A measure-correlate-predict approach for optical turbulence ( C n 2 ) using gradient boosting

Abstract

We present a machine learning-based measure-correlate-predict approach that predicts a multi-year time-series of optical turbulence strength (C n 2) with high accuracy ( r¯= 0.78 at 16 locations) based on a single year of in-situ C n 2 measurements and reanalysis data.