This paper presents a calibration approach to a holistic water resources-economic model, which involves essential water resources and economic components in a consistent model. The model is often formulated as an optimization model, with the objective of maximizing economic welfares/profits from water uses. When we apply the model to a baseline scenario against real world conditions, the economic outputs are often expected to match the observations, since a wide divergence between model outcomes under the baseline case and actual results is not appropriate for policy options starting from the baseline. Following the concept of "positive mathematical programming", which has been widely used in agricultural and applied economics, this paper presents a procedure to convert an existing normative model to a positive one through calibrating the model to the baseline level using both programming constraints and "positive" inferences from base-level observations. A genetic algorithm is used to implement the calibration process.