Investigation of the WRF-Simulated Visibility Parameter in Comparison to Observational Data

Climatological studies of surface visibility in regions lacking surface observations require leveraging a Numerical Weather Prediction (NWP) model, provided that the model has skill in its simulations. With this in mind, experiments are performed with the Weather Research and Forecasting (WRF) model to validate simulated visibilities against actual observations. A triple-nested domain with horizontal resolutions of 16 km, 4 km, and 1 km is applied for a 27 hour forecast centered over Asheville, NC. The size of the inner domain is 273 x 273 km and encompasses six reporting stations. Simulations for the month of March 2008 are examined and compared to surface observations inside the inner domain.

Extinction coefficients are calculated from WRF-derived mixing ratios of cloud water, rain, snow, and cloud ice. These extinction coefficients are then applied to an algorithm that calculates surface visibility. Comparisons of the WRF-derived visibility and the observed visibility are investigated. Direct comparisons between the two datasets are not straightforward due to the large spatial variability inherent in reduced visibilities. Results indicate that the WRF model reproduces the general patterns of reduced visibilities for a range of meteorological events (i.e., rain, fog, snow). Skill is also shown in determining when reduced visibility will occur, although the magnitudes of the reduced visibilities are not well captured. In addition, application of the WRF chemistry model is discussed to include the impact of aerosols to the visibility calculation.