autocorrelation for lead time =1, r = 0.81, therefore, the dynamical model is only improving the forecast by 1%. autocorrelation for lead time =4, r = -0.14, therefore, the dynamical model is improving the forecast by 55%. if lead time autocorrelation coefficients 1 -4 are averaged,however, r = 0.28, therefore, the dynamical model is improving the forecast by only 10% is it true that autocorrelation at lead times =1, 2 is potentially overstating the dynamical models ability to predict ENSO? reference Alessandri and Navarra, 2008