In a hydro-dominated power system, with complex inter-temporal linkages, capacities of individual projects and project configurations are more complicated than using the simple megawattage, often used to denote capacity in thermally dominated systems. In this paper, the long term hydro scheduling problem is solved and used to test novel definitions of hydro-capacities, based on expected cost and cost probability distributions for system operating cost, based on rising load and a within-the-year time distribution of load. The paper expands on previous capacity definitions and tests them numerically in a case study with realistic data from a hydro-dominated system. The principal contribution of this paper are new capacity concepts and practical applications for an example of sizing a given reservoir. The testing is performed using actual data from the Icelandic hydro-dominated power system, with the model HYDOVG delivering optimal schedules for various inflow instances and load cases and using linear programming (LP) CPLEX software to optimize each case and instant.