A reduction in the amount of available clean water is a universal problem, and the harvest of rainwater is one of the methods that can solve this issue. In this study, the smallest reservoir capacity that can collect rainwater falling on roofs is defined, using daily rainfall data spanning a 32-year period for the cities Burdur, Isparta, and Antalya. Reservoir capacities with different roof areas being able to serve different needs are expressed using graphics. Particle swarm optimization (PSO) is compared with classic methods like the mass curve method, and the minimum flaw method. While the mass curve method provides the best results among the classic methods, PSO offers similar results. This study is divided into two sections. The first section includes the optimization model, which is defined for annual fixed needs. The software created in this section is compared with the results obtained from the other methods and its accuracy is proven. In the second section, the software is created according to users' changing needs, which is not available for the other methods. The biggest advantage of the software is to be able to define optimum reservoir capacity in accordance with changing water needs and user risk perception. In this paper, a sample study is conducted to show the difference between fixed needs and changing needs, with the advantage acquired via optimization according to changing needs being shown.