To drive a Lagrangian stochastic dispersion (LSD) model in and over a forest, vertical profiles of turbulence statistics that are characteristic of the forest architecture are needed. Far above the canopy, such turbulence statistics are expected to follow known similarity theory relationships. However, in the roughness sublayer and within the canopy no comparable similarity theory exists. Measured or simulated canopy turbulence characteristics from wind tunnels, LES runs or other forests are not easily generalized to apply universally. In view of this need, turbulence measurements were collected in and over a mixed hardwood forest at the University of Michigan Biological Station (UMBS) AmeriFlux site in the summers of 2000 and 2001. Velocity fluctuations and temperature were measured at seven levels within the canopy (up to the canopy height, h=22 m), using 1-D and 3-D sonic anemometers and fine-wire thermocouples. Six additional thermocouples were distributed over the canopy layer depth. Three-dimensional turbulence and sonic temperatures were also measured above the canopy, on the AmeriFlux tower at 1.6 h and at 2.15 h. Vertical profiles of buoyancy flux, mean horizontal velocity, Reynolds stress, standard deviation and skewness of velocity components were calculated, as these quantities are intended as input to the LSD model. Velocity spectra are computed to explore the potential of estimating the viscous dissipation rate, although preliminary results suggest that the high frequency range of the spectra do not generally exhibit the roll-off predicted by Kolmogorov theory. The analysis of these measurements will be used to determine a multi-layer parameterization framework of turbulence statistics for implementation in the LSD model.

Turbulence profiles and power spectra above the canopy are analyzed in the context of Monin-Obukhov similarity theory (MOST), as determined by stability at the top level (2.15 h), to assess the extent to which MOST is valid as the canopy height is approached. Preliminary results suggest that above the canopy velocity standard deviations converge to MOST predicted values towards the top level, and spectral peaks shift with stability, as expected from MOST. Within the canopy, both turbulence statistics profiles and spectral distributions follow the general characteristics of published results from forest canopies and reflect the vertical canopy architecture (as expressed by the vegetation area density profile).