WRRCTR No.133 Bayes-Markov Analysis for Rain-Catchment Cisterns

Abstract

In many parts of the world, public water-supply systems have shown signs of an inability to adequately service increasing demand. As a result, water shortages have occurred and moratoriums on new development areas have been imposed. And because of recurrent droughts and the rapid acceleration of urban development in recent years, the rain-catchment system and its cost effectiveness are now regaining the attention of researchers and planners as a "new" and important alternative water supply. Rainfall, catchment area, storage capacity of the cistern, and water demand are the four main elements considered in the design, operation and management of a cistern system. Expected weekly rainfall is the main uncontrollable element of concern to the cistern owner; therefore, to meet this problem, expected weekly rainfall probabilities were first simulated by the Bayesian analysis. Then the sequential property contained in the 25-yr rainfall record was utilized to generate the likelihood probability function by using the lag 1 Markov sequential analysis. The Bayes-Markov analysis was subsequently applied to obtain the weekly rainfall probabilities. As a result of these analyses, the Bayes-Markov probabilistic approach for weekly rainfall simulation has shown its usefulness. Design and operational methodologies are also presented in the report.