We analyze some restrictions found in the estimation process, which are general to any measurement technique. The insensitivity of the instrument to large values of outer scale is one of them, as the telescope becomes blind to outer scales larger than its diameter. Another problem is the contradiction between the length of data and the stationarity assumption of the turbulence (turbulence parameters may change during the data acquisition time).
Our method effectively deals with problems such as noise estimation, asymmetric correlation functions and wavefront propagation effects. It is shown that the latter cannot be neglected in high resolution AO systems or strong turbulence at high altitudes. The method is applied to the Gemini South MCAO system (GeMS) that comprises five wavefront sensors and two DMs. Statistical values of L0(h) at Cerro Pachón from data acquired with GeMS during three years are shown, where some interesting resemblance to other independent results in the literature are shown.