Comparison of LIDAR Detection Statistics using Ensemble Average and ‘Single Realization' Dispersion Models

Active remote sensors have various applications, from air pollution studies to defense uses. However, investing in new sensors, or any new technologies, can be very expensive especially if the success of the instrument is not guaranteed. A relatively inexpensive alternative to field testing is the numerical model, which itself can become “expensive” due to the computational effort required. For NCAR’s Raman-shifted Eye-safe Aerosol Lidar (REAL), the difficulty in modeling its performance comes from the turbulence in the atmospheric boundary layer. There are two types of numerical models which can resolve the effects of turbulence, each with its own set of benefits and drawbacks: ensemble average and ‘single realization’ models. Ensemble average models can be run in real-time and are computationally efficient, however they lack resolution, while ‘single realization’ models maintain a high level of resolution, yet become computationally expensive and cannot be run in real-time. This study focused on comparing these two atmospheric dispersion models using REAL detection statistics. The process involved generating scenarios with the dispersion models, adding perturbations to mimic turbulence or creating multiple representations, depending on the model, and running output through code representing the REAL in order to obtain detection statistics. This presentation will discuss the model comparison and present findings regarding the applicability of the ensemble average model as a potential substitute for the ‘single realization’ model in testing REAL.