Machine learning approach for forecasting optical turbulence

The strength of a machine learning algorithm is that given enough data and computation, any trend or pattern appearing within that data can be approximated without explicitly involving underlying understanding. The refractive index structure parameter, C²_n, is a measure of the strength of fluctuations in the refractive index along a path and one of the most significant contributors to beam quality. Weather data including C²_n is collected over a period of several months at the TISTEF (Townes Institute Science and Technology Experimentation Facility) laser range. Once data is aggregated, important features are selected and the data is processed. The AutoGluon framework is used for exploratory analysis of the effectiveness of neural network structures and Keras is used in final model building. The model chosen as a result of this process indicates that machine learning is able to generalize trends in the fluctuations of C²_n over time.