Configuring LES Based on Dropsonde Data in Sparsely Sampled Areas in the Subtropical Atlantic

Abstract

AbstractThis study explores the question of how field campaign data, gathered over a large area that is poorly sampled, can be used to make large-eddy simulation (LES) realizations more representative of the local conditions. To this purpose, dropsonde data recorded during the first Next-Generation Aircraft Remote Sensing for Validation Studies (NARVAL)-South campaign, in the marine subtropical North Atlantic, are blended into the forcing data. Control simulations are driven by time-dependent forcings derived from a combination of analyses and short-range weather forecasts, using weak nudging to prevent excessive model drift. A second set of simulations is driven by forcing data with dropsonde profiles included at the time points of their release. Metrics are designed to (i) quantify the impact on the boundary layer vertical profiles as a result of nudging toward the dropsondes and (ii) use a probabilistic method to allow a fair comparison of the inversion strengths in the simulations and observations. The simulations show strong time variation in the cloud-layer depth on relatively short time scales, which is commensurate with recent observational studies in the area. Nudging toward dropsondes improves the representation of the atmospheric profiles throughout the depth of the boundary layer in all simulations. However, the impact on the inversion strength is less pronounced. All impacts persist for some time after the dropsonde time point, depending on the intensity of the nudging and the nudging time window. The sensitivity of the results to nudging details and vertical resolution is assessed.