Water Demand Estimation should be based on future forecasting approach which will depend on so many factors such as predicted population on the targeted future year, whether or not rapid urbanization is expected, human nature of water uses over 24-hr period, etc. etc. Indeed, it is certainly not a linear problem. So, linear programming may not a good solution for this type of problem. Nonlinear stochastic optimization modeling may be a better solution for a tedious problem in this nature.
Hi, I guess you already know but to emphasize again, the problem should be linear and if it is nonlinear you should calculate how much nonlinear it is. My article 'https://www.researchgate.net/publication/257141691_Optimized_Sustainable_Groundwater_Extraction_Management_General_Approach_and_Application_to_the_City_of_Lucknow_India' has some insights.
It all about how you define your problem. You can estimate water demand for any future year using system dynamics, and them can optimize the factors affecting water demand using linear programming. In this case, it would be a decision-making problem.