Short term hydro-thermal dispatch is a classical problem in power systems operation. This problem is usually solved by a combination of heuristic and classical optimization algorithms that are unsuitable for real time operation. This paper presents a quadratic convex approximation that guarantees a global optimum with fast convergence. Our approximation is based on the semi-definite properties of the nonlinear function that relates the generated power with the volume and water discharge of each reservoir. Simulation results performed in Matlab/cvx demonstrate that the proposed approximation is accurate compared to the conventional Genetic Algorithm (GA) and Particle Swarm optimization (PSO) with the key advantage of being an exact technique suitable for real time operation.