The Impact of the Land Surface Physics in the Operational NCEP Eta Model on Simulating the Diurnal Cycle: Evaluation and Testing Using Oklahoma Mesonet Data

Abstract

On 31 January 1996, the National Centers for Environmental Prediction/Environmental Modeling Center (NCEP/EMC) implemented a state-of-the-art land surface parameterization in the operational Eta Model. The purpose of this study is to evaluate and test its performance and demonstrate its impacts on the diurnal cycle of the modeled planetary boundary layer (PBL). Operational Eta Model output from summer 1997 are evaluated against the unique observations of near-surface and subsurface fields provided by the Oklahoma Mesonet. The evaluation is partially extended to July 1998 to examine the effects of significant changes that were made to the operational model configuration during the intervening time. Results indicate a severe positive bias in top-layer soil moisture, which was significantly reduced in 1998 by a change in the initialization technique. Net radiation was overestimated, largely because of a positive bias in the downward shortwave component. Also, the ground heat flux was severely underestimated. Given energy balance constraints, the combination of these two factors resulted in too much available energy for the turbulent fluxes of sensible and latent heat. Comparison of model and observed vertical thermodynamic profiles demonstrates that these errors had a marked impact on the model PBL throughout its entire depth. Evidence also is presented that suggests a systematic underestimation of the downward entrainment of relatively warmer, drier air at the top of the PBL during daylight hours. Analyses of the monthly mean bias of 2-m temperature and specific humidity revealed a cool, moist bias over western Oklahoma, and a warm, dry bias over the eastern portion of the state. A very sharp transition existed across central Oklahoma between these two regimes. The sharp spatial gradient in both the air temperature and humidity bias fields is strikingly correlated with a sharp west?east gradient in the model vegetation greenness database. This result suggests too much (too little) latent heat flux over less (more) vegetated areas of the model domain. A series of sensitivity tests are presented that were designed to explore the reasons for the documented error in the simulated surface fluxes. These tests have been used as supporting evidence for changes in the operational model. Specifically, an alternative specification for the soil thermal conductivity yields a more realistic ground heat flux. Also, the alternative thermal conductivity, when combined with a slight adjustment to the thermal roughness length, yields much better internal consistency among the simulated skin temperature and surface fluxes, and better agreement with observations.