Surface-layer fluxes in stable conditions

Micrometeorological tower data from the Microfronts experiment are analyzed. Scale-dependencies of the flux and flux sampling error are combined to automatically determine Reynolds turbulence cut-off time scales for computing fluxes from time series. The computed downward heat flux at the 3 m height averaged over nine nights with 7.3 hours each night is 20% greater than the downward heat flux computed at the 10 m height. In contrast, there is only a 1.2% difference between 3 m and 10 m heat fluxes averaged over daytime periods, and there is less than a 2% difference between 3 m and 10 m momentum fluxes whether averaged over nighttime or daytime periods. Stability functions, phi(M)(z/L) and phi(H)(z/L) are extended to z/L up to 10, where z is the observational height and L is the Obukhov length. For 0.01 < z/L < 1 the estimated phi functions generally agree with Businger-Dyer formulations, though the phi(H) estimates include more scatter compared to the phi(M) estimates. For 1 < z/L < 10, the flux intermittency increases, the flux Richardson number exceeds 0.2, and the number of flux samples decreases. Nonetheless the estimates of the stability function phi(M) based on 3-m fluxes are closer to the formula proposed by Beljaars and Holtslag in 1991 while the phi(M) functions based on 10-m fluxes appears to be closer to the formula proposed by Businger et al. in 1971. The stability function phi(H) levels off at z/L = 0.5.