Optimal Prediction of Atmospheric Turbulence by Means of the Weather Research and Forecasting Model

Abstract

The performance of ground-based astronomical observations and free-space optical communication (FSOC) systems suffers from atmospheric turbulence and meteorological conditions. The a priori knowledge of atmospheric conditions several hours before observations allows the programming of astronomical observations (flexible scheduling) to be optimized. In this paper, we present a prediction study based on the Weather Research and Forecasting (WRF) model. It allows the prediction and characterization of a useful set of meteorological parameters relevant to atmospheric physics (e.g., pressure, temperature, relative humidity, wind speed, and direction). Predicted parameters are then injected into an optical turbulence (OT) model to compute the refractive index structure constant C n 2. We performed sets of simulations for Cerro Pachon Observatory in Chile, using the data from the National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR). The main goal is to quantify how accurately numerical weather prediction models can reproduce conditions over the complex terrain of the Cerro Pachon area. In order to produce a reliable forecast, meteorological prognostic skills need accurate representations of the physical parameterization options. Three widely used Planetary Boundary Layer (PBL) schemes and two Land Surface Models (LSM) were tested, analyzed, and compared in order to find the optimal WRF configuration. Predictions are compared to in situ measurements coming from balloon-borne radiosoundings. It is determined that the predicted C n 2 are in good agreement with the radiosoundings measurements with a mean relative error (MRE) under 6.4% at all altitudes when using balloon measurements to deduce some parameters such as the outer scale of turbulence L 0, which is used in the OT model. For a fully operational prediction, the MREs between the predictions and the measurements range from 1.4% to 8% according to the different ways to estimate the L 0 profiles. Seasonal statistics are also presented for different meteorological and turbulence parameters.