We present a machine learning-based measure-correlate-predict approach that predicts a multi-year time-series of optical turbulence strength (Cn2) with high accuracy (r = 0.78 at 16 locations) based on a single year of in-situ Cn2 measurements and reanalysis data.@en