One- and two-equation models for canopy turbulence

The predictive skills of single- and two-equation ( or K-epsilon) models to compute profiles of mean velocity (U), turbulent kinetic energy (K), and Reynolds stresses ((u'w') over bar) are compared against datasets collected in eight vegetation types and in a flume experiment. These datasets range in canopy height h from 0.12 to 23 m, and range in leaf area index ( LAI) from 2 to 10 m(2) m(-2). We found that for all datasets and for both closure models, measured and modelled U, K, and (u'w') over bar agree well when the mixing length (l(m)) is a priori specified. In fact, the root-mean squared error between measured and modelled U, K, and (u'w') over bar is no worse than published values for second- and third-order closure approaches. Within the context of one-dimensional modelling, there is no clear advantage to including a turbulent kinetic dissipation rate (epsilon) budget when l(m) can be specified instead. The broader implication is that the added complexity introduced by the epsilon budget in K - epsilon models need not translate into improved predictive skills of U, K, and (u'w) over bar profiles when compared to single- equation models.