Using an Artificial Neural Network Approach to Estimate Cn2 in Atmospheric Surface Layer

Using an Artificial Neural Network Approach to Estimate C_n² in Atmospheric Surface Layer Yao Wang and Sukanta Basu Accurate estimation of optical turbulence (C_n²) within the atmospheric surface layer is of great significance for both civil and military applications. C_n² values can be reliably measured by various types of research-grade instruments (e.g., small-aperture scintillometer or sonic anemometer); however, due to logistical and financial issues, these types of instruments and associated data are not widely available. In contrast, mean quantities of meteorological variables (e.g., temperature, wind speed) are continuously measured by various organizations around the world with a very high spatio-temporal coverage. Empirical relationships, therefore, are needed to estimate C_n² from these mean quantities.

In this study, an artificial-neural-network (ANN) approach is proposed to estimate optical turbulence within the atmospheric surface layer. Mean quantities of meteorological variables are utilized as the inputs to construct the ANN architecture. In this presentation, we will document the ANN-based model's ability of capturing evolutions of measured C_n² from two field campaigns (Mauna Loa, Hawaii in 2006 and Lubbock, Texas in 2009). In addition, we will evaluate its strengths and weaknesses by comparing with other contemporary approaches (e.g., Monin-Obukhov similarity theory-based formulation).