The partitioning of available energy into surface sensible and latent heat fluxes impacts the accuracy of simulated near surface temperature and humidity in numerical weather prediction models. This case study evaluates the performance of the Weather Research and Forecasting (WRF) model on the simulation of surface heat fluxes using field observations collected from a surface flux tower in Oregon, USA. Further, WRF-modeled heat flux sensitivities to North American Mesoscale (NAM) and North American Regional Reanalysis (NARR) large-scale input forcing datasets; U.S. Geological Survey (USGS) and the Moderate Resolution Imaging Spectroradiometer (MODIS) land use datasets; Pleim-Xiu (PX) and Noah land surface models (LSM); Yonsei University (YSU) and Mellor-Yamada-Janjic (MYJ) planetary boundary layer (PBL) schemes using the Noah LSM; and Asymmetric Convective Model version 2 (ACM2) PBL scheme using PX LSM are investigated. The errors for simulating 2-m temperature, 2-m humidity, and 10-m wind speed were reduced on average when using NAM compared with NARR. Simulated friction velocity had a positive bias on average, with the YSU PBL scheme producing the largest overestimation in the innermost domain (0.5 km horizontal grid resolution). The simulated surface sensible heat flux had a similar temporal behavior as the observations but with a larger magnitude. The PX LSM produced lower and more reliable sensible heat fluxes compared with the Noah LSM. However, Noah latent heat fluxes were improved with a lower RMSE compared to PX, when NARR forcing data was used. Overall, these results suggest that there is not one WRF configuration that performs best for all the simulated variables (surface heat fluxes and meteorological variables) and situations (day and night).

The land surface is coupled to the atmospheric boundary layer through surface turbulent fluxes. Persistent cold-air pools (PCAPs) form in topographic depressions where cold, dense air fills the valley basin and in the presence of air pollution is accompanied by poor air quality. For the first time, the surface turbulence dataset from seven monitors during the Persistent Cold-Air Pool Study conducted in Salt Lake Valley, Utah (December 2010–February 2011), are analyzed. We found that the surface sensible (H) and latent (LE) heat fluxes were lower during strong PCAP events compared with non-PCAPs. The higher ratio of heat flux to net radiation (H/Rn and LE/Rn) for strong PCAPs compared with weak PCAPs is suspected to be related to the presence of boundary layer clouds, which could enhance the turbulent mixing through cloud top–down mixing. The daily average ground heat flux (G) was a similar order of magnitude toHand LE during wintertime. The highest surface turbulent fluxes and energy balance closure occurred in the stability range of −0.05 <ξ≤ −0.02, or under slightly unstable conditions, near the neutral stability range. The median surface exchange coefficient (C_h), a crucial parameter to determine surface turbulent fluxes in land surface models, was slightly higher at the bare land site (BL) than the short vegetation sites (PH and CR) in wintertime, suggesting the importance of dynamic land-use information in numerical models.