Parameterization of visibility in snow: Application in numerical weather prediction models

Several parameterizations of extinction coefficient ( σ ) and visibility ( l v ) as a function of temperature ( T ), liquid water equivalent snowfall rate ( S ), have been developed assuming a gamma size distribution for ice particles and using aircraft data collected in extratropical stratiform clouds. Using surface‐based measurements (SBM) of S , T , relative humidity (RH), cloud ceiling ( l ce ), and σ during the winter months in 2005, 2006, and 2007 at the Centre for Atmospheric Research Experiments site in Ontario, Canada, other parameterizations have been developed and compared with that based on the aircraft data. The analysis of the SBM data indicates that low l v is mainly associated with S . Both aircraft and SBM data indicate that there is a significant dependence of l v on S and a relatively weaker dependence on T . The observed l v is correlated with l ce , but the dependence of l v on RH is relatively weak. There is some nonlinear dependence of l v on wind speed. Using SBM, several parameterizations of σ have been developed using a multiple linear regression method by increasing the number of terms starting with S . The addition of T increases the correlation coefficient (CC) r from 0.85 to 0.87. The addition of RH has no significant effect, but the inclusion of l ce further improves the CC from 0.87 to 0.9. It was also found that both l v and l ce can be described well using the inverse Gaussian probability density function. Model predictions using these parameterizations show that, when the model correctly forecasts the precipitation field, the predicted l v agreed well with observations.