7 %. The relatively high information transfer confirms that these rainfall stations belong to the same hydrological network or hydrological region. The strong dependency among the rainfall stations could suggest that a sensitivity analysis on the number of rainfall stations data could be used during modeling of Infield Rainwater harvesting, in C52A since the information transfer among rainfall stations is relatively considerable. It is suggested that the current study be extended to rainfall stations for quaternary catchments other than C52A.