Implementation of various non-linear similarity functions for stable atmospheric surface layer in the WRF modeling system: An evaluation for three contrasting nights in CASES-99 dataset

Accurate representation of the stable atmospheric surface layer (ASL) in numerical models has been a challenging task due to the occurrence of different physical processes such as radiative transport, turbulent mixing, and the coupling of vegetation with the atmosphere and underlying soil, as well as aspects of land use heterogeneity, etc. The Monin-Obukhov similarity theory has been widely used to parameterize surface turbulent fluxes in numerical models, which utilize similarity functions to account for the effect of atmospheric stability. Over the years, researchers have proposed various forms of similarity functions, depending on different field experiment datasets around the globe. This study incorporates some of the well-established non-linear similarity functions under stable stratification in the revised MM5 surface layer scheme in the Weather Research and Forecasting (WRF) Model version 4.2.2 and hence develop a scheme that has various functional forms of similarity functions for computing surface turbulent fluxes under stable stratification. The proposed scheme has been evaluated in simulating near-surface atmospheric variables and surface turbulent fluxes over the domain centered around the location of the Ranchi dataset site (23.412oN, 85.440oE; India) and the CASES-99 experiment site (37.38oN, 96.44oW; Kansas, USA) during the months of January 2009 and October 1999, respectively. The modified scheme is also evaluated for the three contrasting nights representative of intermittently turbulent, fully turbulent, and radiative ASL based on the CASES-99 dataset. In general, all the newly installed similarity functions are found to be consistent in predicting surface turbulent fluxes as well as near-surface atmospheric variables with respect to the default surface layer scheme and the observed data derived from the flux towers over the two domains. However, this study reveals that all the similarity functions are found to be inconsistent during the fully turbulent night while they seem comparable for the first and last nights based on the CASES-99 dataset.